5

I am fixing a Commodore PC-1 PSU for a friend and want to know if the PSU is behaving as expected. No modern PSU suggestions (e.g. Meanwell), if you don't mind! :)

Question: What are the voltage tolerances and actual current draw for Commodore PC-1 motherboard?

Challenge: I do not have the motherboard, as my friend only sent me the PSU. So I had to simulate the motherboard's capacitance (see 'Adapter board' below).

My main concern is that the +12 V output only actually ever goes up to 12 V if I set the +12 V load to 100 mA and the +5 V load to 3.5 A. If the +5 V output is loaded with 1 A and the +12 V output is loaded with 100 mA, I only get 11.46 V on the +12 V output).

Working on the idea that the motherboard might be tolerant of 5% difference, perhaps anything above 11.4 V is fine? Seems quite easy for the 12 V output to drop below that voltage though.

Commodore PC-1 PSU

In case you're wondering, the burn mark is from the old 8k2 5 W PWM bootstrap resistor that drops about 100 V (dropping about 2 W at 120°C). I installed a new 7 W resistor and lifted it off the board a little (mostly just for fun, since the old resistor was quite functional).

The maximum ratings, according to the schematic (from the manual) appear to be:

  • +5 V 3.5 A
  • +12 V 1.4 A (2.2 A)
  • -12 V 0.2 A

I'm actually not sure what "(2.2 A)" means. Perhaps it's a maximum peak or surge current?

Output measurements

All 3 outputs (+5 V, +12 V, -12 V) have an adjustable DC resistive dummy load attached simultaneously.

Edit: Updated measurements below by measuring closer to the PSU so that voltages aren't affected by wire resistance.

Adapter board: I made a custom little adapter board with caps to connect the dummy loads and simulate the motherboard capacitance (since I don't have it; my friend only sent the PSU). I also tested with the ground spring on the probe across the 5 V 220 µF cap, and that reduced noise by about 10 mV; so you can probably knock off that much from the Vpp measurements below (I'm using the ground lead for convenience).

Adapter board

I based the capacitor values on the block caps from the manual (page 17).

5V block caps 12V block caps

The first voltage is from a multimeter, and the Vpp (peak to peak ripple voltage) is from the scope.

For test #0 to #2, the ripple is caused by the PSU not fully starting; it seems to need about 110 mA current draw on the +5 V output before it stabilises.

Are these voltages normal? What current does the PC-1 normally draw?

Test #0:

  • +5 V 0 mA: 5.14 V (Vpp: 340 mV)
  • +12 V 0 mA: 11.10 V (Vpp: 464 mV)
  • -12 V 0 mA: 9.55 V (Vpp: 2.64 V)

Test #1:

  • +5 V 10 mA: 5.17 V (Vpp: 336 mV)
  • +12 V 10 mA: 10.68 V (Vpp: 1.30 V)
  • -12 V 10 mA: 9.55 V (Vpp: 2.08 V)

Test #2:

  • +5 V 100 mA: 5.19 V (Vpp: 320 mV)
  • +12 V 100 mA: 11.31 V (Vpp: 2.50 V)
  • -12 V 100 mA: 12.22 V (Vpp: 1.28 V)

Test #3:

  • +5 V 200 mA: 5.19 V (Vpp: 26 mV)
  • +12 V 200 mA: 11.19 V (Vpp: 76 mV)
  • -12 V 200 mA (max): 12.22 V (Vpp: 32 mV)

Test #4:

  • +5 V 500 mA: 5.17 V (Vpp: 28 mV)
  • +12 V 500 mA: 11.17 V (Vpp: 48 mV)
  • -12 V 200 mA (max): 12.23 V (Vpp: 32 mV)

Test #5:

  • +5 V 1.0 A: 5.13 V (Vpp: 28 mV)
  • +12 V 1.0 A: 11.20 V (Vpp: 60 mV)
  • -12 V 200 mA (max): 12.27 V (Vpp: 32 mV)

Test #6:

  • +5 V 3.5 A (max): 4.96 V (Vpp: 29 mV)
  • +12 V 1.4 A (max): 11.53 V (Vpp: 60 mV)
  • -12 V 200 mA (max): 12.31 V (Vpp: 32 mV)

Edit: According to Bruce Abbott's answer, the Commodore PC-1 might draw a similar current to the IBM PC (5 V 2.3 A). Assuming the floppy drive would draw 12 V 250 to 500 mA and there's nothing drawing from serial, here's a final test with current draws I'm guessing would be comparable:

Test #7:

  • +5 V 2.3 A: 5.02 V (Vpp: 25 mV)
  • +12 V 500 mA: 11.52 V (Vpp: 40 mV)
  • -12 V 100 mA: 12.33 V (Vpp: 30 mV)

Here's what the scope looks like on the max load test: Max load ripple measurement

General observations

From what I can tell, the PSU regulates the voltage based on the 5 V output, as that's fairly stable (always between 4.8 V and 5.2 V) no matter how I load up the 3 outputs. I observed that the load I put on the 5 V output will affect the +12 V output voltage quite significantly (10.8 V to 12 V).

