Fonte.
qxd 11/5/2007 2:13 PM Page 72
Basic Analog
Power Supply
Design
by Gerard Fonte
Part 1
There is the old saying:
“You can give a man a fish
and he will eat for a day or
you can teach a man to fish
and he will eat forever.”
There are many articles that
give the reader a specific
design for building a power
supply. And it must be made
clear that there is nothing
wrong with these cook-bbook
designs. They often have very
good performance. But they
don’t teach the readers how
to design a power supply by
themselves. This two-ppart
article will start from the
beginning and explain every
step necessary to build a
basic analog power supply.
The design will focus on the
ubiquitous “three-tterminal
regulator” and include a
number of enhancements to
the basic design.
72 December 2007
I
t is always important to remember
that the power supply — either for a
particular product or as a general
piece of test equipment — has the
potential to electrocute the user, start a
fire, or destroy the device it is powering.
Obviously, these are not good things. For
that reason, it is critical to approach this
design conservatively. Provide plenty of
margin for components. A well-designed
power supply is one that is never noticed.
Input Power Conversion
Figure 1 shows the fundamental
design for a typical analog power
supply. It consists of three main compo-
nents: Input power conversion and
conditioning, rectification and filtering,
and regulation. The input power conver-
sion is typically a power transformer
and is the only method considered here.
However, there are a couple of points
FIGURE 1. A basic analog power supply
consists of three parts. The first two are
discussed in this article and the last in
the next installment.
that are important to mention.
The first is that 117 VAC (Volts
Alternating Current) is really an RMS (Root
Mean Square) measurement. (Note that I
have seen ordinary household power
specified anywhere from 110 VAC to 125
VAC. I just measured mine and found it
to be precisely 120.0 VAC.) An RMS meas-
urement of a sine wave is much lower than
the actual peak voltage and represents the
equivalent DC (Direct Current) voltage
needed to provide the same power. The
RMS conversion varies according to the
wave shape and for a sine wave the value
is 1.414. This means that the deviation
around zero volts is actually 169.7 volts
(for my 120 VAC power). The power goes
from -169.7 volts to +169.7 volts each
cycle. Therefore, the peak-to-peak voltage
is actually 339.4 volts!
This voltage becomes especially
important when adding bypass capaci-
tors to the main power lines to suppress
noise from entering or leaving the
power supply (a common situation). If
you think the actual voltage is 120 volts,
you may use 150 volt capacitors. As you
Fonte.qxd 11/1/2007 11:35 AM Page 73
FIGURE 2. The input conditioning is pretty basic, but it must be
remembered that the RMS voltage is not the same as the peak
voltage. The peak voltage of 120 VAC RMS is about 170 volts.
can see, this is not correct. The
absolute minimum safe working
voltage for your capacitors is 200 volts
(250 volts is better). Don’t forget that if
you expect to see noise/spikes on the
line, you need to add that noise/spike
voltage to the peak voltage.
The input frequency is universally
60 Hz in the USA. In Europe, 50 Hz is
common. Transformers rated for 60 Hz
will generally perform well on 50 Hz
and vice versa. Additionally, the fre-
quency stability of the power line is usu-
ally excellent and is rarely a considera-
tion. Occasionally, you may find 400 Hz
transformers available. These are typi-
cally military or aeronautical devices
and are generally not suitable for use on
50/60 Hz power (or vice versa).
The output of the transformer is
also specified as an RMS voltage.
Additionally, the voltage specified is
the minimum voltage expected under
full load. Often there is about a 10%
increase in the rated output with
no-load. (My “25.2 volt/two-amp”
transformer measures 28.6 volts with
no load.) This means that the actual
no-load, peak output voltage for my
“25.2 volt” transformer is 40.4 volts!
As you can see, it is always important
to remember that the rated RMS volt-
ages for AC power are substantially
less than the actual peak voltages.
Figure 2 provides a typical input
power conversion and conditioning
design. I prefer to use a double-pole
switch although it is not absolutely
necessary. It protects against
mis-wired electrical outlets (which is
rare today) or mis-wired power leads
in the power supply itself (much more
common). It is vital that when the
power switch is off, the hot lead is
disconnected from the power supply.
