13.

8V 20A power supply

Regulated DC power supply, short circuit safe, and with current limiter

This PSU has been especially designed for current-hungry ham radio transceivers. It delivers safely around
20Amps at 13.8V. For lower currents, a separate current limiting output, capable of 15ma up to a total of 20A
has been added. Let us see what we have got here. The power transformer should be capable to deliver at least
25A at 17.5 to 20V. The lower the voltage, the lower power dissipation. The rectified current will be ironed by
the C1, whose capacity should not be less than 40.000uF, (a golden rule of around 2000uF/A), but we
recommend up to 50.000uF. This capacity can be built up by several smaller capacitors in parallel. The base of
this design is a simple 12V regulator (7812). The output voltage can be brought to desired value (here 13.8V)
by two external resistors (R5 and R6) using this formula:
U= 12(1+R5/R6)
The low currents (here 15mA) will keep the 7812 in its regular function. As soon as the current rises over
15ma, the voltage drop on R4 will open the Q3, actually handling the high output current. This is a PNP
transistor (Ic>25) and current amplification factor of at least 20. The one that has been tested and proven here
is the 2N5683. The current limiting resistance RL, for the maximum output of 20 Amps should be 0.03 Ohms,
rated at
least 15W. You can use the resistance wire or switch several resistors in parallel, totaling the resistance/power
values. Values for other currents can be calculated by the rule:
RL=0.7/Imax
The RL and Q2 (3A PNP such as BD330) form a short circuit automatic fuse . As soon as the maximum current
reaches 20Amps, the voltage drop over the resistor RL will open Q2, and thus limit the B-E Current of Q3.
Parallel to Q2 is Q1, which lights the LED 1 whenever the current limiting circuit is active. When the fuse is
active, the Q2 bridges the R3, so the full current would flow through the IC1, and damage it. Therefore the R4
is inserted, as to limit the IC1 current to 15mA. This makes it possible to run the IC1 without any cooling aid.
The LED 2 will light up every time the PSU is switched on.
There is an adjustable current limiter in parallel to the fixed output, thus providing adjustable current source for
smaller currents.

This circuit is very simple too. You will notice that there is no current sensing resistor. But it is really there, in a
form of the Rds-on resistance of the N-channel FET, which actually handles the load cutoff from the source. The
function of the FET is shown in the diagram 2. When the current Id is rising, the tension Uds over the resistance
Rds rises very slowly in the beginning, but very fast after the knick. This means, that before the knick the FET
behaves as a resistor but after it, works as constant current source.
The D2, R3 and B-E connection of the Q4 will sense the Uds voltage of the FET1. When the voltage rises
enough, the Q4 will shortcut the FET1 gate to mass, and cut the current flow through the FET 1 off. However,
to enable the FET1 to open, there is certain gate voltage necessary, which in this case is brought up by the
voltage divider consisting of R8, Z1, P1 and R9. So the maximum Gate voltage will be the one of the Z1, and
the minimal will be around 3V6. The Z1 voltage (Uz1) will thus determine the max current flowing through the
FET 1.
The diagram 2 will show that for 5 Amps the Uz1 should be 5V6, and for 20Amps around 9V6. The Capacitor C4
will determine the velocity or the reaction time of the limiter. 100 uF will make the reaction time to be around
100ms, and 1n will make it 1us.
Within the designed limits, the P1 will limit the current output in the range of 15mA to 20A. You can use both
output simultaneously, but the total output current will be limited by the value of the RL. This PSU can be built
also for higher outputs, as long as the transformer will handle the current requirements, and you provide
sufficient cooling for the Q3.
If somebody will be interested, there is a PCB design ready.
Have fun.
Bob

