Power Electronics Laboratory
AC Generation, Control and Transformers
By: Professor Steven B. Leeb
Massachusetts Institute of Technology
Purpose: Learn to analyze and build a resonant converter.
Understand and use silicon controlled rectifiers and TRIACS.
Understand and design a transformer and use one in a flyback converter.
Equipment: Agilent Network Analyzer
Introduction:
PRIOR to starting this labREAD or Skim as needed: Scherz pp. 181-190, pp. 108-118. HH pp. 328-329, pp. 316-320.
READ the “Impact of inductor core loss on your circuits,” the last page of this lab!
In this laboratory, you will design and build a fluorescent lamp ballast.
NOTE: You will design and build circuitry in this Laboratory that generates high voltages.
These voltages can injure you severely. You must have attended the in-class SAFETY LECTURE
to work on this lab. You must NOT energize (turn on) either the fluorescent lamp ballast or the
200-volt dc power supply by yourself, ever. You MUST have a staff member present any time that
you wish to turn on one of these circuits. You MUST wear safety glasses and observe all electrical
safety procedures discussed in class when energizing these circuits with a staff member. There will
be no tolerance for any breach of these requirements.
There are two key dates to be aware of in this lab. Please stay on top of the activities.
We will have a design review almost immediately for two of the three circuits you will eventually build,
the fluorescent lamp ballast and the 200-volt power supply. Design (but do NOT build) these two circuits
in Exercises 1 and 2. Bring the fruits of your design effort to the Design Review. If your design makes
sense, you will receive appropriate PC boards and cores from the staff, and build your circuits in
Exercises 3, 4, and 5 for testing with the staff and demonstration at the final check-off (CO). The DR is
intentionally early. We will need plenty of time to build your circuits and test them with the staff during
office hours.
Your grade on this laboratory will be determined as follows: Fifty percent DR; Fifty percent CO.
KEY CALENDAR DATES for the Laboratory
Design Review (DR)
October 17,18 (NOTE: VERY SOON!)
Final Check-off (CO)
October 27,28
EXERCISE 1: Fluorescent lamp ballast design (Do not build yet! Just design!)
A fluorescent lamp is very different from the incandescent lamps that you are probably used to using in
your home. In particular, a fluorescent lamp bulb cannot simply be “plugged in” to the wall. It requires a
special circuit known as a “ballast”.
A fluorescent lamp is a sealed glass tube with electrodes at either end. During the manufacturing process,
the tube is coated with a material called a phosphor, the white powder you may have noticed on the walls
of a broken tube. The tube is evacuated of air, and refilled with a very low pressure mercury gas. In
operation, an electrical arc or discharge fills the tube in response to a voltage difference between the
electrodes. The arc excites the mercury gas, creating a charged plasma that emits ultraviolet light, which
stimulates the phosphors and makes them glow.
The reason that the lamp cannot be simply “plugged in” to a voltage source, as in the case of an
incandescent lamp, is indicated by the V-I characteristic for a typical fluorescent lamp, e.g., see “Electric
Discharge Lamps” by Waymouth. When the bulb is “cold,” i.e., off and not producing any light, it
presents a relatively high impedance between its electrodes. A high voltage must be applied across the
lamp to initiate an arc. This is the “starting voltage” shown in the figure. Once the arc fires or “strikes”,
the gas in the tube begins to ionize. The impedance with the ionized gas is low compared to the
impedance of the cold tube. If the starting voltage is maintained across the tube, an enormous current
would eventually flow into the tube until something broke, e.g., a fuse blew, the filaments in the lamp
broke, a wire melted, etc. So, once the bulb has “struck”, the terminal voltage must be reduced to a point
on the equilibirium V-I curve that will produce the desired current and illumination. Achieving a stable
equilibrium point on the struck V-I curve is tricky. For example, in the useful current ranges where
illumination is produced, between 100 and 500 mA, an increase in terminal voltage corresponds to a
decrease in terminal current, and vice-versa. This happens because, roughly, as the current decreases in
the tube, the number of charged carriers in the tube also decreases, decreasing the conductivity of the
plasma column in the tube. So a higher voltage is needed to maintain the lower current! Increasing the
current, on the other hand, increases the conductivity of the plasma. A lower voltage is required in this
case to sustain the higher current.
