Control of Nonlinear Systems via PI, PD and PID:

Stability and Performance 1st Edition Yong-Duan

Song (Author) download

https://textbookfull.com/product/control-of-nonlinear-systems-
via-pi-pd-and-pid-stability-and-performance-1st-edition-yong-
duan-song-author/

Download more ebook from https://textbookfull.com

We believe these products will be a great fit for you. Click
the link to download now, or visit textbookfull.com
to discover even more!

Stability and Control of Nonlinear Time varying Systems

Shuli Guo

https://textbookfull.com/product/stability-and-control-of-
nonlinear-time-varying-systems-shuli-guo/

Cooperative Control of Nonlinear Networked Systems

Infinite time and Finite time Design Methods Yongduan
Song

https://textbookfull.com/product/cooperative-control-of-
nonlinear-networked-systems-infinite-time-and-finite-time-design-
methods-yongduan-song/

Nonlinear Circuits and Systems with Memristors:

Nonlinear Dynamics and Analogue Computing via the Flux-
Charge Analysis Method Fernando Corinto

https://textbookfull.com/product/nonlinear-circuits-and-systems-
with-memristors-nonlinear-dynamics-and-analogue-computing-via-
the-flux-charge-analysis-method-fernando-corinto/

Adaptive and Fault-Tolerant Control of Underactuated

Nonlinear Systems 1st Edition Jiangshuai Huang

https://textbookfull.com/product/adaptive-and-fault-tolerant-
control-of-underactuated-nonlinear-systems-1st-edition-
jiangshuai-huang/
Advances and Applications in Nonlinear Control Systems
1st Edition Sundarapandian Vaidyanathan

https://textbookfull.com/product/advances-and-applications-in-
nonlinear-control-systems-1st-edition-sundarapandian-
vaidyanathan/

Dynamic Systems with Time Delays Stability and Control

Ju H. Park

https://textbookfull.com/product/dynamic-systems-with-time-
delays-stability-and-control-ju-h-park/

Nonlinear control systems using MATLAB First Edition

Boufadene

https://textbookfull.com/product/nonlinear-control-systems-using-
matlab-first-edition-boufadene/

Nonlinear Systems and Circuits in Internal Combustion

Engines: Modeling and Control 1st Edition Ferdinando
Taglialatela-Scafati

https://textbookfull.com/product/nonlinear-systems-and-circuits-
in-internal-combustion-engines-modeling-and-control-1st-edition-
ferdinando-taglialatela-scafati/

Biota Grow 2C gather 2C cook Loucas

https://textbookfull.com/product/biota-grow-2c-gather-2c-cook-
loucas/
 i i

“FM” — 2018/9/19 — 16:44 — page 1 — #1

i i

Control of Nonlinear
Systems via PI, PD and
PID
Stability and Performance

i i

i i
 i i

“FM” — 2018/9/19 — 16:44 — page 2 — #2

i i

AUTOMATION AND CONTROL ENGINEERING

A Series of Reference Books and Textbooks

Series Editors
FRANK L. LEWIS, Ph.D., SHUZHI SAM GE, Ph.D.,
Fellow IEEE, Fellow IFAC Fellow IEEE
Professor Professor
The Univeristy of Texas Research Institute Interactive Digital Media Institute
The University of Texas at Arlington The National University of Singapore

STJEPAN BOGDAN
Professor
Faculty of Electrical Engineering
and Computing
University of Zagreb

RECENTLY PUBLISHED TITLES

Discrete-Time Recurrent Neural Control: Analysis and Applications,
Edgar N. Sánchez
Electric and Plug-in Hybrid Vehicle Networks: Optimization and Control,
Emanuele Crisostomi, Robert Shorten, Sonja Stüdli, and Fabian Wirth
Adaptive and Fault-Tolerant Control of Underactuated Nonlinear Systems,
Jiangshuai Huang and Yong-Duan Song
Optimal and Robust Scheduling for Networked Control Systems,
Stefano Longo, Tingli Su, Guido Herrmann, and Phil Barber
Deterministic Learning Theory for Identification, Recognition, and Control,
Cong Wang and David J. Hill
Networked Control Systems with Intermittent Feedback,
Domagoj Toli´c and Sandra Hirche
Doubly Fed Induction Generators: Control for Wind Energy,
Edgar N. Sanchez and Riemann Ruiz-Cruz
Optimal Networked Control Systems with MATLAB®,
Jagannathan Sarangapani and Hao Xu
Cooperative Control of Multi-agent Systems: A Consensus Region Approach,
Zhongkui Li and Zhisheng Duan
Nonlinear Control of Dynamic Networks,
Tengfei Liu, Zhong-Ping Jiang, and David J. Hill
Modeling and Control for Micro/Nano Devices and Systems,
Ning Xi, Mingjun Zhang, and Guangyong Li
Linear Control System Analysis and Design with MATLAB®, Sixth Edition,
Constantine H. Houpis and Stuart N. Sheldon
Real-Time Rendering: Computer Graphics with Control Engineering,
Gabriyel Wong and Jianliang Wang
Anti-Disturbance Control for Systems with Multiple Disturbances,
Lei Guo and Songyin Cao
Tensor Product Model Transformation in Polytopic Model-Based Control,
Péter Baranyi, Yeung Yam, and Péter Várlaki

i i

i i
 i i

“FM” — 2018/9/19 — 16:44 — page 3 — #3

i i

Control of Nonlinear
Systems via PI, PD and
PID
Stability and Performance

Yong-Duan Song

i i

i i
 i i

“FM” — 2018/9/19 — 16:44 — page 4 — #4

i i

CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

© 2019 by Taylor & Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

Printed on acid-free paper

Version Date: 20180919

International Standard Book Number-13: 978-1-138-31764-2 (Hardback)

This book contains information obtained from authentic and highly regarded sources. Reasonable
efforts have been made to publish reliable data and information, but the author and publisher cannot
assume responsibility for the validity of all materials or the consequences of their use. The authors and
publishers have attempted to trace the copyright holders of all material reproduced in this publication
and apologize to copyright holders if permission to publish in this form has not been obtained. If any
copyright material has not been acknowledged please write and let us know so we may rectify in any
future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information
storage or retrieval system, without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access
www.copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc.
(CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization
that provides licenses and registration for a variety of users. For organizations that have been granted
a photocopy license by the CCC, a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and
are used only for identification and explanation without intent to infringe.

Library of Congress Cataloging-in-Publication Data

Names: Song, Yong-Duan, author.

Title: Control of nonlinear systems via PI, PD and PID : stability and
performance / Yong-Duan Song.
Description: First edition. | Boca Raton, FL : CRC Press/Taylor & Francis
Group, 2018. | Includes bibliographical references.
Identifiers: LCCN 2018026890 | ISBN 9781138317642 (hardback : acid-free paper)
Subjects: LCSH: Adaptive control systems. | Nonlinear control theory.
Classification: LCC TJ217 .S67 2018 | DDC 629.8/36--dc23
LC record available at https://lccn.loc.gov/2018026890

Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com

i i

i i
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page vii — #3

✐ ✐

To my family for the understanding, support and love.

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page vi — #4

✐ ✐

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page vii — #5

✐ ✐

Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Preview of Chapters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Classical PID Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.1 The Three Actions of PID Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.1 Proportional Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.2 Integral Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.3 Derivative Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Tuning Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

3 Adaptive PI Control for SISO Affine Systems . . . . . . . . . . . . . . . . . . . . . 9

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2 Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.3 Design Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.3.1 PI Control Design for First-order Nonlinear Systems . . . . . . 12
3.3.2 PI Control Design for High-order Nonlinear Systems . . . . . . 14
3.4 Numerical Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.6 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

4 Generalized PI Control for SISO Nonaffine Systems . . . . . . . . . . . . . . . 25

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.2 System Description and Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.3 Control Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

vii

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page viii — #6

✐ ✐

viii Contents

4.3.1 PI Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.4 Adaptive Fault-tolerant PI Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.4.1 PI Control under Actuator Failures . . . . . . . . . . . . . . . . . . . . . . 35
4.4.2 PI Control under Actuator and Sensor Faults . . . . . . . . . . . . . 36
4.5 Illustrative Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

5 Adaptive PI Control for MIMO Nonlinear Systems . . . . . . . . . . . . . . . . 43

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.2 Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5.2.1 System Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5.2.2 Neural Networks and Function Approximation . . . . . . . . . . . 47
5.3 PI Control Design and Stability Analysis . . . . . . . . . . . . . . . . . . . . . . . 48
5.3.1 Neuoadaptive PI Control for Square Systems . . . . . . . . . . . . . 49
5.3.2 Neuoadaptive PI Control for Non-square Systems . . . . . . . . . 53
5.4 Modified PI Control Based on BLF . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.4.1 Neuro-adaptive PI Control for Square Systems . . . . . . . . . . . . 56
5.4.2 Neuro-adaptive PI Control for Non-square Systems . . . . . . . . 58
5.5 Illustrative Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.7 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

6 Adaptive PI Control for Strict Feedback Systems . . . . . . . . . . . . . . . . . . 67

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
6.2 System Description and Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . 68
6.3 PI-like Control Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.4 Illustrative Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
6.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

7 Adaptive PID Control for MIMO Nonlinear Systems . . . . . . . . . . . . . . . 85

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
7.2 Problem Formulation and Error Dynamics . . . . . . . . . . . . . . . . . . . . . . 87
7.2.1 Error Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
7.2.2 Nussbaum Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
7.3 PID-like Control Design and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 90
7.3.1 PID Control for Square Systems . . . . . . . . . . . . . . . . . . . . . . . 91
7.3.2 PID Control for Non-square Systems . . . . . . . . . . . . . . . . . . . 94
7.3.3 Analysis and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
7.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

8 PD Control Application to High-Speed Trains . . . . . . . . . . . . . . . . . . . . . 99

8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
8.2 Modeling and Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
8.2.1 Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
8.2.2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page ix — #7

✐ ✐

Contents ix

8.3 Control Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

8.3.1 Structural Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
8.3.2 Robust Adaptive PD-like Control Design . . . . . . . . . . . . . . . . 105
8.3.3 Low-Cost Adaptive Fault-tolerant PD Control . . . . . . . . . . . . 109
8.3.4 Comparison and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
8.4 Simulation Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
8.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

9 PID Control Application to Robotic Systems . . . . . . . . . . . . . . . . . . . . . . 117

9.1 Robotic Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
9.2 PID Control for Robotic Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
9.2.1 Square System (joint-space tracking) . . . . . . . . . . . . . . . . . . . . 117
9.2.2 Non-square System (task-space tracking) . . . . . . . . . . . . . . . . 118
9.3 Case Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118

10 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page x — #8

✐ ✐

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/25 — 16:35 — page xi — #9

✐ ✐

Preface

The proportional integral derivative (PID) controller in the form we know it today
emerged between 1915 and 1940. Since PID control is simple in structure and
inexpensive in implementation, it has been undoubtedly the most widely employed
controller in industry. In fact, PID controllers are sufficient for many control
problems, particularly when process dynamics are benign and the performance
requirements are modest. However, there are still problems that limit the
applications of the PID controller. One of those is how to determine the appropriate
PID gains to ensure system stability and desirable performance. The
Ziegler–Nichols tuning procedure, published in 1942, is simple and intuitive, but it
creates a closed-loop system that is very poorly damped and that has poor stability
margins. Since then, various methods for tuning PID gains have been suggested, but
a systematic means is yet to be established and the information about those methods
is scattered in the control theory.
With the increasing control demands for various practical systems which are
generally nonlinear, uncertain and with abnormal actuation (such as asymmetric
saturation, dead-zone module, loss of effectiveness, etc.), traditional PID control
seems to lack theoretical support and is losing efficiency. Thus, a series of control
strategies is proposed to tackle the control problems for all kinds of nonlinear
systems. Although various control schemes have successfully addressed the control
problems for nonlinear systems, the resultant solutions seem quite sophisticated —
not only complicated in structure, but also expensive in computation. As a
consequence, these complex control methods are not much appreciated in practical
applications. By this token, the PID control seems still the most favorable choice for
the control of practical systems if it could be made effective in dealing with system
nonlinearities and uncertainties.
Industrial experience has clearly indicated that automatic tuning of PID gains is
a highly desirable and useful feature. Therefore, the focus of this book is on
PI/PD/PID controller for nonlinear systems with self-tuning gains, wherein an
exposition of adaptive PI/PD/PID control methods developed recently for numerous
nonlinear systems is provided. All these PI/PD/PID controllers are able to
adaptively update the gains through analytic algorithms and there is no need for

xi

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/25 — 16:35 — page xii — #10

✐ ✐

xii Preface

human tuning or trial and error process. Besides, the stability condition (the primary
concern for any control system) is established for the corresponding systems with
PI/PD/PID controller in the loop. Furthermore, in order to make the control scheme
more reliable in practical applications, in this book, the proposed PID control
strategies are equipped with fault-tolerant capabilities to accommodate the
abnormal actuation characteristics which may occur during system operation.
Constraints (due to physical saturation, safety specifications, etc.) imposed on
system outputs or states, together with the issue of prescribed control performance,
are also considered in control design. In the last chapters of the book, the
PI/PD/PID control scheme is applied to practical systems such as high-speed trains
and robotic systems. The effectiveness of the proposed adaptive PI/PD/PID
controller is demonstrated and validated via computer simulations. Several books
on PID controllers are available on the market, but this book exclusively focuses on
PI/PD/PID control with gain auto-tuning mechanisms for nonlinear systems. While
efforts have been made on PI/PD/PID control for nonlinear systems, there is still
much room for further research and development. We hope that this book will aid in
understanding the essence of PID control, providing readers with alternative
perspectives concerning the development of PI/PD/PID controllers for typical
nonlinear systems.