REV1.
I have received several requests for some modifications, and the one I find useful is the addition of an amp
meter. Therefore the slightly modified diagram is included in this revision. All elements within the dotted border
are now placed on the pcb. There is also elements placement design included. Should one have an 25Amp
instrument on hand, there is nothing easier. Just mount it in line and there it goes. However, a ham would
probably find an instrument somewhere in his junk-box , but the scale would be something completely
different, let say an S or Voltmeter. No problem. We already have a shunt for the amp-meter, and it is there
as the Current limiting resistor RL . As already known from before, there is a voltage drop of 0V7 over the
resistor at current flow of 20A. What we now have to do, is to simply measure the voltage drop over the
resistor, and co-relate it with the current. Let us say that our instrument has an internal resistance of 13R, and
has a full scale reading of 60mV. The voltage drop over the RL is 0V7 for 20Amps. Therefore, we need another
resistance in line with the instrument, that would bring the 0V7 to 60mV, or an voltage drop of 640mV.
The formula is simple:
U1:R1=U2:R2
60:13=640:X
X=13x640/60
X=138.66
Therefore, the resistance that has to be inserted in line is around 140R. I suggest to insert a trimmer (VR1) of
around 200R, to fine trim the reading when calibrating the instrument. Using your favorite drawing software,
design your scale to your likings, (at least 20A for the full scale) and insert it in the instrument you have. Due
to the many requests for the PCB Layout, I have included the design here. The exact dimensions of the pcb are
160x100mm. Please remember, that the pcb has to be printed as a mirror image, to obtain higher quality when
transferring it to the copper side of the board.
Wish you a good time.

Bob

High Current 13.8V Power Supply

Rod Elliott (ESP)

Introduction
As is commonly the case, this supply was born of necessity. There is absolutely nothing
special about the circuit, except that as shown, it is quite capable of up to 20 Amps
intermittently or 10A continuous. Simply use a bigger transformer, bridge rectifier and more
capacitors and output transistors to get more current. The basic circuit should be good for
up to 100A or so, using a 5A TO3 regulator IC, but it can obviously be increased further (if
you really do need a 500A supply!). There is no reason that the supply cannot be made
smaller as well (did I hear someone say "Perish the thought." ?). Using fewer transistors
and a smaller transformer it will work from 1A upwards.
This is not a project intended for beginners or powering opamps (or other similar frivolous
purposes :-) it is intended solely for powering (nominally) 12V car audio accessories.
Regulation is not especially wonderful. It could have been made much better, but at the
risk of instability and increased complexity, particularly as the current capability is
increased. As it happens, the relatively poor regulation is actually a benefit - the supply is
intended for testing car power amplifiers and the like, and even with the heaviest wire,
there will always be some voltage drop, and this is mimicked very well by the supply.
As a result, the tests that are carried out using this supply will be much closer to reality
than if a supply with perfect regulation were used. It can also be used as a battery charger
(with care!), as the no-load voltage is very stable.

This project requires knowledge of mains wiring. If you are unfamiliar with (or
justifiably scared of) the household mains supply - DO NOT ATTEMPT CONSTRUCTION.

WARNING - Never use lead-acid batteries indoors unless extremely good ventilation is
provided. Do not smoke or allow a naked flame within 10 metres of a charging battery, as
highly explosive gases are generated during charging. These batteries contain sulphuric
acid, which is highly corrosive and will cause severe burns. Exercise safe working and
handling practices at all times.
Description
The power supply circuit is shown in Figures 1 and 2. A 7812 positive 3-terminal regulator
is used for the main regulation, and this is followed by as many power emitter followers as

needed for the current you require. The transistors are not critical. I used 2N3771 devices
(50V, 20A, 200W) simply because I had a whole bunch of them in my junk-box. These are
pretty much ideal, but I suggest that you use whatever you can get cheaply. If you use
2N3055s (as indicated in the schematic), expect to use four transistors for the first 10A,
and one transistor for each additional 5A peak (or 4A continuous) output capability to
ensure an adequate safety margin. The voltage rating is unimportant, as the main supply
will only be about 22V with an 18V transformer.