This odd behavior makes it difficult to keep the lamp stable on the V-I equilibrium curve. Imagine, for
example, that we plug the lamp into a fixed voltage source and that, somehow, the lamp is struck and a
perfect voltage is applied that will keep the lamp on the equilibrium curve. Now, imagine a slight,
inevitable disturbance that momentarily increases the current in the bulb. This disturbance could be a
slight change in exterior temperature, for example. The voltage across the tube remains fixed, but now
we are “off” the equilibrium curve, with a larger number of charge carriers in the tube compared to before
the disturbance. Off the equilibrium curve, this voltage will push yet more current into the bulb, further
increasing the conductivity. If the voltage remains unchanged, the bulb enters a “runaway” condition
where the current increases until something breaks.
A ballast circuit is used to start the lamp and keep it at a stable equilibrium point once struck. One easy
way to solve the stability problem would be to put the lamp in series with a conventional linear resistor.
For a suitable value of resistance, this series combination would appear to have an overall positive linear
resistance characteristic. We could then apply a high voltage to strike the lamp/resistor combination, and
lower the voltage to run at a stable equilibrium point in steady state. The problem with this plan is that
the resistor dissipates power. This creates heat in the lamp fixture (could cause a fire) and also decreases
the efficiency of the lamp (very expensive when you have a lot of lights).
A practical ballast circuit might look something like this:
(a) Simple ballast circuit
In this ballast, a square wave produced by the input voltage source drives a resonant circuit that consists
of an inductor, a capacitor, and the lamp. When the lamp is cold and off, we can approximate it as an
open circuit. In this case, the square wave source is driving an LC resonant circuit. If the square wave is
at the resonant frequency, and the parasitic resistances in the inductor and capacitor are small, a huge
voltage will be generated across the inductor and capacitor. The capacitor voltage will strike the lamp.
Once the lamp is struck, it has a relatively low impedance. The capacitor is in parallel with this low
impedance, and the resonant quality (Q) of the circuit is reduced dramatically. Roughly, you can imagine
that the lamp “shorts out” the capacitor, leaving the circuit as an inductor in series with the lamp. The
inductor serves as the “ballast” impedance that limits the current through the lamp. The square wave
voltage in the circuit above must be large enough to drive the steady state current (between 100 and 500
mA) into the inductor. An AC voltage is used for at least two reasons. First, AC permits “lossless”
reactive elements to serve as ballast impedance. Specifically, an inductor can be used instead of a resistor
to limit the current through the lamp. If we used a DC input voltage, the inductor would have essentially
zero impedance, and would not serve to limit current. Second, the bulb lasts longer if the two filaments
occasionally swap roles as electron emitter and receiver (cathode and anode).
If the resonant circuit has a “high Q” (low-loss) and is driven near its resonant frequency, the capacitor
voltage will be large enough to strike the lamp. This presents an interesting engineering economics/
manufacturing problem when making a lamp ballast. To make sure that the lamp will strike in
a range of customer operating environments, the LC tank should be tightly tuned with low loss. This will
create a strong “peak” in capacitor voltage at the resonant frequency. Unfortunately, this approach also
means that the frequency of the square wave must be tightly controlled to be very near the resonant
frequency. Therefore, either the L and C values must be tightly controlled and unchanging, or the square
wave frequency must adapt to changes or tolerances in the L and C values. Precision power-level
inductors and capacitors are expensive, and incompatible with the production of cheap ballasts, so we will
tune the clock or drive frequency to the resonant point. In a practical ballast, a feedback circuit might be
employed to “find” the resonant point and ensure that the lamp will strike.