Yongduan Song
Chongqing, China

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page xiii — #11

✐ ✐

Acknowledgments

There are numerous individuals without whose help this book would not have been
completed. My sincere thanks go to Dr. Yujuan Wang, Dr. Danyong Li, Qing Chen,
Zhirong Zhang, Ziyun Shen, Ye Cao, Shuyan Zhou, Xiucai Huang, Kai Zhao. In
particular, I would like to express my gratitude to the following authors for allowing
me to use the materials of the papers for compiling this book.
• Y. D. Song, Y. J. Wang, and C. Y. Wen, “Adaptive fault-tolerant PI tracking
control with guaranteed transient and steady-state performance,” IEEE Trans.
Autom. Control, vol. 62, no. 1, pp. 481-487, 2017.
• Q. Song, Y. D. Song, “Generalized PI control design for a class of unknown
nonaffine systems with sensor and actuator faults,” Syst. Control Lett., vol. 64,
no. 1, pp. 86-95, 2014.
• Y. D. Song, J. X. Guo, and X. C. Huang, “Smooth neuroadaptive PI tracking
control of nonlinear systems with unknown and nonsmooth actuation
characteristics,” IEEE Trans. Neural Netw. Learn. Syst., vol. 28, no. 9, pp.
2183-2195, 2017.
• Y. D. Song, Z. Y. Shen, L. He, and X. C. Huang, “Neuroadaptive control of strict
feedback systems with full-state constraints and unknown actuation
characteristics: an inexpensive solution”, IEEE Trans. Cybern., DOI:
10.1109/TCYB.2017.2759498.
• Y. D. Song, X. C. Huang, and C. Y. Wen, “Robust adaptive fault-tolerant PID
control of MIMO nonlinear systems with unknown control direction,” IEEE
Trans. Ind. Electron., vol. 64, no. 6, pp. 4876-4884, 2017.
• Y. D. Song, X. C. Yuan, “Low-cost adaptive fault-tolerant approach for semiactive
suspension control of high-speed trains,” IEEE Trans. Ind. Electron., vol. 63, no.
11, pp. 7084-7093, 2016.
I also would like to thank the Research Institute of Intelligent Systems at
Chongqing University for their support.
The writing of this book was supported in part by the National Natural Science
Foundation of China.

xiii

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page xiv — #12

✐ ✐

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page xv — #13

✐ ✐

Acronyms

Abbreviations

UUB uniformly ultimately bounded

MAS multi-agent systems

Notations
R field of real numbers
Σ summation
|a| the absolute of a scalar a
kxk the norm of a vector x
max maximum
min minimum
sup supremum, the least upper bound
inf infimum, the greatest lower bound
∀ for all
∈ belongs to
→ tends to
< (>) less (greater) than
≤ (≥) less (greater) than or equal to
≪ (≫) much less (greater) than
Rn the n−dimensional Euclidean space
ẏ the first derivative of y with respect to time
ÿ the second derivative of y with respect to time
y(i) the i−th derivative of y with respect to time
w.r.t. with respect to

xv

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page xvi — #14

✐ ✐

xvi Acronyms

sat(·) the saturation function

sgn(·) the signum function
diag{a1 , · · · , an } a diagonal matrix with diagonal elements a1 to an
P>0 a positive definite matrix P
P≥0 a positive semi-definite matrix P
AT (xT ) the transpose of matrix A (a vector x)
λmax (P) (λmin (P)) the maximum (minimum) eigenvalue of a symmetric matrix P
f −1 (·) the inverse of a function f

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page 1 — #15

✐ ✐

Chapter 1
Introduction

1.1 Motivation

A proportional integral derivative (PID) controller is a control loop feedback

mechanism widely used in industrial control systems. Fig. 1.1 is a block diagram of
the classical PID controller. A PID controller continuously calculates an error value
e(t) as the difference between a desired setpoint r(t) and a measured process
variable y(t) and applies a correction based on proportional, integral, and derivative
terms (denoted P, I, and D, respectively) which give their name to the controller.

Fig. 1.1 Classical PID controller in a feedback loop. r(t) is the desired process input or setpoint,
y(t) is the measured process output, u(t) is the control input, and e(t) = r(t)− y(t) is the discrepancy
between the setpoint and the output.

In PID control, the P-control (proportional control) is the action based on current
behavior of the system, I-control (integral control) is the accumulated effort using
the experience information of bygone state, whereas D-control (derivative control)
is the predictive effort based on the tendency information for the ongoing state.
Since the Ziegler and Nichols’ PID tuning rules were published in 1942, the PID
control has survived the challenge of advanced control theories. The PID’s long life
comes from its clear meaning and effectiveness in practice; thus the PI/PD/PID
control has been widely accepted in industry. Especially, it is suitable for control
systems that can establish accurate mathematical models. However, most practical

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page 2 — #16

✐ ✐

2 1 Introduction

industrial processes are nonlinear, uncertain, and are difficult to establish accurate
mathematical models for. Thus, the traditional PID gains tuning methods cannot
obtain optimal control performance. To deal with this problem, various PID gains
self-tuning methods have been suggested. Though the PI/PD/PID control has been
widely accepted in industry, PI/PD/PID control itself is still short of the theoretical
basis, e.g., the optimality of PI/PD/PID control, performance tuning rules of
PI/PD/PID control, and automatic performance tuning methods have not been
clearly presented especially for the trajectory tracking control of nonlinear systems.
Our interest in revisiting PI/PD/PID control is largely motivated by the fact that,
although various advanced control methods have been developed during the past
decades, the preferred one in engineering practice is still the PI/PD/PID control, due
to its simplicity in structure and intuitiveness in concept. It has gained wide
application in practical engineering systems. However, the well-known PI/PD/PID
control exhibits two major drawbacks that restrict its application to more general
systems. The first one is the determination of the PI/PD/PID gains for a given
system is an ad hoc and painstaking process. Thus far, there exists no systematic
means to guide the determination of such gains that ensure system stability and
performance, although various methods for tuning PI/PD/PID gains have been
suggested in the literature. The second one is that although PI/PD/PID control has
been demonstrated to be quite effective in dealing with certain linear time-invariant
systems, its applicability to nonlinear systems remains unclear and lacks theoretical
insurance for closed-loop system stability and performance. Furthermore, it is
desirable or required to equip such PI/PD/PID schemes with adaptive and
fault-tolerant capabilities yet guaranteeing transient performance.

1.2 Objectives

Firstly, this book attempts to provide readers with an overview of the basic principle
of PID control. Traditional PID control is characterized with constant PID gains and
is oriented for set point regulation; thus it seldom works satisfactorily for general
nonlinear systems with uncertain dynamics and unpredictable disturbances. Besides,
stability has always been the major concern with traditional PID control due to the
lack of the systematic procedure for determining the proper stability-ensured PID
gains for a given dynamic system.
Secondly, through detail theoretical analysis and technical development, this
book intends to show how conventional PI/PD/PID controllers could be extended
and generalized to deal with various systems, such as SISO nonlinear systems,
SISO nonaffine systems, and MIMO nonlinear systems. The emphasis is on how to
enable these controllers with the capabilities of tuning their gains automatically to
compensate for system uncertainties and reject external disturbances. Furthermore,
as nonsmooth actuation characteristics or actuation failures (partial loss of
effectiveness (PLOE) or total loss of effectiveness (TLOE) ) might occur during

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page 3 — #17

✐ ✐

1.3 Preview of Chapters 3

system operation, effort is also made on designing PI/PD/PID control with adaptive
and fault-tolerant capabilities.

1.3 Preview of Chapters

In Chapter 2, a brief review of the traditional PID control with fixed gains is
presented.
In Chapter 3, a generalized PI control with adaptively adjusting gains is
presented for single input single output (SISO) nonlinear systems. We consider two
control schemes: one is for the first-order nonlinear system; the other is for the
high-order nonlinear system. Besides, the developed PI controller is suitable for
nonlinear systems with undetectable disturbances and actuation failures.
Meanwhile, the pre-scribed transient and steady-state performances are
dynamically maintained.
In Chapter 4, a generalized adaptive PI control is developed for unknown
nonaffine dynamic systems. As the control inputs enter into and influence the
dynamic behavior of the nonaffine system in a nonlinear and implicit way, control
design for such systems becomes quite challenging. The proposed control is able to
accommodate both sensor and actuator faults.
In Chapter 5, neuro-adaptive PI control algorithms with self-tuning gains are
developed for a class of multi-input multi-output (MIMO) normal-form nonlinear
systems subject to unknown actuation characteristics and external disturbances. It is
shown that the proposed neuro-adaptive PI control is continuous and smooth
everywhere and ensures the uniform ultimate boundedness of all the signals of the
closed-loop system. Furthermore, the crucial compact set precondition for a neural
network (NN) to function properly is guaranteed with the barrier Lyapunov function
(BLF), allowing the NN unit to play its learning/approximating role during the
entire system operation.
In Chapter 6, a neuro-adaptive PI control for a class of uncertain nonlinear strict
feedback systems with full-state constraints and unknown actuation characteristics
is presented. In order to deal with the modeling uncertainties and the actuation
characteristics impact, the neural networks are utilized at each step of the back
stepping design procedure.
In Chapter 7, it is shown that the structurally simple and computationally
inexpensive PID control, popular with SISO linear time-invariant systems, can be
generalized and extended to control nonlinear MIMO systems with nonparametric
uncertainties and actuation failures. By utilizing the Nussbaum-type function and
the matrix decomposition technique, non-square systems with unknown control
direction are also considered.
In Chapter 8, the PD-like controller is designed for a high-speed train system.
The situation is further complicated if actuation faults occur. The resultant control
scheme is capable of automatically generating the intermediate control parameters

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page 4 — #18

✐ ✐

4 1 Introduction

and literally producing the PD-like controller. The whole process does not require
precise information regarding system model or system parameter.
In Chapter 9, the robust adaptive PID controller is applied to a robotic system.
Under the proposed PID-like control the vibrations are effectively suppressed in the
presence of parametric uncertainties and varying operation conditions.

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page 5 — #19

✐ ✐

Chapter 2
Classical PID Control

In this chapter, the structure of the PID controller and the roles of the three
(proportional, integral, and derivative) terms of the PID controller, together with the
tuning of the PID gains, are discussed.

2.1 The Three Actions of PID Control

A typically PID controller involves three types of control actions: a proportional

action, an integral action, and derivative action, which can be mathematically
expressed as
Z t
de(t)
u(t) = K p e(t) + Ki e(τ )d τ + Kd (2.1)
0 dt
where K p , Ki , and Kd denote the proportional, integral, and derivative gain,
respectively. The role of each term is described and discussed briefly in what
follows.

2.1.1 Proportional Action

The proportional control action is proportional to the current control error, which can
be expressed as
u(t) = K p e(t) = K p (r(t) − y(t)) (2.2)
where K p is the proportional gain. The role of such control is quite obvious since it
implements the typical operation of increasing the control effort when the control
error is large (with appropriate sign). The transfer function of a proportional
controller can be derived trivially as

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page 6 — #20

✐ ✐

6 2 Classical PID Control

C(s) = K p (2.3)

If K p is made large, the steady-state error would be small. But the dynamic response
would become worse because the damping is too low. Apparently, a proportional
controller has the advantage of providing a small control input when the control error
is small and therefore can avoid excessive control efforts. The main drawback of
using a pure proportional controller is that it produces a steady-state error. It is worth
noting that this would still occur even if the process bears an integrating dynamics
(i.e., its transfer function has a pole at the origin of the complex plane), in case a
constant load disturbance occurs. This motivates the addition of a bias (or reset)
term ub , namely, [1, 2, 3]
u(t) = K p e(t) + ub (2.4)
The value of ub can be fixed at a constant level (usually at (umax + umin )/2) or can
be adjusted manually until the steady-state error is reduced to zero, where umax and
umin denote the maximum and minimum value of the control input, respectively.

2.1.2 Integral Action

The integral action is proportional to the integral of the control error, i.e.,
Z t
u(t) = Ki e(τ )d τ (2.5)
0

where Ki is the integral gain. With the integral action, the resultant control makes use
of the past values of the control error to generate its control signal. The corresponding
transfer function is:
Ki
C(s) = (2.6)
s
The presence of a pole at the origin of the complex plane allows the steady-state
error to be reduced to zero when a step reference signal is applied or a step load
disturbance occurs. In other words, the integral action is able to set automatically
the correct value of ub in (2.4) so that the steady-state error is zero [1]. This actually
results in a PI controller with the following transfer function
1
C(s) = K p (1 + ) (2.7)
Ti s
where Ti is integration time constant. The block diagram in Fig. 2.1 shows how
integral action is implemented using positive feedback with a first-order system.
The controller output is low-pass-filtered and feed back with positive gain. The
integral action is used to generate the bias term ub in (2.4) in the proportional
controller automatically, often called automatic reset. For this reason the integral
action is also often called automatic reset. Thus, the use of a proportional action in
conjunction to an integral action, i.e., a PI controller, solves the main problems of

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page 7 — #21

✐ ✐

2.1 The Three Actions of PID Control 7

the oscillatory response associated to an on-off controller and of the steady-state

error associated to a pure proportional controller.