Figure 1 - Basic 10A Power Supply - Power Section

The LEDs are optional, but recommended. 2,2k series resistors (as shown) will give a LED
current of about 10mA, which is pretty much ideal.
The supply is designed to provide very high current, and I used a 300VA toroidal
transformer and two bridge rectifiers, one for each winding. The 40,000uF electrolytic is
one I had to hand, and provides excellent performance. You can get away with quite a bit
less capacitance for the 10A version, but (as always), hum may become a problem if there
is insufficient capacitance. The circuit shown will have a ripple voltage of about 4V at 20A
load, and this is quite acceptable as the regulator IC will remove the vast majority of this
ripple voltage. I was unable to measure any output ripple on my unit at continuous currents
up to 10A, although there must be some - it was just too far below the noise voltage to
measure.
Calculate the capacitance you need from the following formula ...

A full load ripple voltage of up to 5V is acceptable for this application, but feel free to have
less. As ripple voltage is reduced, the dissipation of the output transistors will increase.
This apparently strange behaviour is because the average voltage across the transistors is
greater with lower ripple.
Because the highest rated bridge rectifiers commonly available are 35A, use multiple
transformers (and/or windings) and bridges for more current. This will be a lot cheaper
than trying to get 100A (or more) devices, and overall performance will probably be better
as well. Likewise, use multiple filter capacitors rather than a single large unit - again, these
are cheaper, and will outperform a single very large capacitor. Figure 1 shows the
recommended method of connecting the multiple windings for higher current, which may
be duplicated as many times as needed.

Figure 2 - Basic 10A Power Supply - Regulator Section

As you can see, the regulator is made adjustable over a small range, and will typically give
from 11V to 13.8V at full load. With the no-load voltage set to 13.8V (nominal 12V battery
voltage), the output will fall to 13.5V at about 1.5A, and 12.8V at around 13A. This is fairly
typical of the voltage drops that can be expected in a car installation. Needless to say, if
the supply is designed for more current, then the regulation will remain about the same,
but at the higher design currents.
The output transistors are wired in parallel, with 0.1 ohm 5W wirewound resistors in the
emitter of each. The more transistors you use, the better the regulation and peak current
capability. The resistors used to drive the optional (but highly recommended) ammeter
need only be 1/4W types. These average the individual emitter resistor voltages, and the
result will be much more accurate than driving the meter from only one emitter resistor.
The diode (D1) from output back to input must be a high current type - I suggest a 2A
diode (or two 1A diodes in parallel as I used in my unit). This is used to ensure that the IC
is not damaged if the supply is connected to a battery without mains power. R1 and R2,
the 4.7 ohm 5W resistors feeding the regulator, provide the only electronic protection

available - when the IC current exceeds 1A, the IC input voltage will be reduced and the
output voltage will fall. If you use a high current (TO3 style) regulator, then the value of the
resistors must be reduced, but the diode will need to have a higher rating to compensate
for the increased current back into the main filter cap.

Be warned! There is no diode to protect the unit from reverse polarity if

connected to a battery. A series diode would reduce regulation and be
very expensive, and a parallel diode would short the battery (a typical 12V
car battery can supply several hundred amps with ease!). This is very bad
for the battery, and not too good for the diode, either (it will probably
explode - and yes, I'm serious). An output fuse can be used if desired, but
it will not protect against reverse polarity.
In addition, the supply is perfectly capable of melting flimsy test leads, or
the ground lead on an oscilloscope (for example). Like all high current
power supplies, take great care when building and using this supply, to
avoid the risk of severe burns or damaged equipment.
Protection is with a fuse only, as the supply is sufficiently rugged to withstand almost any
abuse for a short period. The minimal protection provided by R1 and R2 is sufficient to
allow the fuse to blow before any damage is done to the transistors. I briefly considered an
"electronic circuit breaker", but decided against it very quickly since I needed the supply in
a hurry!
The unit I made used a case I had lying around, and although the heatsinking is not
substantial, it is adequate for my needs. Most units will need less heatsink than you might
imagine, since even high power car amps will not draw full power all the time. If you do
decide to make a 100A version (or more), I suggest that you will need quite a large amount
of heatsink - this will not be a real problem (other than financially), since there will be
plenty of room - the power transformer(s) will need to be a minimum of 1,500VA so the
case will have to be quite big. This will leave you with lots of space to play with :-)