The fuse (or circuit breaker) is nec-
essary. Its main purpose is to prevent
fires because, without it, a transformer
or primary circuit short will allow
massive currents to flow causing metal
parts to get red or even white hot. It is
usually a slow-blow type rated at 250
volts. The current rating should be
about double of what the transformer
can expect to draw. For example, the resistor can be increased as required.)
the 25.2 volt two-amp transformer Note that LEDs have really poor
mentioned above will draw about 0.42 reverse breakdown voltages (typically
amps of primary current (25.2 volts/120 10 to 20 volts). For that reason, a
volts x two amps). So, a one amp fuse is second diode is necessary. This must
reasonable. A fuse in the secondary will be able to operate with at least 170
be discussed in the next article. volts of PIV (Peak Inverse Voltage).
The bypass capacitors help to The standard 1N4003 is rated at 200
filter out noise and are optional. Since PIV which doesn’t provide much mar-
the peak voltage is about 170 volts, a gin. The 1N4004 is rated at 400 PIV
250 volt rating is better than a margin- and costs perhaps a penny more. By
al 200 volt rating. You may want to placing it in series with the LED, the
use a “power-entry filter.” There are overall PIV is 400 plus the LED PIV.
many types of these units. Some
contain a standard power connector, Rectification and Filtering
switch, fuse holder, and filter in one
small package. Others may have only Figures 3, 4, and 5 show the most
some of these components. Typically, typical rectification circuits with the
the ones with everything are fairly
expensive, but surplus units can usual- FIGURE 3. The half-wave rectifier circuit
ly be found at very reasonable prices. is simple but it produces a poor output
waveform that is very difficult to filter.
Being able to determine if the Additionally, half of the transformer
primary circuit is powered is important power is wasted. (Note that the filtering
so a pilot light is used. Two typical capacitors are omitted for clarity
because they change the waveform.)
circuits are shown. The neon lamp has
been used for decades. It is simple and
inexpensive. It has the drawbacks that
it is somewhat fragile (being made of
glass), can flicker if the resistor is too
big, and can actually generate some
electrical noise (due to the sudden
ionic breakdown of the neon gas).
The LED circuit also requires a
current limiting resistor. At 10,000 hms,
about 12 mA of current is provided.
Most LEDs are rated for a maximum
current of 20 mA, so 12 mA is reason-
able. (High efficiency LEDs may work
satisfactorily with only 1 or 2 mA, so
FIGURE 4. The full-wave design (top)
produces a nice output. But by
redrawing the circuit (bottom), it can
be seen that it is really just two half-wave
rectifiers connected together. Again,
half the transformer power is wasted.
December 2007 73
Fonte.qxd 11/1/2007 11:39 AM Page 74

Basic Analog Power Supply Design

FIGURE 5. The bridge rectifier bottom). When this is done, it

approach (top) provides full use of becomes clear that the full-wave is real-
the transformer power and with a ly just two half-wave circuits connected
full-wave rectification. Additionally, together. Half of each transformer
by changing the ground reference
(bottom), a dual voltage power power cycle is not used. Thus, the
supply can be obtained. maximum theoretical efficiency is 50%
with real efficiencies around 30%.
very high values of capacitance. The PIV of the circuit is one half of
It is rarely used for an analog the half-wave circuit because the input
power supply. voltage to the diodes is half of the
An interesting and important transformer output. The center tap
thing happens when a filter capac- provides half the voltage to the two
itor is added to a half-wave rectifi- ends of the transformer windings. So,
er circuit. The no-load voltage dif- for the 25.2 volt transformer example,
ferential doubles. This is because the PIV is 35.6 volts plus the no-load
the capacitor stores energy from increase which is about 10% more.
the first half (positive part) of the Figure 5 presents the bridge recti-
output waveform displayed above. cycle. When the second half occurs, the fier circuit which should generally be
(The filter capacitor is not shown capacitor is holding the positive peak the first choice. The output is fully rec-
because by adding it, the waveform voltage and the negative peak voltage is tified so filtering is fairly easy. But most
changes to something like a DC applied to the other terminal causing a importantly, it uses both halves of the
voltage.) It is useful to examine these full peak-to-peak voltage to be seen by power cycle. This is the most efficient
three basic circuits to identify the the capacitor and through that, the design and gets the most out of the
strengths and weaknesses of them. diode. Thus, for a “25.2 volt” transformer expensive transformer. Adding two
Figure 3 shows the basic half-wave above, the actual peak voltage seen by diodes is much less expensive than
rectifier. The only redeeming character- these components can be over 80 volts! doubling the transformer power rating
istic of this is that it is very simple, using Figure 4 (top circuit) is an exam- (measured in “Volt-Amps” or VA).