Fig. 2.1 PI controller involving P and I actions.

2.1.3 Derivative Action

The derivative control takes the following form,

de(t)
u(t) = Kd (2.8)
dt
where Kd is the derivative gain, which makes use of the predicted future values of
the control error. The corresponding controller transfer function is

C(s) = Kd s (2.9)

Upon using Euler formula to approximate the derivative in (2.8), it is derived that

de(t)
e(t + Td ) ≃ e(t) + Td (2.10)
dt

where Td is the sampling period, implying that e(t) + Td de(t)

dt is able to reflect the
value of the control error at time t + Td . So if a control law proportional to this
expression is considered, i.e.,

de(t)
u(t) = K p (e(t) + Td ) (2.11)
dt
then the control input at time t is actually based on the predicted value of the control
error at time t +Td . In order words, the controller (2.11) consisting of P and D terms is
able to enhance the transient response of the closed-loop system. For this reason the
derivative action is also called anticipatory control, or rate action, or pre-act [1]. It
should be stressed that although the derivative action has great potential in improving
the control performance as it can anticipate an incorrect trend of the control error and

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page 8 — #22

✐ ✐

8 2 Classical PID Control

counteract it, it also creates some critical issues that should be carefully addressed in
control design in practice.

2.2 Tuning Methods

As the structure of the PID controller is fixed, the tuning of its PID gains should be
carefully considered for different applications with various requirements. By
properly tuning the three parameters, a PID controller can deal with specific process
requirements. There is no uniform method for tuning the PID gains. The most
commonly used one is the “trial-and-error” tuning, which starts with determining
the proportional gain K p first, then trying to find the integral time constant Ti and
the derivative time constant Td , with which the integral gain Ki is obtained by
Ki = K p /Ti and the derivative gain Kd is set as Kd = Td K p .
The Ziegler-Nichols (ZN) tuning rule is also a popular method used in practice.
The tuning rule is simple and needs only the ultimate information, which can be
estimated easily by simple identification methods, such as the continuous-cycling
method and relay feedback identification method [4]. The ZN tuning rule works
satisfactorily for certain processes. However, because the ZN tuning rule uses only
the ultimate data of the process, its performance is uncertain for those systems with
unusual frequency response characteristics.
For more complicated systems, manual calculation methods are no longer
practical. Software based PID tuning and loop optimization is a must. There are
some software packages that gather the data, develop process models, and suggest
optimal tuning. Some software packages can even develop tuning procedures by
gathering data from reference changes [1, 2].

2.3 Conclusion

A brief overview of traditional PID controller is presented in this chapter, starting

with the three actions of PID control. In order to make the PIDcontrol function
satisfactorily with some specific process requirements, proper methods for tuning
PID gains should be utilized. Although a number of PID tuning methods (e.g.,
trial-and-error tuning, Ziegler—Nichols method and PID tuning software) are
available in literature, the traditional PID controllers with fixed gains are apparently
ineffective in dealing with the increasingly sophisticated systems. Several
PI/PD/PID control design methods for complex nonlinear systems are presented in
the following chapters.

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page 9 — #23

✐ ✐

Chapter 3
Adaptive PI Control for SISO Affine Systems

It is a long lasting open problem to synthesize a general PI control for nonlinear

systems with its gains analytically determined, yet ensuring stability and transient
performance. The problem is further complicated if modeling uncertainties and
external disturbances as well as actuation failures are involved in the systems. In
this chapter, a generalized PI control with adaptively adjusting gains is presented,
which gracefully obviates the ad hoc and time-consuming “trial and error” process
for determining the gains as involved in traditional PI control; collectively
accommodates modeling uncertainties, undetectable disturbances, and undetectable
actuation failures that might occur in the systems; and dynamically maintains
prespecified transient and steady-state performance.

3.1 Introduction

The problem addressed in this chapter is: would PI (proportional and integral)
control be applicable to uncertain nonlinear systems? Our interest in revisiting PI
control is largely motivated by the fact that, although various advanced control
methods have been developed during the past decades, the preferred one in
engineering practice is still the PID/PI control, due to its simplicity in structure and
intuitiveness in concept. Therefore it has gained wide application in practical
engineering systems [5, 6, 7]. However, the well-known PI control exhibits two
major drawbacks that restrict its application to more general systems. The first one
is that the determination of the PI gains for a given system is an ad hoc and
painstaking process. Thus far there exists no systematic means to guide the
determination of such gains that ensure system stability and performance, although
various methods for tuning PI gains have been suggested in the literature [2, 3, 5, 8].
The second one is that although PI control has been demonstrated quite effective in
dealing with certain linear time-invariant systems, its applicability to nonlinear
systems remains unclear and lacks theoretical insurance for closed-loop system
stability and performance. While some efforts have been made in developing

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page 10 — #24

✐ ✐

10 3 Adaptive PI Control for SISO Affine Systems

algorithms for tuning/adjusting PID/PI gains by utilizing generic algorithms, neural

networks, and/or fuzzy system techniques [9, 10, 11, 12, 13] (to name a few), there
still leaves much to be desired in most existing methods in terms of simplicity,
affordability, and effectiveness. The interesting issue to address is therefore: would
it be possible to construct PI-like control capable of dealing with nonlinear
uncertain systems where the PI gains are systematically and adaptively determined
by the control algorithm itself? Furthermore, is it possible to equip such a PI scheme
with adaptive and fault-tolerant capabilities yet guarantee transient performance?
The purpose of this chapter is to present a solution to address these issues.

3.2 Problem Formulation

Consider the following class of uncertain nonlinear systems,

ẋk = xk+1 , k = 1, 2, · · · , n − 1
ẋn = g(X,t)ua + f (X,t) (3.1)

where xk ∈ R (k = 1, · · · , n) is the kth state with x1 = x; X = [x1 , · · · , xn ]T ; ua ∈ R

is the actual control input of the system (the output of the actuator); g(·) ∈ R is the
time-varying and uncertain control gain; f (·) ∈ R denotes the lumped uncertainties
and external disturbances.
As unanticipated actuator faults may occur, we additionally include such
scenarios in the model, where the actual control input ua and the designed input u
are no longer the same in that

ua = ρ (tρ ,t)u + ur (tr ,t) (3.2)

where 0 ≤ ρ (·) ≤ 1, known as the “healthy indicator” [12], indicates the actuation
effectiveness, ur (·) is the uncontrollable portion of the control signal, tρ and tr
denote, respectively, the time instant at which the loss of actuation effectiveness
fault and the additive actuation fault occur. In this chapter, we consider the case that
0 < ρ (·) ≤ 1, i.e., although losing its effectiveness, the actuation is still functional
such that ua can be influenced by the control input u all the time. In addition, tρ and
tr are assumed completely unknown; this fact, together with the unknown and time
varying ρ and ur , literally implies that the occurrence instant and the magnitude of
the actuation faults are unpredictable. The dynamic model considering actuation
failures then becomes

ẋk = xk+1 , k = 1, 2, · · · , n − 1
ẋn = g(X,t)ρ (tρ ,t)u + f (X,t) + g(X,t)ur(tr ,t) (3.3)

The objective is to design a PI-like tracking controller for the system with
lumped uncertainties and disturbances as well as actuator faults as described by

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page 11 — #25

✐ ✐

3.2 Problem Formulation 11

(3.3) such that not only stable tracking is achieved, but also pre-described
performance is ensured, yet all the internal signals are continuous and bounded.
More specifically, the PI-like control ensures that: 1) the tracking error
E = X − X ∗ = [ε1 , ε2 , · · · , εn ]T (ε = ε1 ) converges to a small residual set containing
the origin for any given desired trajectory X ∗ = [x∗ , ẋ∗ , · · · , x∗(n−1) ]T ; 2) the tracking
error is confined within a pre-given bound all the time, i.e., there exist performance
functions µ1k (t) and µ2k (t) such that µ1k (t) ≤ εk (t) ≤ µ2k (t) (k = 1, · · · , n) for all
t ≥ 0. In addition, the convergence rate is controlled by e−a0t for some pre-specified
constant a0 > 0; and 3) all the internal signals in the system are ensured to be
continuous and bounded.
To proceed, the following assumptions are in order.
Assumption 3.1 The control gain g(·) is unknown and time-varying but bounded
away from zero, i.e., there exist some unknown constants g and ḡ such that 0 < g ≤
|g(·)| ≤ ḡ < ∞, and g(·) is sign-definite (in this note sgn(g) = +1 is assumed without
loss of generality).
Assumption 3.2 The desired state x∗ and its derivative up to (n − 1)th are assumed
to be smooth and bounded. In addition, x∗(n) , the nth derivative of x∗ , is bounded by
an unknown constant xm , i.e., |x∗(n) | ≤ xm < ∞, ∀t ≥ t0 .
Assumption 3.3 For uncertain nonlinearities f (·), there exist an unknown constant
c f ≥ 0 and a known scalar function ϕ (X,t) ≥ 0 such that | f (·)| ≤ c f ϕ (·). If X is
bounded, so is ϕ (X,t).
Assumption 3.4 ρ (·) and ur (·) are unknown, possibly fast time-varying and
unpredictable, but bounded in that there exist some unknown constants ρm and r̄
such that 0 < ρm ≤ ρ (·) ≤ 1 and |ur (·)| ≤ r̄ < ∞.
Remark 3.1 Assumptions 3.1—3.2 are commonly imposed in most existing works
in addressing the tracking control problem of system (3.1) [10, 14, 15, 16, 17].
Assumption 3.3 is related to the extraction of the core information from the
nonlinearities of the system, which can be readily done for any practical system
with only crude model information. As for Assumption 3.4, it is noted that most
FDD/FDI based fault tolerant control implicitly assumes that the faults vary with
time slowly enough to allow for timely fault identification and diagnosis [18, 19] or
that one has enough information on the faults to carry out parametric
decomposition [16], while Assumption 3.4 imposes no such restriction, and thus
seems more practical.
Remark 3.2 Note that in practice it would be very difficult, if not impossible, to
obtain the exact values of those bounds involved in Assumptions 3.1—3.4. The
developed PI-like control in this chapter, however, is independent of those bound
parameters, thus there is no need for analytical estimation of such bounds despite
the fact that those bounds do exist in stability analysis.

✐ ✐

✐ ✐
 ✐ ✐

“pidbook20180916” — 2018/9/19 — 17:14 — page 12 — #26

✐ ✐

12 3 Adaptive PI Control for SISO Affine Systems

3.3 Design Details

To help understand the fundamental idea and the technical development of the
proposed method, we start with controller design for the first-order nonlinear
systems, followed by the extension to the high-order case.

3.3.1 PI Control Design for First-order Nonlinear Systems

In this subsection we develop the generalized PI control law for first-order nonlinear
systems with actuation failures as described by (3.2). In this case (3.3) with (3.2)
becomes
ẋ(t) = g(x,t)ρ (·)u(t) + g(x,t)ur (·) + f (x(t),t) (3.4)
where x ∈ R denotes the system state. To facilitate the PI controller design, we first
introduce a filtered variable s as,
Z t
s = ε +β ε dτ (3.5)
0

where ε = x − x∗ is the tracking error, and β > 0 is a free parameter chosen by the
designer.
To establish the main results, the following lemma is needed.
Lemma 3.1 R Consider the filtered variable s defined in (3.5). If limt→∞ s = 0, then
ε (t) and 0t ε d τ converge asymptotically to zero as t → ∞Rwith the same decreasing
rate as that of s. In addition, if s is bounded, so are ε and 0t ε d τ .

Proof. The proof can be readily done by using the L’Hopital’s rule, so is omitted
here.