Construction
Construction is not critical in the normal sense. The regulator IC must be on a heatsink,
and needs the capacitors (as shown in Figure 1) mounted as close as possible to the IC to
prevent oscillation. No PCB is available for this project, and it is not necessary, since the
wiring all needs to be capable of very high currents that would just melt the tracks off a
circuit board. The small signal section (regulator, transistor and bypass caps, etc.) can be
mounted on a tiny piece of Veroboard or similar.
Use the heaviest wire you can for all main power connections, especially for the output.
Any additional resistance you introduce with your wiring will reduce the regulation. I
suggest that you keep the leads to the 0.1 ohm emitter resistors short, and most of the
power wiring will be pretty much self supporting because of the wire thickness.
Wire the current meter with the return point located as closely to the mid point of the
emitter resistors as possible. The accuracy will never be great, but it will be reduced
further if there is a lot of copper in the circuit, because the temperature coefficient of

resistance for copper is quite high. The 100 ohm output resistor will not introduce an error,
since it is connected to only one of the transistors (directly to the emitter). I calibrated my
meter to 10A full scale, but calibration to 20A is quite OK, to allow for the peak current
capability of the supply.
VR2 (any value from 500 ohms to 2k can be used) is used to calibrate the meter. Use an
ammeter and a suitable load, and adjust the pot to obtain the same reading as the external
meter. Make sure that the external meter is capable of handling the current you intend to
calibrate to. The meter scale can be re-marked as 0-10A or 0-20A, and calibrated
accordingly.
If you do not have access to an ammeter capable of at least 10A, then calibration of the
meter will require a known accurate low value resistance, and an accurate voltmeter. You
can calculate the current by knowing the resistor value and the voltage, and adjust the
trimpot until you get the same reading as you calculate. The meter movement is not critical
either - use any meter of 100uA to 1mA with the circuit as shown. You will need to adjust
the feed resistor values for other movements.
I=V/R

Where I is current, V is measured voltage and R is the test resistor value (in

ohms)
Typically, you will need a resistor of about 1 or 2 ohms to calibrate the unit. Power will be
extremely high - a 1.25 ohm resistor with 12.5V and 10A will dissipate 125W. Eight 10W
10 ohm resistors in a bucket of water will work very well, and will allow you to "soak test"
the unit at full power to make sure that everything manages to stay together.
The voltage control may be calibrated, or just place a marker on the panel for 13.8V. If
desired, a voltmeter can also be included in the circuit - if used, this should be wired to the
output terminals.

Appendix
The author's unit is shown in Figures 3 and 4. It was designed as a 10A supply. As I
mentioned, the case is one I had lying around, and although I can't use mine at its peak of
20A for extended periods (not enough heatsinking), it serves the purpose that I needed it
for, which was to test some car amplifiers I had (also lying around). I have found it to be
extremely satisfactory, and since it can be completed in an afternoon, this makes it a
simple project that should give many years of service.

Figure 3 - Prototype Supply (Front Panel Inside View) - 10A Version

The meter was already in the case I used, the toroidal transformer is clearly visible, as well
as the filter capacitor. The bridge rectifiers are on the vertical aluminium bracket between
transformer and filter cap. The control electronics (regulator, transistor and small caps) are
on the piece of Veroboard just to the right of the meter. The cap on the extreme right is the
output capacitor. The regulator is thermally connected to the front panel to provide
heatsinking (don't forget the insulation washer and bushes!).

Figure 4 - Prototype Supply (Rear Panel Inside View) - 10A Version

In the above view, the power transistor mounting, emitter resistor and mains input can be
seen. The small round thing in the top-centre of the photo is the meter setting trimpot. Note
the shroud over the EIC mains inlet to prevent contact with the 240V, although if you look
carefully, you will see that the fuse has no shroud (naughty, naughty. Verrry naughty!) Please do as I say, not as I do! All mains connections should be protected against contact
(I will get around to it one day - I promise).
From this angle you can see that the filter cap is an old compuer grade unit (salvaged from
my trusty junk box), and you can also see that I only used three power transistors. As I
mentioned before, I used 2N3771 devices, and these are much more powerful than the
2N3055s I suggested, but are probably very hard to get (and almost certainly epensive).
The little heatsinks I used are just visible at the back. The mating surfaces were carefully
filed so they were completely flat, and are thermally bonded to the aluminium backplate
with heatsink compound and lots of pressure from the transistor mounting.