only a single rectifier. The bad feature is ple of a typical full-wave, center-tap The only drawback to this design
that it uses only half of the power cycle rectifier circuit. This is used today even is that the power must pass through
making the theoretical efficiency of the though, in most cases, it probably two diodes with a resulting voltage
circuit less than 50% just to start. Often, shouldn’t be. It provides a nice output drop of 1.4 volts instead of 0.7 volts
half-wave rectifier power supplies are that is fully rectified. This makes filter- for the other designs. Generally, this is
only 30% efficient. Since transformers ing relatively easy. It uses only two only a concern for low voltage power
are expensive items, this inefficiency is rectifiers, so it’s pretty inexpensive. supplies where the additional 0.7 volts
very costly. Secondly, the wave shape is However, it is no more efficient than represents a substantial fraction of the
very difficult to filter. Half the time there the half-wave circuit presented above. output. (In such instances, a switching
is no power at all coming from the trans- This can be seen by re-drawing the power supply is usually used rather
former. Smoothing the output requires circuit with two transformers (Figure 4 than either of the above circuits.)
Since there are two diodes being
A N ote o n E fficiency used for each half-cycle, only half of
the transformer voltage is seen by
Let me expain the difference between a full-wave rectifier design and a bridge, each. This makes the PIV equal to the
especially from the point of view of efficiency.
Remember that I said half the power is wasted with a full-wave rectifier design.
peak input voltage or 1.414 times the
It is self-evident if you consider that a full-wave design is really made up of two transformer voltage, which is the same
half-wave designs. Combining two circuits that are each about 50% efficient does not as the full-wave circuit above.
in any way increase the total efficiency of the circuit. It’s still about 50% overall. A very nice feature of the bridge
The loss of efficiency comes from the amplitude of the resultant rectification. A
half-wave circuit clearly provides half the amplitude that is available from the trans- rectifier is that the ground reference
former. So does the full-wave version, but it’s not as obvious. It only uses half of the can be changed to create a positive
total available voltage of the transformer. and negative output voltage. This is
The output of the full-wave rectifier is 1.414 FIGURE A
shown in the bottom of Figure 5.
times the voltage between the center tap
and the end. The output of a bridge is the
full end-to-end voltage of the transformer Filtering
times 1.414. This is shown in the attached
figure. For the 25.2 volt transformer, the
full-wave rectifier develops about 19 volts
Nowadays, nearly all filtering for an
(unloaded). The bridge rectifier develops analog power supply comes from a filter
about 38 volts (unloaded). So, the full-wave capacitor. It is possible to use an induc-
rectifier does indeed waste about half the tor in series with the output, but at 60
available power from the transformer.
Hz, these inductors must be quite large
74 December 2007
Fonte.qxd 11/1/2007 11:37 AM Page 75
and are expensive. Occasionally, they
are used for high-voltage power supplies
where suitable capacitors are expensive.
The formula for calculating the filter
capacitor (C) is quite simple, but you
need to know the acceptable peak-to-
peak ripple voltage (V), half-cycle time
(T), and current drawn (I). The formula is
C=I*T/V, where C is in microfarads, I is
in milliamps, T is in milliseconds, and V
is in volts. The half-cycle time for 60 Hz
is 8.3 milliseconds (reference: 1997
Radio Amateur’s Handbook).
It is clear from the formula that the
filtering requirements are increased for
high current and/or low ripple power
supplies. But this is just common sense.
An easy-to-remember example is 3,000
microfarads per ampere of current will
provide about three volts of ripple. You
can work various ratios from this exam-
ple to provide reasonable estimates of
what you need fairly quickly.
One important consideration is the
surge of current at turn-on. The filter
capacitors act as dead shorts until they
get charged up. The larger the capacitors,
the greater this surge will be. The bigger
the transformer, the greater the surge will
be. For most low voltage analog power
supplies (<50 volts), the transformer
winding resistance helps somewhat. The
25.2 volt, two amp transformer has a
measured secondary resistance of 0.6
ohms. This limits the maximum inrush to
42 amps. Additionally, the inductance of
the transformer reduces this somewhat.
