C128D power supply voltage too high

[petervg]

HigGuys,

I have a C128D with a german power supply that has the 5V output set to 5.5V. I'ld like to dial that back down to 5V, but that seems to be impossible.
When you look at the schematic, the output voltage can be set using R118. When you turn this counter clockwise, the output voltage should drop (which it does downto 5.4V). I'm observing the pulse width on pin 1 of IC101 that as expected lowers when I lower the output voltage. But at 5.5V, the pulse width is barely long enough to keep the power supply running at all. When I turn R118 couterclockwise, the output voltage starts to lower, but when I hit about 5.4V, the pulsewidth becomes too small and the power supply just stops completely.

In the screenshot below you can see the PWM signal on Pin1 of IC101 (yellow) with an output voltage of 5.5V (purple). As soon as I get down to 5.4V, the dutycycly drops to 0 and the power supply just stops. The AC input voltage here is 229Vac with the DC bus on the power supply (measured over C5) is 306Vdc. I don't have a lot of experience with these kind of power supplies...

 

[petervg]

Just an update:

I had already replaced all polarized capacitors on the secondary side (C108, C107, C102 C105 and C106). Also took out T1, T2, T3, T4 and T102 to test them - all are fine. Also the diodes D103, D4, D5, D2 and D110 have been taken out and tested - all fine.
Crowbar circuit was also tested and working (I accidentally raised the 5V line a little too high triggering the crowbar ).

 

[MiaM]

What does P2 pin 1/2/3 go to? It seems like anything connected there could interfere with the regulation.
C113 is also a polarized capacitor. If it's bad then the OP amp might not work properly. In particular it's power rail might have enough ripple for the regulation to not work, but at the peaks of the ripple it might work long enough for the crowbar to still trigger correctly

(Going off on a tangent: note that all text is in italic. This makes me think that they used an East German drawing template thing for the text, as those were mass produced and sold in various places in the west, and their legally binding standardization required italic in technical drawings, and usage of their TGL-something font, that is similar to but not identical to the DIN font).

 

[petervg]

What does P2 pin 1/2/3 go to? It seems like anything connected there could interfere with the regulation.
C113 is also a polarized capacitor. If it's bad then the OP amp might not work properly. In particular it's power rail might have enough ripple for the regulation to not work, but at the peaks of the ripple it might work long enough for the crowbar to still trigger correctly

(Going off on a tangent: note that all text is in italic. This makes me think that they used an East German drawing template thing for the text, as those were mass produced and sold in various places in the west, and their legally binding standardization required italic in technical drawings, and usage of their TGL-something font, that is similar to but not identical to the DIN font).

P2 pin 1/2/3 are for a 12V external fan which I did not connect during the test. I'm going to replace the fan anyway with one that has a larger airflow to make sure the internals are cooled as good as possible
I already had replaced C113 with a new one and just double checked the polarity which is fine. I also verified the power supply of the opamp (sorry, comparator - it's an LM339) and this is about 20V without any ripple and well within the operating limits. Seems this is all good.

I don't competely yet understand how the PWM signal of IC101 is actually used in the primary oscillator, so I'm trying to understand this better by performing a few more measurements. Below I have turned R116 to get my 5V rail as low as possible without the power supply stopping. It's now at around 5.6V and I can see a dutycycle of approx 10% (fluctuates about +/-5%). But it's very close as sometimes the dutycycle drops a bit more and the power supply stops and needs to be powercycled.

Blue: 5V output voltage (is at 5.6V here - can't get it any lower)
Purply: PWM signal at pin 1 of IC101 (27Khz signal with a dutycycle of around 10% which fluctuates around 5-15%, so regulating seems to be done properly)
Yellow: output voltage of the transformer measured at the anode of D102 (labeled "frei")



and below exactly the same measurement, but with R116 turned to get my 5V rail as high as possible before the crowbar kicks in. Dutycycle is now at around 20% and the 5V rail has risen to just about 6V.


So the power supply does seem to work and I feel it responds the way I would expect. But I have no idea why I can't get the 5V rail below 5.6V as the duty cycle would drop too low to keep the power supply running...

 

[MiaM]

sorry, forget what I wrote about P2 pin 1/2/3 - I misread the schematics and missed that these are connected to the 12V and ground rails. Ooops.

Re the PWM and whatnot:

I admit that I haven't tried finding a component list, so I'm just guessing what the components are,but:
I assume that T1 is the big transistor that drives the main current through the transformer, and R11 has a low resistance. If a too high current flows through the transformer, T1 and R11, the voltage across R11 would be high enough for T4 to start conducting, cutting the base feed to T1. (R9 ensures that a current spike won't rush through the base-emitter on T2).

R14, R15 and DZ1 feeding C8 ensures that T1 will periodically get pulses at the mains frequency to start the power supply (and in theory to restart it if it for some reason stopped (like due to a brownout or temporary overload?).

I'm not 100% sure how the oscillator works, but T2 and T3 discharges C6 when T1 receives enough base voltage.

I think the circuit depends on the transformer being inductive, and the delay in the transformer creates a pulse out on the pin 12+15 winding a moment after T1 has sent out a pulse, creating the next pulse, rinse and repeat.

The comparator feeds a separate transformer that feeds current into T4, the same way as an overload condition would do.

Note that T102 is a PNP transistor connected in a way that makes conduct when the comparator output is low. Thus the pulse widths you see aren't 20% and 10%, but rather 80% and 90%.

And since TR2 is a transformer and not an opto coupler, there is a limit where it obviously can't conduct a "100% pules width" (=DC).

This makes me think that the PSU has a limited output current range where it can provide correct voltage, and if you pull too little power it will produce a slight overvoltage.

In other words, connect a dummy load, for example automotive light bulb, that corresponds to the minimum consumption of a bare C128D (I.E. the same as a bare C128 + a 1571), and check the voltages again

 

[petervg]

@MiaM: I didn't have any lightbulbs, so I had to order some power resistors to put a load on the output. And you're correct. Turning the potmeter to obtain an output voltage of 5,6V (below this, the primary oscillator would stop) and loading the power supply with a power resistor that pulls 1A reduced the output voltage to about 5.2V.
But this made me rethink if I would reuse this power supply. I prefer to have a rock solid 5V no matter what the load to power my C128D .

Thanks for the explanation! Really appreciated!

It was fun and very educational to work on this, but I'll be putting this PCB in my spare parts bin and I replaced it with a modern Meanwell power supply. Still need to add some sort of grommet to the mains input cable and in she will go...