Flyback
Flyback
Lab notebook
You should keep a written lab notebook while doing this and other experiments. It should contain
• Diagrams of circuits you simulated or built
• Waveforms, analysis, observations and explanations as prompted by this manual, and things that
you decided to pursue. Be quantitative.
• Enough detail to enable someone to quickly pick up the thread of what you did and observed.
Computer programs that deduce and solve all the equations (Kirchoff, Ohm etc.) of an electronic circuit
are widely used. If used wisely, they can be quite accurate; the typical reason for a mismatch with reality
is that reality does not live up to the idealisations of your circuit diagram.
These notes are written for the LTSpice simulator, but others are available, e.g. many other Spice
derivatives, and Qucs and its derivatives. They are mostly free software, and most run on multiple
operating systems. [spice = simulation program with integrated circuit emphasis; qucs = quite universal
circuit simulator]
LTSpice is built for Windows and Mac. It runs OK within Wine on Linux. Download the latest version
of LTspice (presently XVII) from www.analog.com at
https://www.analog.com/en/design-center/design-tools-and-calculators/ltspice-simulator.html
Check that you can get the same results as shown in section 2.1, and attempt section 2.2.
We begin with a circuit that you may be familar with, to check that we can reproduce its behaviour. Use
LTspice (or other simulator) to draw the circuit shown in Fig. 1.
The resistor, capacitor and ground symbol are selected from the toolbar and placed by clicking. Ctrl-r
(or the rotate symbol on the toolbar, that looks like EM) rotates the symbol before placing. Esc deselects
the part. There is an Undo option in the Edit menu. The voltage source is found as "voltage" in the
list revealed by the Component button (looks like an AND gate). The Wire tool from the toolbar makes
connections.
Use a right click to alter the values of the components from their defaults to: R1 = 1k (1000 Ω);
C1 = 1u (10−6 F); V1: Advanced, small signal AC analysis, AC amplitude = 1.0 (1 volt). Use the Label
Net tool (letter A) to apply vin and vout labels to the appropriate points in the circuit.
Select the simulation required from the Edit Menu » Spice Analysis » AC analysis: Type of sweep = Lin-
ear, 250 points, start = 10, stop = 100k. Place the .ac command that it generates on the schematic.
Save the circuit, and run the simulation by clicking the Run icon (running man). Any errors will
appear in a window. Click on the vout and vin wires to see their voltages plotted as a function of
frequency, see Fig. 2
The plot shows that Vin has a constant value of 1 volt at all frequencies, while Vout quickly falls to-
wards zero at high frequencies. A standard representation used in electronics is the Bode plot, which has
logarithmic amplitude and frequency axes, so right-click on the frequency axis to make it logarithmic.
The amplitude in decibels is automatically a logarithmic scale, but you may prefer to make it logarithmic
in voltage units. The linear sweep does not calculate enough points to make a clean plot at low frequen-
cies, so go back and change the ac simulation to use 50 points per decade (a logarithmic sweep), and
simulate again.
You can alter the appearance of the plots using the Control Panel tool (hammer) » Waveforms. Dark
lines a couple of pixels wide on a light background are advised. The result, Fig. 3, shows why this circuit
is called a low-pass filter.
Low frequencies (below 100 Hz, say) pass from input to output with little attenuation, but high
frequencies (above 1000 Hz, say) do not. It can be quite a useful circuit fragment, even though the cutoff
point is not that sharp.
Interchange the positions of the resistor and capacitor in the circuit of section 2.1 and produce graphs
showing the Bode plots for the new circuit. Comment on the results.
D EMO NOTES : Here is what the high pass filter plots should look like. There should also be a
labelled circuit diagram. Most important features are log/log for amplitude and linear degrees vs. log
frequency for phase. They can be separate or on the same plot, using left and right axes. Unfortunately,
LTspice does not produce such nice plots as Qucs, so we have to live with this. A generally tidy output
and extension of simulation to 1 Hz to show more of the curve are worth commending but not essential.
The crossover from blocking to passing occurs at the same frequency as the LPF. Also there is a phase
lead at low frequency, where the capacitor is a significant series impedance, in contrast to the phase lag
in the LPF case, where the capacitor is a significant
√ load at the high frequency end. When the (angular)
frequency ω = 1/RC, the attenuation is 1/ 2 and the phase change is 45◦ (in both cases).
The need to convert electrical power from one voltage level to another is very common. A USB charger
that converts 240 V mains to 5 V to charge a phone is one example, as are the step-up and step-down
conversions that the phone may need to power its display (∼10 V) and its processor (∼1 V) from its nom-
inally 3.7 V battery. In this section, you will investigate how this is done using inductors and electronic
switches.