However, there is still a large potential
current surge at turn-on.
The good news is that modern sili-
con rectifiers often have huge surge cur-
rent capabilities. The standard 1N400x
family of diodes is usually specified with
30 amps of surge current. With a bridge Filter PIV
Circuit
circuit, there are two diodes carrying Needs Factor Transformer Use
this so worst-case is 21 amps each Half-Wave Large 2.82 50% (theoretical)
which is below the 30 amp specifica- Full-Wave Small 1.414 50% (theoretical)
tion (assuming equal current sharing, Bridge Small 1.414 100% (theoretical)
which is not always the case). But, this TABLE 1. A summary of the characteristics
is an extreme example. Generally, a fac- of the various rectifier circuits.
tor of about 10 is used, instead of 21.
Nevertheless, this current surge is rectifiers and capacitors. (The ripple
not something to be ignored. should be about two volts, as shown.)
Spending a few cents more to use a The bridge rectifier must be able
three-amp bridge instead of a one- to handle the high initial current surge.
amp bridge may be money well spent. A two-amp unit costs about $0.57,
a four-amp bridge is $0.77, and a six-
Practical Design amp device is $0.87. It seems obvious
that spending an additional dime for
We can now put these rules and improved reliability is worthwhile.
principles to use and start to design a Note that the bridge is specified by
basic power supply. We will use the 25.2 what the transformer can supply rather
volt transformer as the core of the design. than what the power supply is eventu-
Figure 6 can be seen as a composite of ally specified for. This is done in case
the previous figures but with practical there is an output short. In such a
part values added. A second pilot light in case, the full current of the transformer
the secondary indicates its status. It also will be passed through the diodes.
shows if there is a charge on the capaci- Remember, a power supply failure is a
tor. With such a large value, this is an bad thing. So design it to be robust.
important safety consideration. (Note
that since this is a DC signal, the 1N4004 Conclusion
reverse voltage diode is not needed.)
It may be cheaper to use two small- Details are an important considera-
er capacitors in parallel than one large tion in designing a power supply. Noting
one. The working voltage for the capac- the difference between RMS voltage and
itor must be at least 63 volts; 50 volts is peak voltage is critical in determining the
not enough margin for the 40 volt proper working voltages for the supply.
peak. A 50 volt unit provides only 25% Additionally, the initial surge current is
margin. This may be fine for a non- something that cannot be ignored.
critical application, but if the capacitor Next month, we will complete this
fails here, the results can be catastroph- project by adding a three-terminal regula-
ic. A 63 volt capacitor provides about tor. We will design a general-purpose,
60% margin while a 100 volt device current-limited, adjustable voltage power
gives 150% margin. For power supply with remote shut-off. Additionally,
supplies, a general rule of thumb is the principles used for this design can be
between 50% and 100% margin for the applied to any power supply design. NV
FIGURE 6. Final design of the power supply
with practical parts specifications. Regulating
the power is discussed in the next article.
December 2007 75
Fonte2.qxd 11/29/2007 10:54 AM Page 60

Basic Analog
Power Supply
Design
by Gerard Fonte

H
owever, it should be remembered non-inverting input of an op-amp. The

Part 2 that many of the basic design

aspects of three-terminal regula-
inverting input of the op-amp measures
the output. If the output is higher than
In the first part, tors apply to any analog regulator
design. In this article, a zero to 30
the reference, the op-amp slews
towards a lower voltage.
we saw that there VDC, zero to one amp, current limited Conversely, if the output is lower
are three main power supply will be developed. The than the reference, the inversion of the
basic cost for parts is under $25. op-amp’s input drives the base of the
components to an transistor harder to allow a higher
analog power supply: Basic Three-T
Terminal voltage to pass. This design concept
input power Regulator is pretty straightforward (although
implementing it can be fairly complex).
conversion and Figure 1 shows a conceptual There are a number of points
design of a three-terminal regulator. about this basic design that need
conditioning, (We will limit our discussion to positive discussion. The first is that the pass
rectification and regulators because the negative regula- transistor is acting as a variable resistor.