I noticed similar behaviour for the Commodore Amiga 1200 PSU, and that seems to work fine.

Test setup

Fun little weekend project. The AC source is isolated, and goes through a dim bulb limiter (bypassed at full load). The dummy loads are DC variable resistance. More details about the adapter board earlier in the question.

Test setup

13
  • 1
    What is the exact question really? The computer will surely draw more than 0.1A of 5V so you have no problem. It is typical that old computer power supplies do not work properly and are out of regulation if you don't give it a proper load, so they might not even start up with too little load. There are no specs but it would be a good assumption that anything within 5% of the nominal is fine, as long as it's not 5% at the supply but worst case on the motherboard.
    – Justme
    Commented Aug 27, 2023 at 9:00
  • 1
    Yes but when you are drawing 1A from 5V and it has dropped to 4.67V, are you loading the 12V at all? They share a feedback path and it won't work if 12V is unloaded.
    – Justme
    Commented Aug 27, 2023 at 9:19
  • 1
    Another thing what I forgot to say earlier is that you are measuring the supply in isolation, likely with resistors as dummy DC loads. So you are not measuring the actual system, and the motherboard is not a DC load resistance. At least the capacitors of the motherboard should be taken into account, as they affect the stability as to the switch mode power supply ripples from the switching, they are an AC load and reduce the voltage peaks.
    – Justme
    Commented Aug 27, 2023 at 9:41
  • 1
    You should maybe just measure with the actual supply powering the actual computer. Or have a modern supply powering the original computer. It certainly does not look like you get overvoltages and the supply did not itself blow up yet on your max load testing. Definitely add capacitors at the load, have wires approximately same length and gauge as original for matching drop at matching currents. And typically ripple and noise measurements with a scope are measured with 20 MHz bandwith limit and over a capacitor. The -12V comes from linear regulator, it may need min 10..20mA but not 200mA.
    – Justme
    Commented Aug 27, 2023 at 13:47
  • 1
    Approximate the capacitance on the rest of the system. And the motherboard and other parts have polarized caps so of course you can use polarized caps, you don't have alternating voltages between positive and negative. But you need to have both bulk electrolytics for the low frequencies and maybe ceramics for high frequency bypass, just like what the motherboard likely has. Forget tantalums, for many reasons. If you look at modern supplies, positive supplies are 5% and negative are 10%, over rated loading between supplies, which you don't know, and that's the difficult part you are asking.
    – Justme
    Commented Aug 27, 2023 at 19:00

1 Answer 1

9

Test #6:

+5 V 3.5 A (max): 4.96 V (Vpp: 120 mV)
+12 V 1.4 A (max): 11.53 V (Vpp: 400 mV)
-12 V 200 mA (max): 12.29 V (Vpp: 88 mV)

Are these voltages normal?

Yes, those voltages are 'normal'.

The switched-mode power supply regulates output voltage by varying the duty cycle of the waveform going into the transformer. This affects both the +5V and +12V outputs proportionally. However the feedback circuit only (or mostly) monitors the +5V rail because its voltage is more critical. If the +5V rail is loaded with high current but the +12V rail isn't, the +12V rail's voltage will increase due to the higher duty cycle required to keep the +5V rail voltage up.

The power supply is designed so the +12V rail will have the correct voltage when 'normal' current is being drawn from the +5V rail. At lower +5V rail current the +12V rail will have lower voltage due to the lower duty cycle.

What current does the PC-1 normally draw?

I don't know (the service manual doesn't say), but it should be similar to an IBM PC. According to the IBM 5150 Technical Reference its 63 watt power supply is rated for a minimum current of 2.3 amps at +5 V and 0.2 amps at +12 V, both with a tolerance of +5% and -4%. Therefore only test #6 is relevant, as all the other tests didn't draw enough current from the +5V rail.

4.96 V is within 1% of 5 V, and 11.53 V is (just) within 4% of +12 V, so your voltages are OK.

In practice the +12V rail could probably go as low as 11.0 V without problems, as it is normally only used to power the floppy drive motor and RS232 serial (both of which have fairly wide supply voltage tolerances). The +5V rail is more critical because if it drops too low even momentarily the digital logic may glitch or reset.

8
  • Excellent answer! I didn’t think to use the IBM PC as a reference, but that makes a lot of sense. Bravo. Commented Aug 28, 2023 at 3:33
  • Do you happen to know roughly the total capacitance of the average IBM PC motherboard? For the ripple measurements. Commented Aug 28, 2023 at 4:53
  • Find the electrolytics on each rail on the PC-1 motherboard and count up the total. I don't think it will make much difference because the PSU already has a lot of capacitance on each output. The ripple levels you measured look ok to me. If you want a better idea of what it will be on the motherboard then just plug it in! Commented Aug 28, 2023 at 5:53
  • Good idea. One small problem tough... "I do not have the motherboard, only the PSU." "I am fixing a Commodore PC-1 PSU for a friend" -- my friend might not be quite technical enough to figure this out (though I will ask). There's always the service manual: github.com/nbolton/amiga/blob/… Commented Aug 28, 2023 at 9:07
  • 1
    Looking good... Commented Aug 29, 2023 at 3:09

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .