Many thanks to Peter VK2BFQ for the excellent revision of the following text

13.8 V / 15 A from a PC Power Supply

Safety Instructions

Caution mortal danger: The following circuit operates at a mains voltage of 230 Vac.
Because of rectification some of the components conduct dc voltage, of more than 322 V.
Work has to be carried out only when the circuit is disconnected from the mains, and de-
energized. Note that capacitors located on the primary side, can be charged with high
voltage for several seconds.Even after switching off the mains voltage.

The major disadvantages of usual linear power supplies are high power dissipation, the
size, and the appropriated weight. When looking for an alternative solution, I decided to
use a switch mode power supply (SMPS). The efficiency of such power supplies is around
70 % to 90 % at a power density of 0.2 W / cm³. Because homebrewing was out of the
question due to lack of time, I tried the modification of a PC switch mode power supply.
The latter are mass produced goods, and available for less than 50 DM.

Fig.1: Block diagram of a primary switching power supply

Brief description of PC SMPS Features

Depending on the PC model, these are rated anywhere between 150 and 300 W. For
supplying socket 7 main boards they have four different output voltages of +5 V, +12 V, -
12 V and -5 V. They are mainly primary switching power supplies with power switches
arranged in a half-bridge configuration. The outputs can drive the usual 20 A (+5 V), 8 A
(+12 V) and 0.5 A (-12 V, -5 V). At approx. 205 W output power and a typical efficiency of
75 % this means a dissipation of only 68 W. I had acquired an unbranded PC power
supply, measuring 140W x 100D x 50H mm, and weighing 350 g. Most power supply units
are designed according to the same principle (half-bridge configuration), and hence the
following described modification should be applicable also, to power supplies from other
manufacturers.

Fig.2: Half-bridge configuration of power switches

Strona 1 z 7

2012-01-21

Regulation

After switching on the mains voltage the circuit operates for a short duration as a free-
running oscillator. This behavior is caused by a feedback winding at the output
transformer T2. As soon as the auxiliary voltage Uaux is present, the pulse width
modulator IC, TL494CN from Texas-Instruments takes over the control function, and
synchronizes the "oscillator".
The error amplifier in the TL494 compares the voltage at the +5 V output (actual value),
with a reference voltage (set value).It calculates the analogue control variable according
to the PI algorithm, and adjusts the pulse width modulator (see Fig. 6). The modulator
sends alternate pulses to the driver transistors Q5 and Q6. The pulse duration is inversely
proportional to the variable control setting. Increased loading on the +5 V output, makes
for wider pulses. Lighter loading causes narrower pulses. As there is a finite minimum
pulse width, a minimum load of 0.1 A is required. Without this load the power supply may
be destroyed. The switching frequency is approximately 33 kHz. fairly normal for PC
power supplies. It is defined by a resistor, and a capacitor located at pin 5 and 6 of IC1.

Monitoring Circuit

Strona 2 z 7

2012-01-21

Several protection circuits are included in the original power supply. Excessive primary
current due to a very high secondary current, leads to a high alternating voltage at the T3
output. If this voltage is above a fixed threshold, the TL494 stops cyclically generating
pulses, and changes to the intermittent mode (on / off). The circuit and the load are
protected likewise against over-voltage at the +5 V output, or short-circuit at the -12 V and
-5 V outputs. Switching off is executed via H-signal to the IC1 protection input (pin 4) as
well.
If you see a KA7500 or IR3MO2 PWM regulator IC on the board, each one is a pin
compatible second source to the TL494CN. IC3 is a LM339 dual comparator type. Some
power supplies are not equipped with this IC, but instead, with a discreet two transistor
monitoring circuit, offering the same functionality.

Mods to the Secondary Rectification

The intent is for all of the available power at the 12 V secondary of T1 to be rectified,
regulated, protected, and filtered to provide a single output of 13.8 V DC at 205 W, or
more, if possible. A first check indicated that the +12 V wire was of the same diameter as
the +5 V wire.
First unsolder, and remove all components on the secondary side of T1, which are
provided for rectification, filtering, and regulation of the four output voltages. On that part
of the board, there are only three remaining components. RC1 to RC3, and the
components for providing the auxiliary power supply Uaux.

Reconstruction of the secondary side.

Break the PCB tracks between the RC members RC1 / RC2 and
both 5 V taps of the T1 secondary winding.

Modify L4 for 12 V at 20 A. Remove windings L4a, L4b and L4c
from the toroid (counting turns of L4c). Rewind the toroid L4* with
a single winding, turn count as old L4c but with 2.5 times the
thickness. Take two wires with 1 mm diameter each, bifilar
wounded.

Install two low ESR electrolytic capacitors of 2200 uF each and
the 100 Ohm bleeder resistor as permanent load.

Use the old PCB tracks from the +5 V section and GND tracks as

Strona 3 z 7

2012-01-21

A simple and clear structure of the secondary rectification was achieved after "stripping"
and "reconstruction".

Mods to the Regulation and Protection Circuit

The part of the circuit responsible for regulation and monitoring has to be modified at
three places. Arrange additional components free standing on the component side of the
PCB.

terminals for L4*. The 100 Ohm resistor and the two 2200 uF
capacitors. Insert L4* at the same place, onto the PCB
component side where the L4b winding was connected before.

The original cooling of the rectifier diode D5 is insufficient.
Adequate cooling is achieved by a finned heat sink measuring 70
x 50 x 30 mm (W, D, H) instead of the old aluminium sheet metal.

Fasten D5 to the heat sink and extend the three leads by 40 mm
long wires. Use isolation material and thermal compound. D5
carries on some boards the abbreviation SKD.

Place the finned heat sink approx. 40 mm above the "stripped"
secondary (see photo) with plastic spacers and long M3 screws
(avoid short-circuit to common).

Connect the anode leads of D5a and D5b with one RC member
RC1 / RC2 each. The cathodes have to be connected to the
nodal point of RC1, RC2 and L4.

Establish two links between the 12 V terminals of T1 and the RC
members by two thick wires. D5 will be fed from the 12 V
winding.

R24* is calculated for 13.8 V output voltage. The voltage at the
(+) input of the error amplifier must be equal to 2.5 V after control
loop stabilization, i.e. half the 5 V reference voltage when the
output is at 13.8 V.

R24* = 20 kOhm = 2 x 10 kOhm in series

Arrange a second universal diode 1N4148 and a 8,2 V Zener
diode in series to D16.

Strona 4 z 7

2012-01-21

The areas marked with dotted frames, show the modified or additional components that
are necessary for 13.8 V output.

Fig. 6: Regulation and protection circuits incl. all modifications

Further Modifications

After commissioning the modified board, the situation in regards to interference looks very
bad. The whole reception range from 3.5 MHz to 30 MHz was disturbed by harmonics of
the 33 kHz switching frequency. S-meter readings showed S5 on 80 m down to S2 on 10
m. As I was testing the board in a metal box, the HF radiation could only get out on the
mains cable and/or DC output leads. The insertion of an additional standard 230 VAC
mains filter, and a home-brewed pi-filter in the output rendered the interference inaudible.

U

sum

= 8,2 V + 2 x 0,7 V = 9,6 V

Simplify the voltage divider (R36, R42, R45 and D14) in the
short-circuit protection circuit. For this remove R36 and D14.
Connect the free end of R42 to common (GND) and replace R45
with one of higher value to ensure no shut-down at normal
operation. The voltage across R42 must be less than 1,7 V (I
chose 1,2 V).

R45* = 15 kOhm

Insert an additional 230V / 2A mains filter to the primary side,
close to the place where the mains cable enters the enclosure

Strona 5 z 7

2012-01-21

Testing the Power Supply

Phase 1: These tests have to be carried out at a low DC supply voltage in order to avoid
component destruction in case of possible errors. The 13.8 V output is loaded with a 12
V / 50 W car headlight globe, and a 15 V / 1 A lab power supply is connected to GND, and
Uaux. The TL494 IC generates control pulses, with a maximum pulse duration. Check the
signals at Q5 and Q6.

Phase 2: During the second test phase, the galvanic isolated primary side of the circuit is
supplied by the lab supply also. For this purpose make a short cable link between Uaux
and U+ as well as between GND and U-. The PWM controller tries to offer 13.8 V at the
output at maximum pulse duration. The latter cannot be successful, due to the low 15 Vdc
input voltage, and the present transformer ratio. With an oscilloscope, measured signals
at the measuring points TP1 (emitter Q1 against emitter Q2) and TP2 (cathode D5
against GND) must look like that, shown in figure 7.

Phase 3: Nor disconnect the lab supply from the primary side only. Instead connect a 48
V / 1 A mains transformer to the L1 and N terminal in order to feed the board with a
galvanic isolated Ac voltage. 60 Vdc at C1 and C2 is in Europe defined as a non-
dangerous voltage rate. 48 VAC at the input causes a rise of the output voltage up to +6
V.

rear wall.

Insert a 20 A pi-filter to the DC output , behind the +/- DC
terminals at the rear wall.

The power supply enclosure must absolutely consist of iron sheet
metal to screen magnetic fields. Aluminum plates protect only
against electrical fields.

Optional on the primary: Replace the 220 uF smoothing
capacitors C1 and C2 by 470 uF capacitors. This reduces
primary ripple, which helps output regulation at full load.

Strona 6 z 7

2012-01-21

If everything is all right up to now, one can proceed with the exciting test at 230 Vac. The
laboratory power supply, the 48 V transformer, the measuring instruments and all
provisional cable links attached for the test etc. must obviously be removed. The car
lamp, is further needed as a load, and for the functional checks. If after application of the
230 Vac mains voltage, the lamps light up brightly, the output voltage amounts to 13.8 V,
and no undefined noises, or smells are noticeable, you have won the first round. If a non
recognisable error has passed the pre-testing, the two switching transistors, and copper
tracks say good-bye, with a more or less loud bang.

For the following load test, some high power resistors with 1 Ohm resistance, and
sufficent power rating are required. The current flowing with this load should not cause
excessive heating of the rectifier diode, and the switching transistors during a 5 minute
test period.

Warning: Check temperature of components only if the mains voltage is switched off

Cooling of the switching transistors Q1 and Q 2 at a continuous current of 15 A has to be
proved in any case. When exchanging the small heat sinks, note that they form an
electrical connection between copper tracks on some boards. Replace the missing
connection by wire links. As one can see on the photo, I did not take this measure, for
further power improvement.

Operation Experience

The modified board was permanently installed in the SP120 speaker cabinet that matches
my transceiver. The mains lead exit from its rear, which also carries the DC terminals, an
on-off switch, the additional mains filter and a small 12 V blower. A green LED power-on
indicator was inserted in the front panel into a 5 mm hole. I had installed the small blower
just in case, but found it superfluous. At the low duty cycle of CW and SSB, none of the
components are getting hot. The power supply has been in use for several years, and has
given no problems.

Fig. 8: Modified power supply board i

Strona 7 z 7

2012-01-21