filtering, and tors are virtually the same, other than
polarity.) It is important to appreciate
It reduces the input voltage by limiting
the current. As you can imagine, a
regulation. We how this works in order to properly fixed resistor in this situation must be
examined the first understand and apply analog regulator able to dissipate plenty of heat, so it’s
concepts. Fundamentally, the regulator the same for the transistor. We will
two components and is a negative-feedback servo system. examine this in more detail later.
in this article we The resistor and zener diode provide The second point is that there is a
a stable voltage reference to the voltage drop across the pass transistor.
will examine the All bipolar semiconductor
regulation aspect. junctions display this
effect. This means that
We will concentrate the input voltage must
on the basic three-
terminal regulators FIGURE 1. The basic
approach to the three-
that are cheap and terminal regulator is a
simple negative feedback
easy to obtain. op-amp circuit. Although
in practice, it’s a bit more
complicated than this.
60 January 2008
Fonte2.qxd 11/29/2007 10:55 AM Page 61
FIGURE 2.
The practical
implementation
of the LM317
adjustable regulator
requires a few
additional
components
for optimal
performance and
reliability. The
text describes the
additional parts
and their function.
be higher than the
output voltage for proper operation.
For the standard three-terminal volt-
age regulators, this voltage is about
1-2.5 V, depending on temperature
and current drawn. (However, for
proper design purposes the 2.5V
“drop-out” voltage should always be
employed.)
If the desired input-output voltage
difference is less than this 2.5 V,
the circuit will no longer have the
headroom to regulate properly
and drop out of regulation. (There
are other “low-dropout” or LDO
regulators available that can operate
with much less input-output differen-
tial — some as low as 0.1 V. However,
care must be taken to provide proper
capacitive loading or else they may
not function properly.)
The third point is that it takes
some small amount of time for an
output change to propagate through
the op-amp and transistor (loop
delay). During this time, the output
is not being properly regulated.
Obviously, the faster the op-amp and
transistor, the shorter this transient
will be. However, manufacturing
fast, high-power transistors is not easy
or cheap.
Additionally, the faster the transis-
tor and op-amp circuit is, the more
prone to oscillation it is. Having your
power supply oscillate is not a good
thing. Most three-terminal regulators
have a transient response (closed-loop
delay time) of about 30 μs.
The last point concerns adjustable
voltage regulators (Figure 1 shows a
fixed-voltage regulator). Without
getting into a long technical discus-
sion, it is difficult to provide a stable
and cheap reference voltage below
1.2 V. This limits the minimum output
FIGURE 3. By
substituting the load
for the adjustment
resistor, the LM317
becomes a current
regulator (the voltage
changes according
to the load). This is a
very useful circuit
that is not often used.
voltage of adjustable regulators. switching power supply which is
However, we’ll see a cheap and easy typically quite noisy.
solution to this problem, later. The diodes are used to protect
the regulator against unexpected
Additional Components voltage reversals. If the input should
Needed be shorted to ground, capacitors on
the output and/or the adjustment pin
For the fixed voltage regulator, will still maintain their voltage. They
only an input and output bypass will then discharge backwards through
capacitor (connected to ground) are the respective pins towards the input.
required (typically 0.1 to 1.0 μF). This can destroy the device.
These are to stabilize the op-amp If there is no adjustment bypass
in the regulator. If the large filter capacitor used (Cbyp), then the Dadj
capacitor is in close physical proximity can be omitted. If you are sure that
(less than 6”) to the regulator input, the device being powered will never
then the input bypass capacitor can have 10 μF or more load capacitance,
be omitted. then Do can be omitted, as well.
In theory, the output capacitor
can also be omitted. But if your circuit Adjustment Calculations
to be powered just happens to have
a load capacitance of 500 pF to It is important to remember that
5,000 pF, then the regulator might (for adjustable regulators) the
oscillate. Using a 1.0 μF output bypass adjustment pin acts as a non-inverting
capacitor forces the op-amp into a feedback point to the internal op-amp.
stable operating mode. It is always a As such, the op-amp will do whatever
good idea to spend a few pennies on it can to maintain the voltage on this
this to be sure you have a reliable pin (which is specified to be 1.25 volts
power supply. below the output). So if we apply a
Figure 2 shows the proper design voltage (through a resistor) to this pin,
for an adjustable regulator. The input the output will go up. If we short this
and output bypass capacitors (Ci pin to ground, then the output will fall
and Co) are the same as in the fixed to about 1.25 volts. If a negative 1.25
regulator implementation. volts is applied, then the output will
The two resistors (Rb and Radj) go all the way to zero volts.
are required to choose the output Generally, a voltage divider is
voltage. Typically, Rp is set at 240 used to pass some of the output
ohms to provide a proper current voltage back into the regulator to set
(typically about 50 μA) into the the output voltage (as shown in Figure
feedback of the op-amp. Radj is 2). The equation is Vout = Vref (1 +
used to vary the output to the Rp/Radj) + (Iadj x Radj). Vref is 1.25 V
desired value. The capacitor connect- as noted above. The second term (Iadj
ed to the adjustment pin (Cbyp) x Radj) is the error correction for the
is used to improve ripple/noise 1.25 V reference. Generally this is
rejection. This is especially useful if quite small and is often ignored. If this
the raw DC input is coming from a is ignored and the typical value of 240
January 2008 61
Fonte2.qxd 11/29/2007 10:55 AM Page 62
Basic Analog Power Supply Design
FIGURE 4. This is a nice bench power
supply that provides 0-30 volts at
one amp. It’s also current limited to
help eliminate unnecessary smoke
generation in your circuits. Meters to
monitor the voltage and current can
be added, if desired.
Ω is used for Rp, then the equation
becomes Vout = 1.25 (1 + 240/
Radj). Usually, this is accurate to
within a few percent.
A curious thing happens
when we replace Radj with a
power load instead of a resistor
(see Figure 3). The output voltage
varies with the load. We see that
this makes sense because the
power load variation is acting as a
variable resistor. The result is that
the current to the load remains
fixed but the voltage varies. In
other words, the voltage regulator
has changed into a current regula-
tor. Instead of a constant voltage,
we have a constant current.
It is easily seen that the maxi-
mum current allowed depends
upon Rp. If Rp is made 1.25 Ω, then
up to a full amp of current can be
supplied. Note that this current is
limited by the 1.5 A maximum
output of the regulator and the
maximum output voltage obtainable
(which depends upon the input
voltage). This current regulation is
a very useful feature (of these
regulators) that is not often used.
Power Dissipation
Basically, the regulator acts like
a variable resistor in series with the
power supply. The amount of power
dissipated is simply the voltage
difference between the input and
output pins times the current
drawn. This creates some interesting
situations.
For example, suppose you
provide 35 V to the input of an
adjustable regulator and expect to
draw one amp at five volts and
also at 32 V. At five volts, there
will be a 30 V drop across the
regulator. With an amp of current,
the regulator will have to dissipate
30 W while providing five watts to
the load. It’s only 14% efficient. At
62 January 2008
32 volts, there will be a three volt
drop which means that the regulator
must dissipate just three watts while
supplying 32 W to the load. This is
91% efficient.
It is clear that keeping the input
voltage as close to the output as
possible results in better efficiency
and less heat. (Note that with an input
voltage of 35 V and a current require-
ment of one amp the total power used
will always be 35 W. The more power
the load uses, the less the regulator
must dissipate.)
Often times, multiple voltages are
needed for a circuit. There is nothing
wrong with placing a five volt
regulator after a 10 V one. In this way,
the voltage drop for the five volt
regulator is shared between two
devices. However, if the two voltages
have significantly different current
requirements, this approach may not
be optimal. (You can use Ohm’s Law
to calculate the effective resistance
of the regulator given the desired
voltage and current. The power is then
calculated to be the resistance times
the current squared.)
Applying the Theory
We can now combine the infor-
mation from the two articles to build a
practical power supply. We’ll start
with a basic supply and then discuss
how to change it to add features that
may be useful. Figure 4 provides a
practical circuit for an adjustable,
current limited and voltage regulated,
one amp power supply that goes from
zero volts to about 30 volts. (This
particular circuit has proven to be very
useful and adaptable in a number of
high-reliability designs for my clients,
usually with fixed current limit and
fixed output voltage.)
The raw DC circuit was described
in detail last month and will not be
discussed further except to note a
few minor changes. The power trans-
former is now specified as a 24 volt,
center-tapped, 1.0 to 1.5 amp device
instead of 25.2 volts at two amps. The
reason for this change is three-fold.
The first is that some regulators
are limited to 35 V (most others
are rated to 40 V and there is a high
Fonte2.qxd 11/29/2007 10:55 AM Page 63
voltage version — LM-317H — that
can go to 60 V). Note that this is
technically the voltage differential
between input and output, not
from input to ground. If the output is
maintained above ground, then the
input can be higher.
However, for a general-purpose
power supply you must assume the
output will be shorted to ground from
time to time. This means that the input
voltage must be limited to about 37 V
because there will be about 1.4 V
dropped by the bridge diode and
an additional 2.5 V dropped by the
current regulator.
Also, there is no reason to provide
two amps of primary DC current if the
maximum output is rated at one amp.
(Note that the rating is less than the
absolute maximum the circuit can pro-
vide.) The third reason is for regulator
heat control. This is also the reason
for the added switch which will be
discussed in more detail below.
The main filter capacitor has been
reduced from 10,000 μF to 4,700 μF
because the power supply current
specification has been reduced from
two amps to one amp.
The adjustable current regula-
tor is just three parts: an LM-317, a
fixed resistor, and a variable resistor.
The fixed resistor is necessary to limit
the current through the resistor string.
Obviously, as the resistance gets
lower, more current flows. With a 1.25
V differential between the output and
adjustment pin, the one ohm fixed
resistor limits the current to 1.25 A.
If this resistor was eliminated and
the adjustable resistor was set to 0.1 Ω
for example, the current would be
(theoretically) 12.5 A (using Ohm’s
Law). This would either destroy the IC
or cause it to shut down from thermal
overload. Neither is good.
Also, the power through that
small resistance would be 15.6 W
(power is equal to current squared
times resistance). This would burn
the variable resistor. By forcing a
minimum of 1.0 Ω in the circuit,
the current is limited to a maximum
1.25 A and the variable resistor is
protected.
The worst case power dissipated
by the fixed resistor is 1.25 W. A
three watt resistor (rather than a
wo watt device) is chosen to be
conservative. The worst case power
requirement of the variable resistor
occurs when it is equal to the fixed
resistor or 1.0 Ω.
In this instance, there will be
about 0.4 W of power through it. This
means that a one watt potentiometer
is adequate, but a two watt unit will
be used to be very conservative.
Remember, it’s better to spend a little
bit extra on the power supply than to
spend a lot on ruined circuits.
The voltage regulator section is
simply the design from the beginning
of this article. The only change is to
use a variable resistor (R6) instead of a
fixed resistor (Radj).
Heatsinking is an important issue.
The voltage regulator must be able to
dissipate up to 32 W of power, worst
case. This creates a significant
problem. The typical TO-220 case is
limited to 15 W as per the spec sheet.
One poor solution is to reduce the
maximum power rating to about 0.5 A
for lower voltages. This is changing the
specs to match the design instead of
designing to match the specs. There is
another approach that requires a bit
of creative engineering that will be
detailed below.
Curiously, the current regulator is
very efficient. The voltage drop is
typically 2.5 V (worst case) so the
maximum power dissipated at one
amp is 2.5 W. A simple and inexpen-
sive clip-on heatsink is sufficient here.
Fixing the Heat Problem
As noted, there is too much heat
to be dissipated at low voltages. The
problem comes from the large voltage
drop from the input to the output of
the regulator. The solution is to use
a center-tapped transformer and
switch between taps. Center-tapped
transformers are common and
generally cost no more than non
center-tapped transformers.
When the switch is connected to
the top end of the transformer, the full
voltage is applied to the rest of the
circuit and the full range of voltages is
possible. As long as the current draw
is kept below about 0.5 A, there will
be no heat problem. However, if the
power supply needs to provide a high
current at a low voltage, the switch is
used to connect to the center-tap and
reduces the input voltage by 50%.
Thus, only about 18 volts is applied to
the rest of the power supply.
So, if five volts at one amp is
needed, then only 13 W of power
must be dissipated by the regulator as
heat (instead of about 30 W). This
allows the standard TO-220 style reg-
ulator to be used within the specified
limits. (If you want to get fancy,
you could monitor the adjustment
voltage and use a separate circuit to
automatically switch between the
transformer windings.)