Making More Powerful Units

Since many readers may want higher power than the unit shown, here are some
guidelines for bigger units.
Don't expect to build a 100A version (or more) in one afternoon.
Use one 2N3055 for each 5A of peak output current (4A continuous) - Each
transistor will dissipate about 40W
Assuming a current gain of 20 for the 2N3055s (fairly typical), one 5A TO-3
regulator will drive up to 100A (use 25 transistors)
For more current, use a boost circuit around the regulator IC (up to 500A output,
with 100 transistors!). I shall leave details of the boost circuit to you (it is very
commonly used, and many examples exist on the Net).
Consider using higher power transistors to reduce component count. The cost will
probably be higher though, and heatsink performance will not be as good due to
higher thermal resistance between junction and heatsink.
Use multiple transformers and bridge rectifiers, rather than a single really big one of
each
The transformer(s) need to be rated at 300VA for each 10A continuous. 100A
requires 3kVA
Transformers can be overloaded by up to 200% for short periods (50% of the time
on load, and 50% off). Other overload ratios can be calculated (but excess or
continuous overload is not recommended!)
See the article on Power Supply Design to learn about capacitor ripple current (this
will be extreme!)
See the article on Heatsinks to learn more about the best way to mount the
transistors.
The above is not extensive, but you get the idea. for most applications, the unit shown will
be sufficient. I doubt that too many constructors will want to build 500A supplies, but if you
do have a need for such a monster, then this circuit should do the job quite well. Hmmm ...
500A at 13.8V is 6.9kW - I'm almost tempted to build one for the hell of it (just kidding :-)
Even a 1kA (1,000 amps) unit is not impossible with a few minor modifications (including
the regulator boost circuit), but for anything over the basic 10A unit shown, some extra

heavy duty connectors and fuses will be a good idea. I am doubtful that this will be needed
for most normal applications
For what it's worth, if you do need much beyond the basic 10A supply, use the supply as
shown connected to a car battery. It can safely be left connected permanently if the supply
is set to 13.8V (check the temperature though - lead acid batteries have a temperature
dependent "float charge" voltage). The unit is then a battery charger, but will not introduce
any hum onto the battery output voltage (unlike conventional chargers, which are not
smoothed).

Desktop Power Supply from a PC

Updated May 14, 2006
(See the narrative and disclaimer at the bottom of the page)

A completed 145 watt ATX power supply with switch, binding posts, labels and feet. Notice the zip
ties in the ventilation slots
that hold the load resistor.

This ATX PS board has leads for +5 (RED), -5 (WHITE), +12 (YELLOW), -12 (BLUE) volts,
Ground (BLACK) and switch (GREEN). Dell power supplies manufactured between 1996 and
2000 do not follow the industry standard pinout and color codes. The fan has also been unplugged
for better viewing. Since this PS was converted for use in the logic and robotics labs, the selected
voltages were tapped. Other users may want combinations of +3.3 V (ORANGE), +5 V and/or +12
V if they are converting one of the newer supplies. For R/C applications, the 5 volt output can also
serve as a desktop source to drive receivers and servos. If used as a power source for the micro and
sub-micro servos, you must be careful not to drive the servo to either endpoint to avoid stripping the
smaller gears in these units. Most standard servos have sufficiently robust gear trains and will
simply stall if pushed to the mechanical stops..
Measured voltages on this particular PS (1996 P5-100 MHz Gateway) were about 5.15 and 11.75
volts. The remaining leads have been clipped off at the circuit board.

View of the case top with fan, binding posts and switch. The switch (SPST) and binding posts are
available at Radio Shack or other electronics suppliers.