A practical schematic implementation of the simple ballast circuit in (a) is shown in Figure (b) below:
(b) Practical Fluorescent Lamp Ballast
As a starting point for designing and building your ballast, assume that you will reuse the totem board that
you built at the solder clinic in Laboratory 2 for the go-cart. This totem board is a little “over built” for
the ballast; for example, the FETS on this board will handle more current that we need for this exercise.
So this totem board should be fine for the ballast. This time, we will run the totem pole with a fixed 50
percent duty cycle. You will design the “new” components, the resonant circuit driven by the totem pole,
in this exercise. This new resonant stage consists of six components: a blocking capacitor Cblock, an
inductance L, a capacitance C, an F4T5 fluorescent lamp, and a two resistor divider for measuring the
lamp voltage. Our goal is to strike and run the fluorescent lamp. Your circuit should meet the following
specifications:
1.) The totem pole driver circuit should run with a 0.5 duty cycle.
2.) The totem circuit should be able to produce any switch frequency between 28 kHz and 40 kHz.
3.) The resonant frequency of the L-C resonant circuit should be between 30 kHz and 40 kHz.
4.) The capacitance Cblock should be a non-polarized capacitor designed to standoff or block the dc
or average component of the totem pole square wave. It should look like a low or negligible
impedance at the resonant frequency.
5.) The totem pole rail voltage should be adjustable between 0 and 40 volts – plan to replace the
connection to the 36-volt battery at the top of the totem pole that we made for the go-cart with a
40-volt adjustable power supply.
6.) Assume that the F4T5 lamp requires 300 volts peak to strike. This number is not firm, it will
change with temperature and bulb age, but it is a reasonable working estimate. When the totem
rail is at its maximum value of 40 volts, the resonant circuit should be designed, when driven at
resonance, to produce an approximately sinusoidal waveform with a peak voltage of 400 volts to
make sure that we can strike the lamp.
7.) In steady state with the lamp “on,” assume that the lamp will exhibit an apparent DC resistance of
approximately 500 Ohms (it could be anywhere between 400 and 600 Ohms), and will require a
peak current of approximately 100milliAmps (i.e., an approximately sinusoidal current with a
peak-to-peak amplitude of 200 milliAmps).
Please prepare the following for your design review and lab report:
•
•
•
•
•
•
•
•
•
Prepare a schematic indicating the modifications that you plan to make to your go-cart totem board and
control circuitry to ensure 50 percent duty cycle. Also indicate how you will make the triangle
wave oscillator (which drives the LM311 comparator) vary between 28kHz and 40kHz. Plan to use
a large, easy to adjust potentiometer so that we can adjust your clock frequency to the resonant
frequency of your resonant circuit. Limit the adjustment range of the pot by adding fixed resistors,
so that the lowest frequency we can make by adjusting the pot is 28 kHz and the highest is 40kHz.
Select values for the impedances Cblock, L, and C to meet the specifications.
Assuming that the lamp is on and exhibiting a DC resistance of 500 Ohms, and ignoring ESR in the
capacitors and inductor, compute the peak voltage across the lamp for your choices of Cblock, L, and
C. What current will flow in the lamp?
Design an inductor with inductance L by selecting an appropriate Micrometals Iron Powder core
from the available class components. Determine the number of turns and wire gauge size you plan to
use. Analyze the performance of your inductor. When will it saturate? What is its parasitic series
resistance? How will this affect the operation of your circuit?
Design another inductor with inductance L by selecting an appropriate Ferrite core from the
available class components. Determine the number of turns and wire gauge size you plan to use.
Analyze the performance of your inductor. When will it saturate? What is its parasitic series
resistance? How will this affect the operation of your circuit?
Compare the two inductor designs using the procedure described on the last page of this lab. Choose
the best inductor for your ballast, and plan to build with this design after your design review.