Consider the circuit of Fig. 4. Its behaviour is governed by the expression for the voltage across an
inductor, VL = −L di/dt. Before time t = 0, the current i in the circuit is zero. At t = 0 the switch is
closed. By applying Ohm’s and Kirchoff’s laws, you should show1 that the current has the form of an
exponential approach to V /R,
V −t
i= 1 − exp , where τ = L/R.
R τ
1 Write down Kirchoff’s voltage law, separate the variables i and t, then integrate, subtitute the limits (0, 0) and (t, i), and
di
D EMO NOTES : L dt + iR = V ; di
dt = L1 (V − iR); di
V −iR = dt
L. Integrate to give
1 1 V − iR R iR t V h t i
− ln(V − iR)|i0 = t|t0 ; ln = − t; 1 − = exp − ; i= 1 − exp −
R L V L V L/R R τ
Thus the current builds up, at a rate determined by the L/R time constant, to V /R. The exciting part
happens when the switch is then opened and the current stops. If this is done instantaneously (dt = 0),
di
then L dt = −∞, so an infinite voltage appears at Vs . While this is a good way of generating an increased
voltage, a numerical circuit simulator cannot handle instantaneous infinities, and in any case they do not
occur in a real circuit, so we will next simulate something a little more realistic.
Set up a simulation of the circuit shown in Fig. 5. Set the values of the resistor and and inductor to 1 Ω
and 1 mH to give a time constant τ = L/R of 1 ms.
Si4892DY
Component: voltage
Simulate: Edit Simulation Command (right click, Advanced)
Transient Functions = PULSE
Stop time = 10ms Vinitial = 0
Von = 10
Ground
Ton = 1ms
Tperiod=2ms
The MOSFET M1 (see Fig. 6) acts as the switch in this circuit. When a positive voltage of a few
volts is applied to its gate electrode, it conducts strongly between source and drain. Without the gate
voltage, it switches off. Here we replace the electronics that would normally drive the gate by a source
of rectangular voltage pulses, V2. In contrast to this, voltage source V1 is set to a constant 1 volt, like a
battery.
drain
VDSS = 31V
RDS(on) = 0.01 Ω
gate ID = 50A
S
G D
source
Figure 6: Si4829DY MOSFET. RDS is the resistance when on; VDSS is the voltage at which the isolation
breaks down when off.
Because the circuit is neither linear nor time-independent, an AC simulation is of little help. Instead,
we use a transient simulation that numerically integrates the differential equations of the circuit.
By running the simulation, you should be able to make a plot like Fig. 7, which shows the switch
control Vg in blue, the steady rise of the inductor current in red, and a large voltage transient at Vd in
green. To make a plot like this, click on the Vg and Vd wires when the cursor changes to a voltage probe,
and click on the body of the inductor when it shows a little ammeter. Plot Settings » Add Plot Pane
allows you to add a second graph, then you can drag the curve labels between graphs.
Record the waveforms that you get, and carefully annotate your graph to show how you understand
the relationship between the opening and closing of the switch, the varying current in the inductor, and
the voltage that is generated. Be as quantitative as possible. What detemines the size and duration of the
voltage spike? The fat voltage trace near zero volts is worth exploring too. You can zoom into the plotted
data using the magnifying lens buttons on the toolbar.
D EMO NOTES : Things we’d like to see:
• Current rises as in sect. 3.1 while switch closed. After 1 ms, t/τ = 1, so expect 1 − 1/e = 0.63
times the final current of 1 A. Numerical results will differ slightly because the simulation correctly
models the on-resistance of the switch and the rise time of the generator.
• While switch is being opened (again, the opening time of around 20 µs is modelled correctly), the
current falls and a large Vd is generated throughout this time.
• The voltage does not exceed ∼37 V because the transistor starts to break down above 31 V (see
Fig. 6). Again, this is modelled correctly.
• The voltage is actually oscillating after the high voltage pulse (L1 and internal capacitances in the
transistor).
A continuous high voltage output would be more useful than pulses. Add a diode (from the toolbar), a
capacitor of capacitance 1u and resistor of resistance 50k as shown in Fig. 8. The idea is that the 1 µF
capacitor C1 stores the output voltage, while the 50 kΩ resistor R2 represents a load circuit that makes
some use of the high voltage power.
Simulate this circuit, plotting also the capacitor voltage Vc, which you should see does not fall away
like Vd. Annotate your plot and use it to explain when the diode conducts and when it does not, and
the shape of the Vc waveform. Also simulate for longer (Edit the .tran 10ms to be e.g. 30ms) and
explain the steady state behaviour.