The proposed generalized PI control is of the form

Z t
u = −(k p1 + ∆ k p1 (t))ε (t) − (kI1 + ∆ kI1 (t)) ε (τ )d τ (3.6)
0

Different from the traditional PI control that involves constant gains, the PI gains
here consist of two parts: 1) constant gains k p1 > 0 and kI1 = β k p1 > 0, with k p1 and
β being chosen freely by the designer and 2) time-varying gains ∆ k p1 (t) and ∆ kI1 (t)
determined automatically and adaptively by the following algorithm,

ĉψ 2
∆ k p1 = , ∆ kI1 = β ∆ k p1 (3.7)
ψ |s| + ι
with
σ1 ψ 2 s2
ĉ˙ = −σ1 γ1 ĉ + (3.8)
ψ |s| + ι

✐ ✐

✐ ✐
Another Random Scribd Document
with Unrelated Content
The same weights denote the quantity of stars which the rocket will safely
carry: thus an ounce and a half of stars may be put into the head of a 6⁄8
rocket; and 3 drams into a 3⁄8 rocket. Along with the stars is to be put in 1⁄9 the
weight of the stars of bursting powder; this may be pure meal powder; or a
mixture of 8 meal powder, 1 fine charcoal, well sifted together: or half meal,
half grain; thus the quantity of bursting powder for a 6⁄8 rocket will be 24⁄9 =
22⁄3 drams; and for a 3⁄8 rocket, 3⁄9 = 1⁄3 of a dram. It is advisable to keep
nearly to these directions, for the weight of the stars; but it is not necessary to
be minutely exact; but to ascertain whether the head is too heavy or not, take
a table knife, or any thing an inch broad, and lay the rocket flat and horizontally
upon it in such a manner that the commencement of the head lies close to the
back edge of the knife, the cutting edge lying towards the choke: if the head
pitches over, it is too heavy, and some of the stars must be taken out; if it
balances, it is correct.
Rockets should be fired from two staples, or two screweyes, fixed in a post, one
near the top, the other half a yard below, as in fig. 70. They should never be
propped against a wall, a chairback, a gate, or railing, as they might fall,
especially on a windy night, after the touch-paper was lit, and before the fuse
had caught. Every possible care should be taken in guiding them, as it is too
late to think about any mischief they may cause, after they have once started.
In making rockets, it is essential, above all things, to have good nitre and
charcoal. The best way, with fresh materials, is to weigh out as much nitre,
charcoal, and sulphur, as will make one small rocket. Have the nitre as fine as
possible, and dry it over the fire in a 6-inch frying pan, which should be kept for
the purpose. If the rocket ascends well you will know that the articles are pure,
and you can proceed to use them; but if the rocket does not rise, you may
conclude the articles are adulterated, the nitre with salt, or that the charcoal is
perhaps merely deal sawdust, burnt in a retort. In this case you must buy the
nitre in crystals, and the charcoal in sticks. To powder the nitre, put it into a
pipkin, pour on it a little water, set it on the fire, make the water boil, and keep
stirring the nitre with a piece of wood, until it is dry and of a fine powder.
Charcoal is best ground up in a coffee-mill. It must first be broken into small
pieces, about the size of coffee-beans. After being ground it should be sifted
through the sieve, the interstices of which are about 1⁄40 of an inch square. To
get a correct idea of this size, lay down a line on paper, an inch long: mark it off
into eight equal parts by taking the divisions from a foot-rule; then carefully
divide one of the eighths into five equal parts. But the best way is to make a
square deal sieve, about 8 inches square, and 3 inches deep; then nail on the
bottom a piece of perforated zinc, with quarter-inch flemish tacks. The
perforations are circular, and should be 20 to the inch, measured diagonally;
that is, a diamond-shaped inch, or inch rhombus, consisting of 20 oblique rows,
each row containing 20 holes; 20 times 20 = 400. These holes will be about the
same size as the ones mentioned in the sieve, because the solid parts between
the perforations are also about 1⁄40 of an inch. A strip of deal, 1⁄4 of an inch
thick, should be nailed round the bottom, to keep the zinc tight. A square box
for a receiver, and another for a lid, should also be constructed, otherwise you
will be smothered in sifting charcoal. Making use of the perforated zinc sieve
just described, all the charcoal that goes through, fine and coarse, should be
used for rockets. It need not be shaken much, but brushed through with a
sash-tool. Construct a second sieve with 15 perforations to the lineal inch. Iron
borings for gerbes should be sifted through this; use all that goes through.
Construct, also, a third sieve with 9 perforations to the lineal inch. Nothing can
equal iron and steel for making sparks; but neither will keep long from rusting,
after coming in contact with nitre. Coke grains, about the size of pinheads, are
a fair substitute. Beat coke into lumps about as large as peas, then grind it in a
coffee-mill, and brush it through the 9-perforation sieve; sift the fine dust from
what passes through: throw the dust away, and keep the grains. Porcelain may
be powdered in an iron mortar, for gerbes; brush through the 9-perforation
sieve; sift out the dust with a fine sieve; throw it away and keep the grains. The
intense heat of the focus of the choke renders them incandescent; and, from
their weight, they are projected to a considerable distance. They are inferior,
however, to coke grains, as the latter are to steel filings. We may say,
alliteratively; positive, porcelain; comparative, coke; superlative, steel.
Zinc may be obtained with half-inch perforations; a size useful for garden
sieves, bottoms of soap boxes, &c.
Before grinding a fresh substance in a coffee-mill, it must be taken to pieces,
brushed clean, and screwed up again. Fine lawn or hair sieves should be used
for sifting chemicals; excellent sieves may be made with book-muslin: the
cylinders may be 4 inches diameter, 3 deep; the muslin should be cut into a
circular form, and hemmed round a piece of string; it may then be slipped over
the drum or cylinder, and secured; or it may be pasted up and round the sides,
and if above 4 inches diameter, two pieces of string may be crossed over the
middle to strengthen it.
Charcoal may be made by putting some dry pieces of willow, alder, poplar,
sycamore, maple, or almost any kind of wood, except the pine or turpentine
tribe, into an old iron saucepan, covering them with perfectly dry sand, and
setting the saucepan in the middle of a fire, to remain red hot till the wood is
completely burnt through. Remove when judged sufficiently charred; and when
cold, not before, pour away the sand.
Sulphur is used in the state of sublimed sulphur, sulphur sublimatum, or flowers
of sulphur, and, when mixed with nitre, requires no preparation; but as it is
always more or less impregnated with sulphuric acid, as is readily shown by
testing it with litmus paper, it might, on coming into contact with chlorate of
potash, cause spontaneous combustion. To prevent this, it is necessary to wash
the sulphur. For this purpose put it into a pan, and pour upon it boiling water, in
which some salts of tartar (carbonate of potash) have been dissolved; stir it
well and break down all lumps. Let it stand to subside; pour off the supernatant
liquor; fill up with cold water and let it stand, to again subside. Make a conical
bag, fig. 33, with a piece of linen or calico, sewed at the top, round a ring or
hoop of wire, or cane, or whalebone; fasten a string to it, by which to hang it
up. Put the washed sulphur into it, and hang it under a water-tap; turn the
water gently on, and let it drip all night; this will wash away every trace both of
acid and alkali. Afterwards hang the bag up three or four days till the sulphur is
dry; it may then be bottled, and kept exclusively for colours.
Oxalate of soda may be made thus—procure 3 lbs. of carbonate of soda, the
common washing soda used by the laundress, not bicarbonate of soda; boil it
up in a saucepan with just as little water as will suffice to dissolve it. Dissolve,
in another vessel, 1 lb. of oxalic acid in boiling water, and pour it into a deep jar,
capable of holding two or three quarts; a wash-hand jug will answer. Now put
to this the dissolved carbonate of soda, with a table-spoon, a spoonful at a
time. A violent effervescence takes place. The soda is to be slowly added till
effervescence ceases. It should be tested with a strip of litmus paper, to see if
the acid is perfectly neutralized.
To prepare litmus paper, dissolve 1⁄4 of an ounce of litmus in an ounce and a
half of water; when thoroughly dissolved, and the water is of a dark blue colour,
take some white blotting paper, and with a sash-tool or camel's-hair pencil, go
over it on both sides with the litmus solution. When dry, wet some of these
prepared pieces, with the brush dipped into vinegar: this will turn them red.
Dry, and preserve both. They may be cut into strips, half an inch broad; the
blue strips will be tests for acids; the red, for alkalies. Wet a strip of the blue,
and touch it with oxalic acid, it will turn red; wet a strip of the red, or the piece
just reddened, with carbonate of soda, it will turn blue.
To make sulphuret of copper, procure some thin sheet copper, about as thick as
a card; cut it into pieces, and put it into a crucible, with sulphur, a layer of
sulphur, and a layer of copper alternately, till full. Set the crucible in a clear fire,
and keep it red hot for an hour. Remove it; when cold, break it up, and grind it
in a coffee-mill. Sift it in a lawn or book-muslin sieve as fine as possible. Half-a-
pound of copper and a quarter of a pound of sulphur may be employed.
There is a black sulphide or sulphuret of copper produced by passing
sulphuretted hydrogen through a solution of protoxide of copper: this is useless.
For want of a coffee-mill, charcoal may be beaten in a leather bag, with a
hammer.
A variety of rocket fuses will be found in the Tables; the first is as good as any,
and will answer for all sizes from 3⁄8 to 12⁄8. As a rule, the fiercer fuses,
containing meal powder, may be used for small rockets; but are, by no means,
necessary.
A rocket, when starting, makes a roar; but this is not on account of the
fierceness of the fuse, but of the extent of the surface ignited. Rocket
composition, laid in a train, burns very slowly.

Rocket Stars.
Rocket stars are made in three or four ways. First cut, or chopped, or naked
stars. This mode is used for nitre stars only: chlorate of potash stars require
different methods. It is a singular circumstance that, though chlorate of potash
stars are much fiercer than nitrate of potash stars, yet the latter light without
any trouble, while the former, if made in the same way, would be almost sure to
miss.

To Make Cut Stars.

Wet the composition with thin starch, or dextrine solution, or gum water,
sufficiently to bind; press it into a flat mass, on a slate, or Dutch tile, with a
knife, or small trowel, till about 3⁄8 of an inch thick. Indent the surface with the
edge of the knife, in parallel lines, about 3⁄8 of an inch apart, and cross these
with equidistant indentations, at right angles. Set the mass by, to dry gradually.
When nearly dry, break it up into little 3⁄8 cubes, and lay them out, to dry
thoroughly. The broken edges will be rough, and will catch easily.

Dry Pill-box Stars.

Take a sheet of note paper, and cut it into four equal parts; each part will be
about 41⁄2 inches by 31⁄2. Paste and roll them on a 31⁄2/8 brass tube, so as to
have the cases 41⁄2 inches long. To make these into pill-boxes, perfectly true,
like those used by the druggist, they must be cut in the lathe. For this purpose,
turn a cylindrical piece of wood, fig. 27, which is to fit easily into the case,
except at a, where it is to be turned sloping a little larger, so that when the case
is slipped over it, it will bite at the part a; otherwise, on putting the chisel to it,
it would slip round, without getting cut. Mark the case, with a black-lead pencil,
at every half-inch; suspend it in the lathe; and cut it at the marks. Next procure
a 7⁄16 inch punch, with which to cut out the bottoms. These are to be made of
card, or bristol-board. Lay the card on a piece of sheet-lead, or the grain end of
a piece of beech, and give it a smart blow or two with a hammer; keep on
punching till the punch contains a dozen or more discs, then push them out.
When a number are ready, press them into the pill-boxes, with a rammer that
fits loosely.
The composition is to be put into these dry, and driven in with a solid rammer,
and the little mallet, before described.
This was the original way of making them, but is perfectly unnecessary. Roll the
tubes as directed, of two thicknesses of paper, with a little bit to lap over. Cut
them across, with one sharp clip, with a strong pair of scissors. This will slightly
flatten them; but they may readily be restored to the cylindrical form, by
slipping them on a piece of wood, and rounding them to shape with the fingers.
No bottoms need be provided, no punch used.

To Fill the Boxes with Dry Colour.

Rest the box on a flat surface, put in some composition, and drive it in with a
brass or boxwood drift and the little mallet, till half full, as in fig. 28. Then fill up
the box with more colour, set a little bit of match upright in the side, and mallet
it in, till the box is almost full. The drift for this must be cut flat on one side, to
allow for the match. On the top put a very little dry meal powder, or shell-fuse,
and press it in with the finger. Cut a piece of double-crown, about an inch
broad, and long enough to go rather more than once round the pill-box: paste
it all over; lay the pill-box on it, and roll it up; tuck in one end, to make a
bottom, and press the other end round the match, and on the meal powder, or
shell-fuse, till it assumes the form of fig. 29.

Another Way.
Set the pill-box on a flat surface, put in a very little meal powder or shell-fuse,
then some composition, and mallet it in till full. Roll this up in a piece of double-
crown as before; tuck in the bottom, and set by to dry; when dry, put into the
other end—the end containing the film of meal or shell-fuse—a piece or two of
thin match, and tie it in, as in fig. 30. Dry pill-boxes are best for making
chameleon stars; these are half one colour and half another. Make a dozen
stars, half yellow and half green, and a dozen more half blue and half crimson;
put these into a rocket; they will burst green and crimson, and change to blue
and yellow.

Bottomless Pill-boxes.
The cases are the same as before. To fill them, damp the composition as for
Roman candle stars; put a bit of quickmatch into the case, as at fig. 31, and
press in the composition. This is usually done with the fingers, but is not very
pleasant work, especially with lac solution. A cleaner way, though more tedious,
is to fit the case on to a little foot, with a side notch in it, fig. 60; then slightly
mallet in the composition. Or a notch may be cut in the side of the box, fig. 61,
and the match put in, as fig. 62, and slightly malleted.
Instead of making the cases entirely of writing paper, they may be made half of
writing paper and half of coloured double-crown, to indicate the colour of the
star.
Another way is to sift a thin layer of French chalk over a sheet of paper, and to
roll the stars in it, one by one, as they are punched. When dry, brush off the
superfluous chalk, and prime with a bit of match, tied across the mouth.

Enveloped Stars.

Pump and drive the stars exactly as for roman candle stars, They may be 31⁄2/8
diameter, and 5⁄8 long; or they may be formed with figs. 1, 2, and 3, and the
side pin of fig. 2 may be removed to c, in which case the stars will be 5⁄8
diameter, and 31⁄2/8 deep; the former will be an oblong cylinder, the latter an
oblate. Cut a strip of red, blue, green, or yellow double-crown, of a suitable
breadth, and long enough to go twice round the star. Paste the strip all over, or
gum it at the edge only, and lay the star upon it, as at a, fig. 41, with a bit of
match behind it; then roll it up and put by to dry. When pasted, it shrinks and
holds the match tight; when gummed at the edge this is not always the case; it
may then have a piece of thin binding-wire twisted round it. It will have the
appearance of fig. 32. In fig. 31, the match is embedded, and dries in the
damped composition; in the enveloped stars the stars are dried first, and the
match lies outside and blows through. Enveloped stars show well in the air.
The word envelope, as used in these pages, must not be confounded with the
same word as applied to the coverings for letters. It is rather synonymous with
the term wrapper; you lay an ounce of tobacco on a piece of paper, roll it up
and tuck in the ends. So with cases: you roll them up in a piece of paper, and
leave an inch vacant at each end to receive quickmatch, &c. This is termed the
envelope, that is, the wrapper.

Rocket Heads.
Heads for small rockets may be made of two or three rounds of paper rolled
dry, and secured on the inner and outer edges with paste. After sticking it on
the case, which it may be made to fit, as in fig. 36, pinch the top in like a
choke, only tight, and tie it round with twine or flax. For coloured rocket heads,
6
⁄8 and upwards, the head may be enlarged by fixing a collar round the top of
the case. To make the collar for a 6⁄8, roll a case on a 9⁄8 former, and when dry
cut it into short lengths in the lathe, as recommended for pill-boxes; if
unprovided with a lathe, saw it with a fine-toothed saw. The advantage of the
enlarged head is, that it brings the stars nearer to the rocket, and prevents it
from being top-heavy. Another way is to make the heads tapering, and the tops
conical, as in fig. 39. One part is rolled on fig. 34, the other on fig. 35; or, a
cone may be made of a circular piece of paper, without a former. Cut the circle,
along the radius, to the centre; bend it into a cone; secure the edge with
sealing-wax, and paste paper over it to overlap the rim: snip the edge with the
scissors; paste it inside, and secure it to the other part. But the quickest way of
all is to make a long paper bag, which may be made to fit to the greatest
nicety. For this purpose take a tape-measure; or lay down, on the edge of a
strip of writing paper, 7 or 8 inches long, and 1 inch broad, 6 inches, divided
into eighths, transferred from a foot-rule. Suppose the 6⁄8 rocket has a collar,
which, on being measured by the paper just alluded to, is found to be 42⁄8
inches round; add to this 3⁄8 for lapping over, making 45⁄8 inches. Cut a piece of
imperial brown 45⁄8 broad, and as long as the case, 6 inches. Make this into a
paper bag, 21⁄8 inches broad. Be careful that the corners are perfect; a strip of
double-crown may be pasted over them. When dry, pinch the mouth open till
cylindrical, and merely allow the stars to drop in without forcing them; this will
keep the top edge of a wedge shape, fig. 37, and answer the purpose of a
cone. If the paper is thin, the bag must be made of two thicknesses; the paper
will, then, require to be 9 inches by 6; 43⁄8 of this will have to be kept dry, and
the other 45⁄8 pasted.
A head, made in this way, if required to hold gold rains, or serpents, can
instantly be reduced to a cylindrical shape, by pushing the rocket-case right up
it, to the top; this will cause the corners to stick out, like two horns; press them
down, and secure them with sealing-wax. Attach a cone, if desired.
Heads made like figs. 38 and 39 may be of three thicknesses of paper, pasted
all over.
Fig. 39 shows the manner of tying on the sticks. No variation must be made in
their lengths, and it is not advisable to alter their size. To adapt it to the wood,
however, a slight alteration might be permitted. For instance, instead of 3⁄8
square, it might be 21⁄2/8 by 31⁄2/8, a slight increase one way, compensated by
a slight diminution the other. It must, however, on no account, be so increased
and diminished, as to approach the shape of a lath, as such stick would vibrate,
and cause the rocket to quiver. When the heads are a paper bag of the shape
of fig. 37, the stick must be tied on, as indicated by the dotted lines.
If the stick is suited to the rocket, it will, when suspended on the finger almost
against the mouth, as at f, fig. 40, lie, not quite horizontal, but slightly sloping
downward. The wood should be dry pine, free from knots. The sticks are
generally cut with a carpenter's cutting-gauge. If the learner has a lathe he will
find a 6-inch circular saw convenient for cutting them.
For amateurs, a 6⁄8 rocket is a good size; large enough, and small enough. If 5⁄8
and 4⁄8 are made, two hollow drifts will be sufficient; for 3⁄8, one hollow drift.
Very small rockets, 2⁄8, are made for children; they are rammed solid, and a
hole is pushed up them with a bradawl.
Rockets in former times, before the present days of competition, were charged
in moulds. These were of gun-metal, bored truly cylindrical, furnished with
hinges, to open and admit the case; they were then screwed up, and might be
charged as hard as possible. Names were given them according to the bore of
the mould, that is, the external diameter of the cases; a 4⁄8 was termed an
ounce rocket; a 5⁄8, a two ounce; a 6⁄8 a quarter pound; a 71⁄2/8 a half pound;
a 9⁄8, a pound; a 12⁄8, a two pound.
These names were determined by the weight of a leaden ball of the same
diameter as the bore. Now a sphere of lead, 71⁄2 inches diameter = 90 lbs., or
1440 ounces; consequently one of 15 inches = 720 lbs., similar solids being to
each other as the cubes of their like dimensions: the latter sphere being twice
the diameter from top to bottom; twice the diameter from left to right; and
twice the diameter from front to back; 2 × 2 × 2 = 8.
The external diameter of the rocket being 71⁄2 inches, the internal would be 5
inches, and this would be a 90-pounder: hence, for an inch rocket, we have the
proportion,

53 : 1440oz. :: 13 : 11·52oz.
so that a 12-ounce, or 3-quarter-pound rocket, ought to be a trifle above an
inch.
Required the weight of a 6⁄8 rocket. 5 inches = 40⁄8.

403 : 1440oz. :: 63 : 4·86oz.

So that a 6⁄8 is a trifle too large for a quarter-pound.

If, conversely, we require to know the size of a half-pound, or 8-ounce rocket,

1440oz. : 403 :: 8oz. : 3200/9

and ∛(3200/9) = ∛(9600/27) = (∛9600)/3 = 7·08

So, properly, a half-pounder is a trifle over 7⁄8. The names in use enable the
makers to understand each other, but they are not mathematically correct, and
are of no utility to an amateur.
The following table shows the true weight of leaden spheres, the dimensions
being taken in inches.

Weight. Diameter. Weight. Diameter. Weight. Diameter.

1 dram ·264 1 oz. ·664 1 lb. 1·672

2 drams ·332 2 " ·836 1¼ " 1·8

4 " ·418 4 " 1·056 1½ " 1·91

8 " ·528 8 " 1·328 1¾ " 2·02

12 " ·604 12 " 1·52 2 " 2·112

And 2⁄3 of the above numbers multiplied by 8, give the correct names for
rockets in eighths of an inch.
Required the true size of a half-pounder.
(1·328 × 2 × 8)/3 = 7·08.
A trifle above 7⁄8, as before stated.

A cast-iron ball 6 inches in diameter weighs 30lb.

Cast iron is about 40⁄63 the weight of lead.
5280 feet = 1 mile; 3280 feet = 1 kilometre.
 WHEEL AND FIXED CASES.
Wheel cases may be 55⁄8 inches long. Cut the paper, without waste, into 4
strips, each 29 inches long. Fixed cases may be 71⁄4 inches long: cut the paper
the other way of the sheet, into 4 strips, each 221⁄2 inches long. The extra
length, for fixed cases, is to allow of their being reported; that is, filled at the
end, with an inch, or more, of grain powder, to make a bang like a squib. The
cases are to be gauged to the thickness of roman candle cases, but choked like
rockets; 5⁄8 internal, and 7⁄8 external, are a good size. Two solid drifts will be
required for wheel cases; one 51⁄2 inches long, besides the head; the other 3
inches: for the fixed cases it will be necessary to have one of 71⁄4 inches. They
may be turned of box, beech, or ash; but gun-metal drifts are best, though by
no means indispensable. If they are cast in gun-metal, they will simply require
filing in the rough places. The cases are to be rammed solid throughout, on a
foot, fig. 42, turned in one piece, with a nipple, and a pin 1⁄3 of a diameter
thick, and just high enough to keep the choke clear.
In all wheel and fixed cases, whatever the remainder of the fuse may be, begin
by putting in one scoop of starting fire; this, when malleted firmly in, should fill
about 3⁄8 of an inch in the case. The subsequent fuse can be selected from the
Tables. The choke may be protected with a little clay, before the starting fire,
like rockets, if thought desirable: no clay is to be used anywhere else. Wheel
cases containing steel-filings are termed brilliant.
To prime the cases, very slightly damp some meal powder, by sprinkling it with
a few drops of water; mix and chop it up, or mince it, as it were, with a knife:
put a little into the mouth of the case, press the nipple, fig. 42, into it, and work
it round; this will prime the choke and the mouth at once, still leaving the choke
clear; or, paint the choke and mouth with the sash-tool, with meal paste, just
sufficient to wet them, then plunge them into dry meal, give them a rap to
shake off the superfluous dust, and lay them by to dry.
Wheel and fixed cases need not be choked at all, but plugged 1⁄4 of an inch,
with plaster of paris, as directed for roman candles. When dry, charge them by
setting them flat on the block, without using the foot. Mallet in a scoop of
starting fire, then the other. Afterwards, when a number are charged, bore
through the centre of the plaster, with a shell-bit, 1⁄3 the diameter of the case;
insert a piece of match in the hole; wash the face of the plaster over with meal
paste, and plunge into dry meal. A sheet of double-crown makes 8 wheel case
envelopes (2.2.2) 10 by 71⁄2, so that the void at each end, to receive the
match, is nearly 1 inch.
It is not possible to devise a formula that will indicate the exact quantity
required for wheel cases, as the fuses vary; but, representing the length, in
inches, by I; and the diameter in eighths by e, the following will help to serve
as a guide.

(I e2)/7 = drams.
Suppose a wheel case 55⁄8 inches long, and 4 eighths diameter,

(55⁄8 × 4 × 4)/7 = 90⁄7 = 13 drams.

Suppose a fixed case 71⁄4 inches long, and 5 eighths diameter,

(71⁄4 × 5 × 5)/7 = 181

⁄7 = 26 drms. = 1 oz. 10 drms.
 GERBES.
Gerbes, so called from the French word for wheat-sheaf, which they resemble,
are fixed choked cases: they do not show well on wheels. As they contain
grains of iron, they must not be under 6⁄8; for private exhibitions a good size is
9
⁄8, 9⁄8 internal diameter; 12⁄8 external; 3⁄8 the diameter of the choke; 111⁄4
inches, length of case. Charge the case on a nipple, exactly like a wheel case. It
is advisable to put in, first, a little clay, to protect the choke, as the fire, being
fierce, would, otherwise, enlarge it, and diminish the ascent of the sparks. Upon
the clay drive in a scoop of starting fire, and fill up with gerbe composition.
This, when containing iron borings, is termed Chinese fire; the pieces, Chinese
trees.
The most magnificent of all, however, is the coloured gerbe. For this, some
green, blue, and crimson grains, or small stars must be prepared. They may be
quarter-inch cubes, cut as directed for the chopped nitre stars. A far better way,
however, of preparing them, is to remove the pin a, of fig. 2, up to c, so that
the stars, driven in the tube, will be 5⁄8 inch diameter, and 3⁄8 thick. When these
are dry, chop them into 4 pieces, by holding a knife, or chisel, across them, and
giving it a smart blow with the mallet. They are harder made this way.
To charge the cases: having driven in the clay, and the starting fire, put in 7 or
8 stars, then a scoopful of fuse No. 1 or 2, then 7 or 8 more stars, and another
scoop of fuse; mallet the whole 4 layers down firm, with blows not too heavy at
a time, but many times repeated. Then put in 7 or 8 more stars, another
scoopful, 7 or 8 more stars, and another scoop, and mallet the 4 layers as
before; and so repeat. The two layers and the two scoopfuls may fill up, when
malleted in, 3⁄4 of an inch, in a 9⁄8 case.