Select appropriate capacitors for Cblock and C from those available. Estimate the maximum
combined ESR that you will be able to tolerate in these components and still achieve striking voltage
at resonance across the lamp. What voltage rating is required for these capacitors? What type of
capacitors should they be (electrolytic, metal film, etc.)?
On a PC card sheet that shows the “plain” card (print it from the web site), sketch the proposed
physical layout of your resonant circuit. Plan to lash the F4T5 bulb, which is approximately 6 inches
long and three fifths of an inch in diameter, to the card using ty-wraps through the large board holes
in the prototyping area. Indicate where you will place the components that correspond to Cblock, L,
and C. Indicate where you intend to bring the totem mid-point and ground onto the board to drive the
resonant circuit.
In parallel with your fluorescent lamp, plan to add a voltage divider consisting of a 1 Mohm (1e6
Ohms) resistor in series with a 30 Kohm resistor, as shown in Figure (b). We can measure the voltage
across the 30 kOhm resistor to get a “low-voltage copy” of the voltage across the lamp. Plan to bring
•
the connections across the 30 Kohm resistor using a twisted wire pair to a convenient place on your
plain board, for easy access with a scope probe. We’ll want to be able to measure this voltage while
the lamp is operating without having to probe deeply into the high voltage area of the plain board.
Neatly summarize your work for the DR. Also complete Exercise 2 before the DR.
Impact of inductor core loss on your circuits:
The core material that we use to make inductors adds loss in your circuits. Hysteresis and eddy current
losses occur in the core material when an alternating magnetic field is applied. These losses create heat,
and can be modeled as additional resistance at an appropriate place in your circuit. In the fluorescent
lamp ballast, for example, these losses may significantly impact your ability to strike the lamp.
Here is a procedure for qualitatively evaluating the relative performance of an inductor design in the
lamp ballast in comparison to another inductor design. This procedure is expedient in that it does not
require an iterative solution procedure, e.g., on a computer. This procedure does not give quantitatively
accurate estimates of the losses and ballast circuit striking voltage and operating point, however. To
evaluate the approximate effect of losses on your ballast’s ability to strike a bulb, do the following steps
for each candidate inductor design:
•
Calculate approximate operating frequency f
•
Calculate operating current I assuming (hoping) that you will get the striking voltage that you want across the lamp
•
Calculate peak flux density in the inductor, Bpk (for example, see the Micrometals catalog. Read the
catalog carefully to be sure you understand and are using the appropriate units:
•
Examine the core loss density vs. Bpk graph in the material catalog to estimate the core loss density.
•
Multiply by the volume to get total core loss in Watts.
•
Calculate the effective series resistance due to core loss, Rcore (= Power loss over square of RMS current)
•
Use Matlab to calculate the peak of your ballast transfer function for the ballast model shown below. Be
sure that your circuit “resistors” account for all possible loss mechanisms – capacitor ESR, blocking
capacitor ESR, inductor wire ESR, Rcore , the effective core loss resistance, etc.
•
Compare the transfer function peaks for different candidate inductors to evaluate their relative
merits.
•
Be prepared at your design review to identify and explain the various approximations and
assumptions in the method outlined above for comparing inductor performance.
© Massachusetts Institute of Technology, 2010 October
Reproduced with Permission
Courtesy of the Massachusetts Institute of Technology and of Agilent Technologies, Inc.
This experiment has been submitted by the Contributor for posting on Agilent’s Educator’s Corner. Agilent has not
tested it. All who offer or perform this experiment do so solely at their own risk. The Contributor and Agilent are
providing this experiment solely as an informational facility and without review.
NEITHER AGILENT NOR CONTRIBUTOR MAKES ANY WARRANTY OF ANY KIND WITH REGARD TO THIS
EXPERIMENT, AND NEITHER SHALL BE LIABLE FOR ANY DIRECT, INDIRECT, GENERAL, INCIDENTAL,
SPECIAL OR CONSEQUENTIAL DAMAGES IN CONNECTION WITH THE USE OF THIS EXPERIMENT.