Power supplies in today's computers are known as SWITCHMODE or Switching Mode power
supplies and require a load to continue to operate after being switched on (the term switching mode
actually applies to the technique of A/C to D/C conversion and not to the power up action). This
load is provided by a 10 watt, 10 ohm wire wound load resistor (sandbar - about $0.80 at Radio
Shack) across the +5 volt supply. Some inexpensive power supplies may fail if forced on without a
load. The sandbar has been zip tied to the case with a small amount of heat sink compound applied.
Without cooling, the resistor will get very hot and may fail prematurely. With this arrangement, the
resistor will remain barely warm to the touch.
Be warned that many of the heat sink greases can be quite toxic and any excess should be cleaned
up and disposed of properly. Also be sure to thoroughly clean your hands and tools after use. While
most heatsink compounds are rated to 160 to 170 C, some may dry out over time and their
effectiveness will diminish -- a periodic check for good contact between the case and resistor is a
recommended practice.

Additional comments
Disclaimer: The information presented should not be considered a "HOWTO"
article, but merely a documentation of my conversion process. Modern PC
Power Supplies can produce high output current levels that may cause internal
overheating in the PS or damage to devices connected to them. Any individual
attempting their own conversion is cautioned to carefully research their PS
specifications and to be mindful of the associated voltages and power. DO NOT
work on your opened power supply with it plugged in!!!!
The PS in the picture is a 145 watt ATX salvaged from a 1996 P5-100 MHz Gateway -- I salvage
all usable parts from the older PC's before dumping them. This one is set up for a logic lab, hence
the +5, -5, +12, -12 volt taps. We also use the +5 to drive servos in the robotics lab. This supply
does not have a 3.3 V source, but the newer supplies do. INTEL has continued to modify the ATX
specifications to include additional power connectors to support the increased power requirements

of the newer motherboards. Before any modification is attempted, you should be sure of the type of
power supply you are working with and the output currents being produced at each voltage level.
Higher wattage supplies can generate fairly hefty levels of current and may overheat or damage
devices attached to them. See the Table of Representative Current Levels for other power supplies.

Wiring coming off an industry standard circuit board will be:

ORANGE
YELLOW
BLUE
RED
WHITE
BLACK
GREEN
GRAY
PURPLE
BROWN

+3.3 V
+12 V
-12 V
+5 V
-5 V (May not be present on recently manufactured
supplies)
GND
POWER-ON (Active high -- must be shorted to ground to
force power up)
POWER-OK What is this??
+5 V STANDBY
+3.3 V REMOTE SENSING Design Guide Update

*** Note that the 1996-2000 Dell's did not completely follow this color coding -- check your
voltage levels with a meter before wiring ***

The yellow, red and black wires will likely be grouped together with a clip. Some of the PS's will
have a detachable plug for the fan and some will have the fan permanently attached to the circuit
board. If the fan is attached, I usually clip the wires then re-solder and cover with heatshrink tubing
-- this gives more working room while modifying the PS and allows me to lube the fan.
If you are going to use only the +12v and +5v, you may clip the other wires at the circuit board
level or leave the unused wires about an inch long, gather common colors together, slip a piece of
heatshrink tubing over the bundle and shrink -- it is an easy way to corral and insulate loose ends.
For the +5 / +12 volt PS, you will need the following combinations:
GREEN / BLACK
RED / BLACK
YELLOW / BLACK
RED / BLACK
ORANGE / BROWN

Power on Switch (Use a SPST switch; a momentary

switch will not work)
Pre-Load Resistor (See text for recommended values and
a possible substitution)
+12 volt source
+5 volt source
See the Design Guide Update

I use a single common post (GND -- black) for all voltage sources. Our loads are light and we don't
require separate grounds for each.
Leave 3 black wires -- switch, load resistor and common (GND) binding post

Leave 2 red wires -- 5 volt binding post and load resistor

Leave 1 yellow wire -- 12 volt binding post
Leave the green wire -- power on switch
If sense wires are present, refer to the Design Guide Update
If you expect to place high current demands on your powersupply, it may be prudent to run two
wires to each binding post -- while it is very unlikely that the 18 AWG wire will overheat, there
have been some instances of melted wires and connectors occurring on high demand motherboards.
As an aside, you can get 7 volts from the +5 V and +12 V outputs -- the +5 V is considered the
negative (GND) and +12 the positive -- some geeks will use this combination to run their fans at a
lower speed to reduce noise.

I've followed all the instructions, but the output voltage on the +12 V side is still low -- what can I
do?? Many of the R/C folks are converting power supplies for the purpose of driving field chargers
and are finding that voltage levels below 12 volts are sometimes insufficient to power their
chargers. Read these TIPS for some options that may help increase this voltage level, provide a
little theory, identify the connector pinouts found on most PC supplies and give a few
troubleshooting hints.
Cut everything else off even with the board. I usually cut the power harnesses so I can keep as much
together as possible. The wires remaining in the power supply should be left long and cut to length
as needed. If you leave them too long, they will get in the way when boxing it up, especially if the
fan is internal rather than external. Be sure that they stay out of the way of the fan blades. Also be
sure to reattach the fan -- some supplies will not function without the fan attached - in any event,
you need the cooling. This PS in the pictures has the fan mounted on rubber shock mounts and is
extremely quiet. I will also disassemble the fan and lube the bearings while I have the PS open.
Since these are salvaged, the fans have been in use for some time and normally the bearings are dry
-- I use a high grade sewing machine oil from SINGER. Any light oil will work, just don't use
WD40 -These power supplies are called SWITCHMODE or SWITCHING MODE power supplies and must
have a load to function -- hence the 10 ohm 10 watt load resistor on the 5 volt line. These resistors
are known as wire wound or sandbar resistors and can be purchased from Radio Shack for about
$0.80 each. This resistor will get hot and should have some sort of heat sink. The technique I use
keeps them amazingly cool and is easy to do -- just pick the flattest side of the resistor, apply some
heatsink compound (see warning above about toxicity) and attach to the case. I will usually hit the
inside of the case with a file to remove any stamping flash on the ventilation slots. The switch
(single pole, single throw) and binding posts can also be found at Radio Shack or other electronics
supply houses.

Resistor Substitute
A viable alternative to using a power resistor is to substitute an 1157 automotive signal lamp in its
place. This is a dual filament lamp and its load, with both filaments powered, is usually sufficient to

maintain Latch_On and to raise the voltage on the 12v rail to an appropriate level for most needs.
Your options are to solder a 5v line (red) to both positive pins on the lamp and ground the base to
DC ground or to pick up a twist-lock socket when you buy the lamp. The advantage of using a
socket is the ease of replacement should the lamp fail. If you don't feel comfortable with your
soldering skills, it is also a little easier to work with the wiring on the socket rather than the pins on
the lamp. Just remember that the socket housing is the ground and the two wires in the base are to
be attached to the 5v rail. More importantly, you must be very careful that neither the bulb base nor
socket housing touch any of the internal components in the power supply. These lamps may be
purchased at any automotive supply store and most Walmarts.
I prefer the use of resistors since the final converted product is wholly self-contained and I have
more control over the applied load, but the use of a lamp does simplify finding and installing
components. It also makes a very obvious Power_On indicator!
I usually deal with on-line suppliers such as Jameco, Digikey, Mouser, etc. because we are buying
in larger quantities and Radio Shack is too expensive for large numbers of items. However, you
should be able to convert your PC supply for $5.00 or $6.00 dollars -- less if you have a junk box of
parts. I suppose you could add an LED indicator with a 220 ohm dropping resistor to the 5v rail to
show the PS is turned on, but the fan is a pretty good hint. We have had supplies running 24/7 for
months without problems -- just electricity consumption.
The PS has some fairly hefty electrolytic capacitors and can still give a bit of a shock immediately
after being unplugged -- let it sit a couple of minutes before poking around inside. Obviously, you
can get whacked if you are inside the case with it still plugged in -- probably won't kill you, but you
WILL turn it loose (never mind how I discovered this bit of information).
If you have any questions, comments or corrections, feel free to mail me.

Updated May 14, 2006

This document was created with Win2PDF available at http://www.daneprairie.com.

The unregistered version of Win2PDF is for evaluation or non-commercial use only.