Instead of choking the cases, plaster of paris is far preferable. Let it be 3⁄8 of an
inch thick, and well dried, before charging. After the cases are charged, bore a
hole through the plaster, 1⁄3 of a diameter, that is, with a 9⁄8 case, 3⁄8 of an inch
diameter. For this purpose it is not necessary to have another shell-bit; bore it
with the 3⁄16, and enlarge it with a penknife. It is better arched under, till
conical, as shown in fig. 48. Prime with 4 or 5 pieces of match, and wash with
the sash-tool.
The gerbe being finished, make a cylindrical box, or paper bag, of 2 or 3
thicknesses of paper: fill it with a number of crackers, and a scoopful of meal
powder, and fasten it to the gerbe.
No single piece is more effective than a coloured gerbe; the stars will be
projected 30 feet, or more: they may be put in, mixed; or, one layer may be
blue, another green, another crimson. Twist a piece of wire, deprived of its
elasticity, round the neck, and another piece round the bottom, and leave long
ends; it can then be fastened by them to the top of a post. To remove the
elasticity from iron wire, lay it in the fire till red hot; withdraw it with the tongs,
and put it aside, to cool slowly. If copper wire is used, it will bend without
preparation. String must not be employed, as it might burn, and let the case
fall. Common pins, patent short whites, deprived of their elasticity, are useful
for connecting the parts of lustres together.
 FLOWER POTS.
These are choked cases, charged with spur fire: the fire is somewhat slow, so
the cases must be short: 4 inches long, and 5⁄8 diameter is a good size. Rub the
composition thoroughly up in the mortar; the vegetable black produces
beautiful star-like sparks, totally dissimilar to any other. Put a little composition,
at a time, into the case, and jolt it with the roman candle rammer.
Vegetable black, introduced into a star, causes it to tail, like linseed oil. Light
such star on the hob; it will burn, and leave a residue, unaltered in shape; blow
upon this continuously with the mouth, or, better still, with a pair of bellows:
the supply of oxygen will cause it to boil up, in a state of fusion, when it will
begin to throw out clusters of the peculiar starlike sparks, before mentioned,
bright and yellow as new sovereigns.
Vegetable black is a pure lamp black; some samples of lamp black make equally
good stars, but others are worthless. Greater reliance can be placed upon
vegetable black. Vegetable black and lamp black must not be mixed with
linseed oil, as such mixture is liable to spontaneous combustion.
Roll up a tube for pill boxes, of two thicknesses of brown paper. When dry, cut it
into pieces about 11⁄8 inch long: choke one end, like a wheel case; set it on a
nipple,and charge it with spur fire, till full within 1⁄16 of an inch: fill up flush with
a little plaster of paris, pressed in flat with a knife: prime the choked end, and
put a number of such cases into a rocket head, or shell.
 PORT FIRES AND SHELL FUSES.
These are unchoked cases, like roman candles; 6 inches long, 3⁄8 internal,
41⁄2/8 external, is a good size. They should be rammed, as hard as possible;
and, for this purpose, it is best to have a mould. Now, if a case is rolled of such
a size that it will exactly fit into a brass tube, and is charged, in it, very hard, it
will swell, and it will be almost impossible to get it out again; but if it be made a
trifle smaller, so as to just slip through the tube; then, if a piece of writing
paper be rolled, dry, round it, once or twice, so as to make it a tight fit, and the
case is charged, it can be pushed out, like a pellet from a popgun, leaving the
writing paper, generally, in the tube, or mould, and the case will come out
without a wrinkle.
Let the composition be put in, very little at a time, and well driven with a solid
rammer and mallet. Fig. 59 represents the mould; the foot b fits the tube a; the
tenon c fits the case; a wire, d, goes through 2 holes in the brass tube, and a
hole through the foot; a nut, e, to keep the wire from jarring out, is made of a
piece of indiarubber: make a hole through it, with a bradawl, and slip it on the
wire; or, a screw-eye may be passed through and held with a leaden, or
wooden nut.
 TOURBILLIONS.[A]
A tourbillion, so called from the French word for whirlwind, is a case made to
rotate and ascend at the same time, forming a spiral of fire, and ending in the
shape of an umbrella.

To Make a Tourbillion.
Roll the case like a roman candle case, but gauge it to the thickness of a rocket
case. Let the inner diameter be 6⁄8; the outer 9⁄8; the length of the case 71⁄4
inches, fig. 43. To charge the case, have a mould, as directed for port-fires; and
let the tenon rise exactly 3⁄8 of an inch up the case. Put in a little composition at
a time, and mallet it as firmly as possible, till within exactly 3⁄8 of an inch of the
top of the case; so that there will be a vacancy of 3⁄8 of an inch, at each end.
Fill each of these ends flush with plaster of paris. It is, better, too, if you can
manage to fill the middle half-inch of the case with plaster of paris. It can be
effected with care, and will hold the screw, hereafter to be described, more
firmly.
Construct a wooden box, fig. 44, consisting of a bottom and two sides only,
firmly screwed together. Each of the pieces of wood is to be 71⁄4 inches long,
and 1⁄2 an inch thick. The internal breadth of the box is to be exactly 9⁄8 of an
inch; and its internal depth exactly 41⁄2/8 or 9⁄16, so that when the tourbillion is
laid evenly in it, and pressed down to the bottom, half of the case will be in it,
and half out of it. At a point b, fig. 44, on the top of the side, half an inch from
a, make an ink mark: and, at a point d, half an inch from c, make another ink
mark. Fig. 45 is the bottom of the box. At a point w, 5⁄8 of an inch from the
end; and, at a point z, 5⁄8 of an inch from the other end, make holes with a fine
bradawl, truly, in a line down the middle of the wood, as between side and side.
The distance w z is 6 inches; divide it into 3 equal parts, in the points x and y,
two inches asunder. Bisect x y in the point s. Procure 5 carpet pins, fig. 47: they
will, probably, be 3⁄4 of an inch long. Drive them through the holes w, x, s, y, z,
inverting the box for the purpose, so that they shall stand bolt upright in the
box. Now screw, or nail a piece of wood over the bottom of the box, entirely to
cover it, to prevent the carpet pins from getting displaced. It will be seen, that,
if the tourbillion be now laid evenly in the box, and pressed down till it rests on
the bottom, the projecting pins will make 5 holes, in the under part of the case.
While it is thus lying, with a stiletto, such as used by sempstresses, for making
eyelet holes, prick the side of the case over the line b, of fig. 44; and, also, over
the point d. There will now be 7 holes; 5 underneath, 1 to the right, and 1 to
the left: the latter are the places for the whirlers, or holes of rotation: 4
underneath, for the lifters, or holes of ascension; the centre one, s, receives a
nail or screw. Take a bradawl, fig. 49, 3⁄16 of an inch diameter; and slip over it a
shield, consisting of a piece of wood with a central hole up it, like a pop-gun, of
such a length, that, when it is slipped on, only 1⁄4 of an inch of the bradawl
protrudes; or, instead of a bradawl, fix in a handle, a wire of equal length,
namely 1⁄4 of an inch, and file it to a point. Push this into all the holes, except s,
making 4 holes underneath, and 2 horizontal holes, one left, one right: all these
holes will be exactly of the same depth, on account of the shield: see that they
are bored perfectly true, the horizontals exactly 90 degrees above the others, or
1
⁄4 of the circumference.
The next thing required, is a piece of hooping, or curved stick, about 6⁄8 of an
inch broad, and as long as the case, 71⁄4 inches. In the centre of this, bore a
hole, and countersink it; then, with a screw, an inch and a quarter long, screw
the hooping, at right angles, on the bottom of the case, through the point s of
fig. 45, which must be enlarged to receive the screw. A touch of glue may still
farther hold the wood in position. It will now assume the shape of a cross, like
figs. 50 and 51. Fig. 50 shows the under side of the case; fig. 51 the upper.
From w to x lead a bit of naked match; push the ends into both holes, and
secure with a little wetted meal, pressed in with a knife. Do the same with y
and z. Paste a piece of double-crown, 3⁄4 of an inch broad, and of sufficient
length, and cover each of the two pieces of match, with two layers of the paper.
Turn it over, like to fig. 51; connect the holes a and b with a bit of naked match;
and, under the centre of it, slip another piece of naked match, having a piece of
touch-paper round the protruding end: cover the match with two thicknesses of
pasted paper, in the same manner as the under holes. The tourbillion is now
complete. See that it will balance, and swing round easily, when laid upon a
level surface. The proper way to fire it, is from a flat sheet of iron, or a
flagstone. Light the touch paper; the fire will communicate to the side holes,
and set it in rotation. As soon as 1⁄8 of an inch of fuse has burnt from each end,
and the piece has got well into action, the 4 under holes will catch, and cause it
to ascend.
Instead of this mode of making a tourbillion, some charge it with an inch of
solidly rammed clay, in the middle; fasten the stick, by crossing it with binding-
wire; bore a hole through the middle of it, and of the clay, and slip it over a
tapering-wire, standing upright in a block, like the spindle of a rocket. Four
holes only are then used; two of rotation, and two of ascension; and the whole
are fired at once, the match starting from one of the under holes, going to the
side hole; over, across, to the other side hole, and on to the other under hole.
Instead of making them with clay in the middle, there is yet a better method of
having two cases, each about 4 inches long; and gluing, or fastening them with
tin-tacks on a centre-piece, turned with a tenon at each end, fig. 52; two
balancing arms, one on each side, must then be fixed to the centre-piece.
In a windmill, as is well known, the vanes, or sails, are set at an angle. There is
a toy, made of two slips of tin, forming a cross, and set at an angle, sloping
upwards, called the flying dutchman; this, when spun with a string, from a
handle like a humming-top, flies up into the air, on escaping from the string.
Steel Fliers, with two vanes, are used by sportsmen to practise shooting flying.
Small balloons, some years ago, in a room in the Polytechnic Institution, free
from a current of air, were guided or driven by a similar contrivance, moved by
clockwork. The screw-propeller of a ship acts on the same principle. I think it
possible that, if two vanes were fixed in the central piece of wood, set at an
upward angle of 10, 15, or 20 degrees from the horizontal, they might assist
the ascension, and so cause the tourbillion to reach a greater height; or, the
vanes alone might cause it to rise, upon 4 side holes, two to the left, and two
to the right, causing it to rotate. The design is shown at fig. 54, the shape of
the vane at fig. 53. I have not yet tried it, so offer it only as a suggestion.
 FOOTNOTES:
[A] Tourbillion, from tourbillon, like postillion, from postillon, the i being inserted to approximate the
pronunciation of the French. In pavilion, from pavillon, and vermilion, from vermillon, one l is
dropped; so in battalion, from bataillon; while medallion, from medaillon, retains the ll.
 SAXONS.
These are unchoked cases, charged like a tourbillion, but pierced only with holes of rotation, for the
purpose of turning a coloured fire. Drive them in a mould, as directed before. A good size is, 5⁄8
internal, 7⁄8 external, 6 or 7 inches long. Let the tenon enter the case 3⁄8 of an inch: charge the
composition firm till within 1⁄2 an inch of the top, which leave vacant. Remove it, and fill the 3⁄8 occupied
by the tenon, with plaster of paris. Have a centre-piece, turned like fig. 55, with a tenon, 5⁄8 diameter at
each end, 1⁄2 an inch long. Glue a case on each tenon. Let the centre-piece be 6 inches long, exclusive
of the tenons; so, if the saxon cases are 7 inches long, each, the entire length, as fig. 56, will be 20
inches. Make a hole at a, and another at c, with a shielded bradawl, 3⁄16 of an inch diameter. Put a bit of
naked match in the hole a, carry it round x and y, along to c and on to z. It must be pushed into c with
a blunt wire. Cover it with two thicknesses of pasted paper, like the tourbillion. Leave the match
exposed at x, y, and at z, and brush it over with meal paste. If the central piece of wood were now put
on a horizontal spindle, and fire communicated to the match z, the holes a and c would cause it to
rotate, and produce a white circle of fire. This, however, would be hardly worth making; but, by fixing at
b, a little case of coloured fire, a splendid effect is produced. This case of colour is usually tied to a nail,
driven in at b; but a little tenon of wood may be glued there instead, and the case of colour must be
then charged, with a vacancy at the bottom, to fit on the tenon. The case of colour must be timed to
burn as long as the saxon; rather more than an inch will be sufficient.
Saxons are sometimes made by charging a roman candle case with an inch of clay in the middle, and
boring a hole through the clay, to receive a spindle. Only one half of the case burns at a time; a leader,
placed at the bottom, near the central clay, conveys the fire to the other end; and continues the
rotation. For distinction, they are called Chinese fliers.
 FIVE-POINTED STARS.
These are cases about 21⁄2 inches long, and 1 inch diameter. Make a bottom to the case with 1⁄4 inch
thickness of plaster of paris, so that it looks like a large pill box. Charge it solid, and at 3⁄8 of an inch
from the extremity, that is, 1⁄8 of an inch beyond the plaster bottom, round the circumference make five
holes, as for saxons; run a bit of match round, connecting the holes. These, when fired, stand out at
right angles, the plaster towards the spectator, so that the fire resembles a gas star, with 5 points, as in
fig. 130.
 SQUIBS.
Take a sheet of 60-lb. imperial brown, and a sheet of 12-lb. white demy. The imperial, as said before, is
29 by 221⁄2. Cut it into 24 equal parts (2, 4, 3), that is, first into two equal parts, down the natural fold
of the paper; then each into four equal parts, at right angles to the first fold; and each of these into
three equal parts, at right angles to the second fold. Each piece will then be 55⁄8 by 45⁄8. The demy is
221⁄2 by 171⁄2. Cut the sheet into sixteen equal parts (2, 4, 2), each piece will be 55⁄8 by 43⁄8. A piece of
brown and a piece of white will make a case 55⁄8 inches long. For a former, procure a piece of brass
wire, or stair-rod, about a foot long, and 1⁄4 inch diameter. Lay eight pieces of the demy evenly on each
other; draw the thumbnail of the right hand from the farther edge of the paper straight over the middle
towards you, a few times. If properly performed, this will draw piece behind piece; proceed till about 1⁄8
of an inch of each is left exposed, in the same manner as a pack of cards would arrange themselves, if
set upright, and allowed to fall forwards; something after the manner of the laths of a venetian blind, or
slates upon a roof, imbricated. If you cannot acquire the knack of doing this, you must so lay them, one
by one. Paste the edges of all the eight pieces thus lying. Place one of the pieces of brown paper before
you; lay the former, or stair-rod, across it, nearly in the middle; bend the paper over it, and press it in
with the fingers of both hands; roll for an inch, or so; lay it on the middle of a white piece; bend the
white over; infold the brown in it, and roll forward, till it catches the pasted part, and sticks. After a
number have been prepared, and are dry, choke them, as directed for rockets. The wire of the choker,
fig. 24, should be about 1⁄16 of an inch thick. Now take a piece of square steel, or iron wire, 12 inches
long, the thickness of the wire being about 3⁄5 the diameter of the stair-rod, that is, 1⁄4 x 3⁄5 = 3⁄20 of an
inch; if it be found difficult to procure square wire, file a round piece. Bore a hole down a bradawl or
chisel handle, 2 inches in depth, and 1⁄4 diameter; fix the square wire in it, with melted lead, like fig. 4;
the lead is for the purpose of giving weight to the blows in charging. Black the wire all over with ink,
and allow it to dry. The next requisite is a tin funnel, without a neck, of the size and shape of fig. 65.
Any tinman will readily make one to order; but if the learner procure a soldering-tool, he can construct
one himself. For this purpose, dissolve a piece of zinc in a little hydrochloric or muriatic acid, till the acid
is saturated: heat the tool, and dip the tip end, momently, into it; the acid combines with the oxide of
copper formed by heating, and the zinc adheres to the clear surface of copper produced; it will now
easily take up the solder: the joint to be soldered must be clean, and also touched with a feather dipped
into the acid. It is best to make a funnel with a piece of writing-paper, first, for a pattern; when this is
got correct, the tin can be cut according to it. By describing a circle with a 3-inch radius, and cutting out
a sector of 100°, the correct size is obtained at once; 1⁄8 of an inch breadth, outside the radius, is to be
allowed to lap over. The bore of the squib is 5⁄20 of an inch; the size of the ramming wire 3⁄20; the hole
at the bottom of the funnel may be half way between, 4⁄20 or 1⁄5 of an inch diameter. Push the wire
rammer through the funnel, till it protrudes 2 inches beyond the bottom; observe the part of the wire
which is now level with the top of the funnel; withdraw it, and file a bright mark round the part: it will
be about 41⁄2 inches from the end. The wire, having been inked, shows the bright mark more plainly.
Make a deal box, 3 inches square, and 4 inches deep, and nail a bottom to it. Also have a little nipple,
fig. 73, with a wire, as drawn; it can be secured to a flat board.
To charge the cases. Set the choked end of one on the nipple, insert the funnel in the other end, put
the wire rammer down through the funnel, and let it fall to the bottom of the case: put in some fuse,
and jolt the rammer up and down, till the case gets so far filled, that the bright, filed notch, before
described, is on a level with the top of the funnel. It is necessary to sit at a table to charge the cases;
the eye is, then, on a proper level to see the mark. As the cases are thus charged, set them in the
square box, choked end downwards, till a number are filled. Now take out a handful, invert them over a
sheet of paper, and give them a few taps with the rammer; this will shake out a great deal of loose
composition, that has got puffed up, in the case. Return them to the deal box.
The next operation is to bounce, or bang them. For this purpose take a long slip of paper, 7 inches
broad. Stretch it lengthwise before you. Lay a number of the cases upon it, so that all the choke ends
lie flush with the left edge of the paper; then roll them up in the paper; turn and set them upright on
the choke ends; the paper will now stand up more than an inch above the cases, as a tumbler, three
parts full of water, stands up with an empty space above the water. Rest them on a sheet of paper, and
pour in a quantity of F grain powder; this will fill every case; loosen the paper wrapper, and allow the
superfluous grain powder to fall on the paper. Set the cases, choke downwards, in the square box. As
they are all full, it is necessary to get a little out of each. For this purpose, lay the square box,
containing them, horizontally on a sheet of paper, on one of its sides; turn it gently over, and lay it on
the next side; a little of the gunpowder will spill out; turn it gently over again, and lay it on the next
side; and so proceed, till every case has about 3⁄8 of an inch empty. The ends of the cases are now to
be closed. Take a case, in the left hand, as in fig. 63, and wind a piece of string, or whipcord, three or
four times round it, holding the end of the string, a, firmly with the thumb; then, with the right hand,
bring the end, b, back over all the folds, as in fig. 64; pull the end, B, tight, and the case will be closed.
Melt some common bottle sealing-wax in a pipkin; carry it to a distance from the fire, and dip the ends
of the squibs into it. Next, prime them, by pressing the choked ends into very slightly damped meal
powder, as directed for wheel cases. Take a piece of touch-paper, 15 inches by 10. Divide it into 48
pieces (2, 2, 3, 4); each piece will be 21⁄2 by 11⁄4. Roll a piece round the primed end, twist to a point,
and fasten it with a bit of carpet thread, hitched three times round it.
 SERPENTS.
These are simply squibs, made short, in order to burn out quickly, as they are intended to be thrown
from mines, and would otherwise lie too long on the ground. They are best made entirely of brown
paper, pasted all over, exactly like wheel cases, and choked in the same manner. Cut the paper 6 inches
by 21⁄2, and roll them on the squib former, so that they shall be 21⁄2 inches long. Ram them with the
funnel and wire, but continue the ramming till the guide-mark stands a little higher than the level of the
funnel; until, in fact, the cases get half full. Bounce, close, and dip them in melted wax, like squibs. The
best way to prime them is to paint them with the sash-tool, slightly, then plunge them into dry meal, so
that they may be, as nearly as possible, like quickmatch. If they were pressed into wet meal powder, of
the consistency of mortar, and then rubbed smooth on a Dutch tile, or slate, or plate, they would dry
hard and glazed. In this case, if they were fired from a mine, in all probability three-fourths of them
would miss. In brushing them with the sash-tool, with meal paste, turn the brush well round, to leave
little in; so that, after being pressed into the dry meal, they may not be choked up, but the cup shape
still be left. They are not to be touch-papered, being intended only for mines, rockets, or shells.
 PINWHEELS.
Procure some 16-lb. double-crown white paper.