Fixing the Minimum
1.2 Volt Problem
Most people want a power supply
to go all the way to ground. It turns
out that there is an easy way to
accomplish this: just add two diodes
in series with the output. Each diode
drops the voltage by about 0.7 V
(depending on load). So, two diodes
will reduce the output by 1.4 V. In
actual operation, only about 1.0 V is
dropped at no load. This makes the
minimum output no load voltage 0.2
V. This is virtually zero.
If you really want to, you can
add another diode in series to drop
the output further, but it seems
unnecessary. The resistor to ground is
included to drain off any leakage
through the diodes. Without the
resistor, the output can float up to the
voltage applied to the diodes.
It should be noted that you can
pull the output of the regulator to
ground if you apply a negative voltage
to the adjustment pin. This is some-
what complicated if a negative voltage
is not easily available (as in this case).
The addition of two diodes is a quick
and easy fix.
Minimum Current Discussion
As noted on the schematic
shown in Figure 4, the minimum
current limit is about 5 mA. For
virtually all applications, this is not a
problem. Very few components will
January 2008 63
Fonte2.qxd 11/29/2007 10:59 AM Page 64
Basic Analog Power Supply Design
be damaged with 5 mA applied to
them. However, if the voltage is set to
30 V, then 150 mW of power is
present at the output. This could be
too much for some components. It
must be remembered that the voltage
regulator uses some current for its
operation (in addition to the current
dissipated during the process of regu-
lation). As it happens, the regulator
draws about a constant 5 mA with a
30 V input, regardless of the voltage
output setting. Therefore, while the
current regulator provides a minimum
of 5 mA, this current is used by the
following voltage regulator and it is
not seen at the output to a great
degree (although in certain special
cases it might be). This is not a perfect
solution, but it is reasonable.
If desired, a larger current
adjustment resistor (R5) can be used.
A 1K potentiometer will reduce the
minimum current out of the current
regulator to about 1 mA.
Adding Meters
It’s always nice to be able to
measure the voltage output and
current used. Adding a voltmeter is
simple. Just place it across the output
to measure the voltage.
The current meter is a bit more
complicated. You can place an
ammeter in series with the output to
measure the current used. This will
work most of the time. The problem is
with the series resistance added by
the meter. Depending on the meter,
this might affect the circuit you are
powering. Basically, it increases the
power supply impedance. This is
especially true for low current ranges
where the series resistance may be a
hundred ohms or more (my cheap
Power S upply P arts L ist
ITEM DESCRIPTION
R1 30K 1W
R2, R3 10K
R4 1.0 Ω 3W
R5 250 Ω 2EW adjustable
R6 5K
Resistors 1/4 watt 5% unless
otherwise specified
64 January 2008
meter has 800 Ω of resistance on the
500 μA scale).
If you don’t mind an error of a few
mA, there is a very convenient place
to measure the current. All you have
to do is measure the voltage across
the leads of R4 in the current limiting
circuit. This is a one ohm resistor
and will provide one volt per amp of
output current.
The basic accuracy depends upon
the precision of the resistor. A 1%
resistor will be accurate to 1% except
for the few mA used by the voltage
regulator circuit that follows. (Note
that you can measure small load
currents by noting the current before
connecting the load and after it
has been connected. The difference is
due to the load. Your basic limitation
here is the ability to measure small
voltages. Five mA of current will pro-
vide only 5 mV across the resistor.)
The proper method of measuring
the output current is to add a
small series resistor to the output and
measure the voltage drop across it
(just like we did with R4). Typically,
this resistor is 0.1 Ω or less and
requires a bit of careful design. There
are current monitor ICs available for a
couple of dollars that greatly simplify
the problem of converting a current
into an easy to measure voltage.
Conclusion
Understanding basic analog
power supply concepts allows the
design of a robust, general-purpose
power supply for under $25.
Adjustable three-terminal regulators
are inexpensive and their shortcom-
ings can be compensated for without
too much trouble. Higher current and
higher voltage regulators are available
so that larger power supplies can
be built.
Instead of providing only a circuit
diagram and instructions, these two
articles have attempted to provide a
foundation of information needed to
help anyone design a power supply
for their own needs. Power supply
design is a basic skill every engineer
and hobbyist should have. After all,
every project needs to be powered in
some way. NV