D EMO NOTES : Diode conducts when Vd > Vc by about 0.7V (you can see Vd is clipped to about
0.7V above Vc in the simulation). This charges C1, but when Vd falls away, the diode cannot conduct
in the reverse direction, so C1 remains charged. Successive pulses at Vd add to the charge in C1. Vc
ramps downwards because it discharges by passing current through the load R1. If there were enough
time for C1 to discharge all the way, you’d see an exponential decay to zero volts, but the initial 2 ms
of this decay is almost linear. Running the simulation for 30 ms shows that eventually Vc levels out to
35-36 V with a steady ripple.
3.4 Efficiency
Ideally, all the power entering a circuit like Fig. 8 would be delivered to the load. Considering just one
of the pulses, compare the maximum energy stored in the inductor with the amount delivered to the
capacitor [the energy stored in an inductor is 21 LI 2 , that in a capacitor is 12 CV 2 ]. Also note the power lost
in R1 and comment on your result. You may find it helpful to use the cursor on the graphs to read values
(shown at the bottom of the window).
D EMO NOTES : Peak current in the inductor is 0.67 A, hence 2.24 × 10−4 J stored. Capacitor steps
from 34.8 to 36.25 V, an increment of 5 × 10−5 J (exact values will vary), so the switching process is not
completely efficient. Average current in the resistor is about 710 uA. Over the 2ms between pulses, it
dissipates the 5 × 10−5 J that was put into the capacitor – as it must, to be in a steady state.
A disadvantage of the circuit of Fig. 8 is that the output (e.g. 5 V dc USB power) cannot be isolated
from the input (e.g. derived from 240 V ac mains). This could be a safety hazard. Now we add a second
winding on the inductor to allow us to take power out of it without any electrical connection; the input
and output circuits are coupled only by their shared magnetic flux.2
Convert your model to Fig. 9 (the Cut tool [scissors] is useful for removing circuit elements). Tell
LTspice that the two inductors are coupled magnetically by using the Spice Directive button (looks like
.op) » Edit text on the schematic » How to netlist = SPICE directive to add the text K L1 L2 1 in the
text box. Place the generated text near the inductors. Notice the reversed direction of the diode. For
convenience in plotting, the Vc and ground terminals have also been interchanged. Run the simulator,
and examine how the signals compare to those in the non-isolated circuit of Fig. 8.
If the two inductors are not oriented the same way, as indicated by the circle near one end, only a
small Vc will be seen. Investigate this and explain why.
D EMO NOTES : V = −N dφ dt applies to both inductors, and the flux φ is the same in each of them,
owing to the 100% coupling (1) of the K directive, so voltages should be the same. The dominant voltage
spike has a certain sign, which has to match the orientation of the diode to make it into the capacitor.
2 The
term flyback relates to the movement of the electron beam in vintage cathode ray tube televisions; the magnetic flux
was used to deflect the electron beam as well as to generate a high voltage.
See if you can get more voltage out of the circuit. Hint: a step-up transformer needs more turns on
the secondary than the primary.
√
D EMO NOTES : Making L2 bigger by a factor N implies N more turns, and correspondingly greater
voltage, but to get very far with this, you also need to increase R2, since the power available is limited
by the current in L1 and the switching rate.
In the separate practical “Switch mode power supply” experiment, you can build a circuit very similar to
Fig. 9, and use it to generate a few hundred volts from a 1.5 V battery. A real mains power supply uses
the same principle, as shown in Fig. 10. You should be able to identify in it all the key elements of Fig. 9,
alongside its input power and control circuits.
Figure 10: A practical mains power supply (ON Semiconductor NCP12400 data sheet)
D EMO NOTES : Anyone moving to the switch mode lab experiment after this should go straight to
“section 4: Higher voltages” in that experiment.
Enthusiatic students should be able to pick out some of the main components of the realistic circuit, thus:
Correspondence between Fig. 9 and Fig. 10:
• L1, L2 & K - TR
• M1 - Q1
• D1 - D1
• C1 - COUT
• R2 - RLOAD
• V2 - IC1 and surrounding components
• V1 - Everything from ac input to CBULK
In case anybody asks: D2 and friends damp the inductor and keep the transient voltage under control.
Circuitry around OK1 senses the output voltage and tells IC1 to back off once it is high enough. No-
tice that OK1 is optically isolated, to avoid breaching the input/output electrical isolation. The bottom
winding on TR makes a secondary power supply for the chip (D3, C1).
Recreational reading
• http://www.righto.com/2012/10/a-dozen-usb-chargers-in-lab-apple-is.html
• https://spectrum.ieee.org/tag/switching+power+supplies
MSC 2020-10-13