A sheet is 30 × 20 inches.
Cut off a strip 30 × 3
Leaves a piece 30 × 17

The small strip may be used for odd purposes. Divide the 30 by 17 into 10 strips, each 3 by 17, for
penny pinwheels. If the paper were cut the other way of the sheet, it would not wind smoothly, but
crumple up. The fibres, from some cause or other, appear to arrange themselves in one direction, like
the grain in wood. For halfpenny wheels, divide the sheet into two pieces, each 10 by 30; then cut each
of these into 12, each 10 by 21⁄2. For a former, have a straight piece of iron, or steel, wire, 1⁄8 of an inch
thick, and 24 inches long. One end of this must have a basil, or sloping enlargement upon it, which is
thus made. Take a piece of double-crown 4 inches square; lay it straight before you, and cut it
diagonally, from the right top corner to the left bottom corner; remove the left half; paste the right; lay
the wire upon it, so that the rectangular corner is 5 inches from the right end of the wire; roll it up, and
press it smooth; the sloping edge will now form a spiral, commencing at 9 inches from the right end of
the wire, and being enlarged at 5 inches from the end. Paste a thin strip of paper over the whole, to
protect the spiral edges. It will now be something of the shape of fig. 27, only turned the reverse way.
The 5 inches at the end are for a handle. To roll the pinwheel pipes, lay the strips of paper evenly upon
each other, and work them back with the thumbnail, as before. Paste the edges; lay half-a-quire of
double-crown, or of blotting-paper, flat upon the table, to roll on. Place one of the strips of pasted paper
upon it; lay the wire upon it, the basil end being towards the right hand; the wire is to be laid, not quite
parallel with the near edge of the paper, but slightly sloping, about an inch and a quarter distant from it
on the left, and not quite an inch on the right, or basil end. So much of the basil must lie on the paper
as will make the mouth of the pipe large enough for the nozzle of the funnel to enter half-an-inch. The
basil, of course, gathers up the paper more quickly than the wire, and brings the edge straight, on the
completion of the pipe. Now to roll, bend the paper over the left end of the wire, press it in with the
fingers, and begin immediately to move it forward; at the same time draw the fingers of the right hand
rapidly along, pressing in the paper, and rolling forward, keeping the right, or basil, end of the wire
tight, and slightly pulled towards you: the paper round the left end of the wire will have thus become
rolled half round the circumference of the wire, before the basil end has begun. There is very great
difficulty, at first, in rolling so long a pipe as 17 inches; it is advisable to begin with the halfpenny pipes,
which are much shorter; the same wire-former will serve for both. If you cannot succeed, get a person
to help you; sixteen fingers can manage the matter easily; but it is best to master the difficulty yourself.
Have a funnel as nearly the size and shape of fig. 66 as possible. It is best with a lid soldered on the
top, with a circular hole in the middle, about the size of a shilling, to prevent the composition from
getting spilt. A slightly tapering neck is also to be soldered on at the bottom. Take especial care that the
mouth of the funnel dips into the neck, and not have the neck sticking up in the funnel. The joint inside
must be perfectly smooth. A charging wire will now be required; this should be a square steel wire, as
large as the neck of the funnel will admit, to move easily up and down; let it also have a piece of lead at
the end, to give it weight. Take a piece of brass tube, about 3⁄8 of an inch diameter, and an inch long.
Hold the wire in it, with a piece of flannel, or woollen cloth, and pour in melted lead. To charge the
cases, first bend a quarter of an inch of the small end of the pipe, over the edge of a knife, into a hook,
to prevent the composition running out; insert the nozzle of the funnel in the enlarged end; hold the
part of the pipe now round it, with the thumb and forefinger of the left hand, slip the wire, through the
funnel, down to the bottom of the pipe, which must rest, for a moment, on the table; pour in the
composition, jolt the wire up and down; lift the pipe from the table, keep jolting the wire, at the same
time turning it round, and let the pipe swing slightly to and fro, till filled. A guide-mark must be filed
round the wire, as for squibs, at such a distance as that, when it is level with the top of the funnel, the
bottom may protrude about 3⁄8 of an inch. When a number of pipes are charged, close the large, or
basilled, end, which is the one to be lit, by means of a piece of whipcord, or thin twine, wound round it,
as directed in figs. 63 and 64. Soak a towel in water; wring it as dry as possible; spread it out flat; lay
the pipes in a row, side by side, upon it, like rushes in a chair bottom; roll them up in it, and leave them
for 10 or 15 minutes. Then wind them upon the usual circular blocks, and fasten the end with sealing-
wax. Be careful that the wax is not in a flame at the moment of touching the case; if necessary, blow it
out. Accidents will arise, both from the wax and from the candle, if care is not used. Measure, with a
tape, round the wheel, now wound; suppose 4 inches; cut a piece of blue double-crown, 4 inches
broad, and any length: cut this into strips about 3⁄16 of an inch broad, and 4 inches long; paste a slate
all over; lay these strips, 7 or 8 of them, side by side, flat upon it; paste their upper surface; lay one
across each pinwheel, and bend it over to make the ends meet in the centre on the other side. If you
wish the pinwheel to be of double or triple size, after you have charged one pipe, paste the small end of
a second pipe, outside, and stick it in the other. Let it dry; then charge the additional pipe. Observe, the
pipes must not lie too long in the towel; if the nitre gets dissolved, it soaks into the paper, which
becomes, virtually, touchpaper, and one pipe ignites the other. The wire formers must not be allowed to
get rusty; therefore, before laying them aside smear them with tallow, or olive oil. The wire is much
facilitated, too, in delivering, if, just before using it, it is slightly oiled, and then wiped, apparently,
perfectly dry. After the wheels are finished, they should be spread out to dry, as soon as possible. It is
not usual to prime them; and, as sold, they are very difficult and troublesome to light; for private use,
they may have a short piece of match inserted in the mouth, and a small bit of touch-paper wound
round it. French nails are now made, of wire, very thin; those, about an inch long, are much better than
pins, for firing the wheel upon, as they have larger heads, and prevent its falling off. Pinwheel blocks
may be obtained of Merrick, Shuttle Maker, 155, Bethnal Green Road; the same person sells triangular
and vertical wheels; caprice, furilona and pigeon frames; rocket sticks; mine bottoms; saxon centres,
&c.; also, plait mills, made to order.
 CRACKERS.
A good tough paper for these is 24 mill, 19-lb. double-small hand.

A sheet is 30 × 20 inches.
Cut off a strip 30 × 4
Leaves a piece 30 × 16

The strip 30 by 4 may be used for squibs, or any odd purpose. For penny crackers cut the 30 by 16 into
8 strips, each 33⁄4 by 16. For halfpenny crackers divide the sheet into 2 pieces, each 10 by 30. Cut each
of these into 10, each 10 by 3. For a former have a straight piece of iron, or steel, wire, 3⁄20 of an inch
diameter, and 24 inches long. This will not require a basil. Have, also, another wire, of half the
thickness, for an opening wire. Lay the strips on blotting-paper, as before; and place the wire parallel
with the edge of the strip, as there is now no basil. When a number of pipes are rolled and dry, hold
one flat on a table; and, with the handle of a knife, or tooth-brush, rub the pipe flat along, all except
the first half-inch, held in the left hand, to form a little cup. Now push the opening wire through it, to
partially open it. Bend the bottom, over a knife, into a little hook, as before directed, for pinwheels. Take
up a quantity; wind a strip of paper round them, as in banging squibs, and pour in a quantity of FFF, or
canister powder. Unless the powder is very fine it will not run down. Pipes may be filled with F grain
powder, one at a time, by putting a funnel into the cup part, and tapping the pipe with a wire. You can
hear the powder run down, and easily judge when the pipe is full. The powder is now to be crushed
into meal, by rolling the pipes through a plait mill, fig. 71. This is furnished with a wooden screw, at
top; the screw presses upon a cross-bit, which bears upon two movable collars: these rest on the axis
of the top cylinder, by which means the pressure can be adjusted as desired. The cylinders of plait mills
are of beech, or boxwood; the latter are, of course, the best; but for making great numbers of crackers,
it is necessary to have the frames of iron, and the cylinders of steel. If a mill cannot be procured, the
pipes may be rolled with a brass tube, or rolling-pin, or passed through an American mangle; but a very
good, and far less expensive plan, is to lay them on an anvil, or flat-iron turned upside down, and beat
them with a smooth-faced hammer; this is a tedious process, but it makes good crackers. The cup end,
having served its purpose, is to be flattened, and bent into a hook. Now lay the pipes in a damp towel,
like the pinwheels. Take two pieces of deal, each about 6 inches long and 11⁄2 inch broad: let the one
for the halfpenny crackers be 3⁄4 of an inch thick; the other, for the penny, 1 inch thick. Saw and chisel a
piece out of each, about 4 inches long and 1⁄2 an inch broad, as drawn, fig. 67, so that it looks like the
first and third fingers of the hand held straight up, with the second finger, between them, shut. Next
procure 20 pieces of wire, each 2 inches long, and about 1⁄16 of an inch thick. Support the cracker-frame
in a vice: lay the cracker across the opening; place a wire upon it, as at a, fig. 68; bend the pipe over it;
lay on it another wire, on the other side, as b; bend back, and lay on it the wire c; and so on,
alternately, till the cracker is bent up. Lift it out of the frame, and let the wires drop. The distance
between the wires, in the halfpenny crackers, will be 3⁄4 of an inch; in the penny, 1 inch; if made
according to the directions given. The outside of the frame should be very slightly tapered, or it will be
difficult to lift the cracker up out of it. The penny size will probably take 14 wires; the halfpenny 10
wires. Cut the end flush, to expose the crushed powder; wind round it a piece of touch-paper, about
13⁄4 inch long, 11⁄4 broad; it need not be pasted; fold up the cracker; pass a piece of flax or thread
twice round it; twist the thread in and out, backwards and forwards, among the folds; and the cracker is
complete. If the paper does not appear sufficiently thick to make a good report, cut the strips broader,
and fewer to the sheet; but always cut them up the short way of the paper, or they will not bend
properly. The paper for pinwheels and crackers is 30 inches by 20; the longest pipe that can be cut is 20
inches: as an experiment, try a pinwheel and cracker the other way of the paper. If the cracker is
intended to be put into a jack-in-the-box, shell, or rocket head, push a bradawl up the mouth, insert a
bit of quickmatch, to project a quarter of an inch, and secure it with a little wetted meal, pressed in with
a knife. The way of bending it is shown at fig. 69. It is not necessary that a cracker should contain very
much powder; the loudness of the bang depends more upon the thickness of the paper than upon the
quantity of the powder. So with maroons; more string and less powder, are better than more powder
and less string. A very good paper for crackers is 30-lb. royal cartridge; cut the strips 3 inches broad;
16-lb. double-crown may also be used, the strips 5 inches broad. Crackers for mines may be made with
brown paper.
To make a cracker with 30 or 40 bangs, it is necessary to join 3 or 4 pipes together, before putting in
the grain powder: the mode of making a joining will be understood by attending to the following
directions. Cut a piece of paper 3 inches square, and lay it straight before you. At one inch from the
right top corner, make a mark; at one inch from the left bottom corner, make a mark; draw an oblique
straight line from one to the other, and cut along it with a pair of scissors. Without disturbing the
relative position of the two pieces, draw the right piece a few inches towards the right; paste the farther
edges, as usual; lay the wire close along the near edge of the right piece, and roll it up; it will have an
external spiral; roll up the left piece in the same manner; it will have an internal spiral; when both are
dry, paste the external spiral, and screw it, as it were, into the other piece. If managed carefully, and
brought up till the edge of the paper forms a straight line, the joining will be as firm as if the paper had
not been cut, presenting only a spiral edge, going once round the pipe.

Leader Pipes.
These are for piping quickmatch; they are rolled exactly like pinwheel pipes, on wires of different
thickness, to suit the size of the match. They must be large enough to admit the match easily, without
much pushing, which would break it; otherwise, the more nearly they fit the match, the more rapidly it
blows through. The fire is conveyed from the tail of one case to the mouth of another, by a short piece
of pipe, d, fig. 80, with the match projecting at both ends; when the fire has to be conveyed to two
other cases at once, a fresh bit of piping is slipped on, and the match left exposed, as at a; or a piece is
cut out of the side, as at b; each end should be bent into a hook, as at c, to prevent its slipping back.
Wheel cases are to be enveloped by rolling a piece of double-crown twice or thrice round them, two
inches longer than the case, as fig. 82, shown by the dotted lines. Suppose you have six wheel cases
ready charged; lay six pieces of double-crown flat before you, and work them back with the thumbnail,
as before described; paste the 6 edges; lay the 6 wheel cases in a row, side by side, and draw the paste
brush across the middle, as if pasting your knuckles; then roll one in each envelope, so that it projects
an inch at each end. Put the leader pipe in, and tie it with a piece of carpet thread, thin twine, or waxed
yellow flax or hemp, as at fig. 83. To join one pipe to another, to lengthen it; suppose you have a piece
of match 40 inches, long, and two pipes of 20 inches each. Slip both the pipes on the match, so that
they touch in the middle; take the end of one, so touching, gather it round the match; pinch the end of
the pipe tapering, push it into the other, and bind a piece of pasted paper round, to secure the joint. In
cutting a matched pipe straight across, of course the match inside gets cut flush; do not leave it so, but,
with a pair of pointed scissors, cut away a quarter of an inch of the pipe all round, and bend the
exposed match, as at c, fig. 80, before putting it into the envelope of the case. Besides a strong pair of
pointed scissors, a small pair, 3 inches long, with rounded ends, to carry constantly in the waistcoat
pocket, will be found convenient for many purposes.
 MAROONS.
Take 3 inches of a 6⁄8 or 3⁄8 rocket case. Fix, in one end, a cork, half an inch long; put in 2 inches of F
grain powder; on this, another cork. Wind string tightly round it, lengthwise, 6 or 8 folds, side by side;
bend it to a right angle, and wind 6 or 8 more folds; and so on, till covered; then wind crosswise; and
again, at right angles, as in fig. 74. Dip it into melted glue, and put by to dry. When dry, make a
bradawl hole through one part, to reach the powder. Make sure of this, by inverting it, and letting a little
drop out. Insert a short port-fire, having a piece of match at the bottom, and touch-papered at the top.

Another Method.
Take a rocket case, 5 inches long, and unchoked. Put in one solid inch of plaster of paris. When this is
dry, pour in 3 inches of F grain-powder; on this put another solid inch of plaster of paris. When dry,
wind string tightly round it, up and down the cylindrical part, not the ends, till of two thicknesses. Paint
the string over with melted glue. Make a bradawl hole in the middle of one side; fasten to it a squib,
without a bang, having a piece of quickmatch at the end, to enter the hole. Cover the joining with two
layers of pasted paper. On the other side, below, fasten a short piece of deal, like a rocket stick, with a
point, to put into the ground, fig. 72. After lighting the touchpaper, remove to a distance; as bits of
string are likely to get driven into the face, on the explosion.
To fire a salute with maroons, at regular intervals of time. Charge a port-fire, and saw it into inch
lengths; envelope each in a piece of double-crown, 3 inches broad, and long enough to go thrice round
the port-fire. Hang the maroons to hooks, or otherwise suspend them, on a frame, a foot distance from
each other, as a, b, c, fig. 75. Underneath them fasten, with binding screws, or tie to nails, the port-fires
x, y. Connect the port-fires with one another, and with the maroons, by leaders, in the usual way. On
lighting at w, the first maroon explodes, and the first port-fire catches; this, having burnt, lights the
second maroon and the second port-fire; and the port-fires being of the same length, the intervals of
time between the explosions of the maroons will be the same.
 GOLD AND SILVER RAINS.
These are little cases, 21⁄2 inches long, rolled on a 1⁄4-inch former, and filled with the funnel and wire.
They may be primed like fig. 29 or 30, or like squibs. Put them, mouth downwards, into rocket heads.
 PEACOCK'S PLUMES.
These are a combination of rain and star. Roll them like pill-box cases, on a 3⁄8 inch former, about 13⁄4
inches long; charge one end, 3⁄8 of an inch deep, with coloured fire, driven in dry; fill up with gold, or
silver rain, with a film of shell fuse at the top, to bind. Cut a bit of match, 21⁄4 inches long, lay it outside
the case, so that it projects 3⁄8 of an inch at one end; envelope it in a piece of double-crown, fig. 90;
tuck in the paper, to press the match, at one end, on the colour; twist the other to a point. Both ends
thus light at once; and the rain appears, like a coloured star, with a tail.
To prime a case with match laid flat on the mouth. Take a piece of thread, or fine string, and fold, or
bend it, in the middle, as at fig. 76. Tie a knot near the bent end, as at a, fig. 77. Bring the knot, a, up
to the side of the case, as at fig. 78; pass the loose ends round, and tie in a knot, at b. Lay the bits of
match flat across, as at e, fig. 79; bring the threads together, and tie them at c. Or bend tape match
across, and tie, as at d.
To get a fine thread through a long pipe, or the hem of a bag. Take a piece of copper wire, and bend it
round at one end, as z, fig. 88. Pass the end, z, forward, and push it through to the other end of the
bag, &c., then bend it to the form of x, fig. 89; pass the string through the loop x, and pull the wire
back.
 SAUCISSONS.
These are a large kind of serpent, charged on a nipple, like a wheel case, with solid drift and mallet.
They may be 5⁄8 or 6⁄8 or larger; about 31⁄2 inches long. Drive in brilliant fire, or gerbe, fig. 84, 11⁄2 inch;
fill up to within 1⁄2 an inch of the top with F grain powder; and plug the end with plaster of paris, or a
bit of wood, fastened with a tack or two. Press a piece of touch-paper, or double-crown, into the shape
of a deep pill-box; fill it with F grain powder; fit it to the mouth of the saucisson; tie round the choke;
brush, with meal paste, the outside, at bottom, and dip into dry meal. These saucissons are to be fired
in a volley. Procure, say 2 dozen, iron tubes, a, b, c, &c., each a foot long; fit them with a wooden
bottom, fig. 92, having a tenon, t, an inch long; let it be fastened with a screw on each side. Bore a
hole through, to make a communication with the mortar formed by the tube. Take a board, an inch
thick, of suitable length and breadth; bore in it 2 dozen holes, of a size to fit the tenons; glue these in,
so that the tubes, or mortars, stand upright, in rows, side by side, like the pieces on a chess-board.
Invert it. Nail a rim all round, so as to make a box, 2 or 3 inches deep. Cut a groove from hole to hole
of the tenons; connect all the holes with naked match, also push a bit of match up all the holes in the
tenons; now fill the box with sawdust, and nail a board on, to serve for a bottom, and to keep the
sawdust in. Invert it; and put a saucisson, mouth downwards, into each mortar. Fig. 84 represents a
single saucisson; w, w, w, fig. 85, saucissons in the mortars. On firing the match at s, it is evident the
cases will be driven out rapidly, one after the other. The sawdust prevents the flash igniting the whole at
once.
 PEARL STREAMERS.
Paste brown paper all over, and roll up a case, of four or five thicknesses, on an inch and a quarter, or
an inch and a half former, like a rocket or other case; when dry, cut it in the lathe (see fig. 27) into inch
lengths—inch-deep bottomless pill-boxes. Set one on a foot, fig. 9, to enter about 1⁄8 of an inch; mallet
in the pearl streamer fuse, till nearly full, then a little meal powder; remove it from the foot, and press
in flat, with a knife, a little plaster of paris, to form a bottom. They will have the appearance of bungs;
fire them in volleys, like saucissons, from suitable-sized mortars. Primed end downwards, of course.
Match may be tied on, as in figs. 78 and 79.
 BLUE LIGHTS & STAR CANDLES, OR STAR LIGHTS.
These are little cases, charged with the funnel and wire; the latter are filled with spur fire.
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

textbookfull.com