SMPS (tl494) bench supply unstable

[gvc]

I have a 'Voltcraft VSP 2405HE' dual output bench supply where one of the PSU's is unstable. The PSU has 2 independent TL494 based SMPS boards, fed by a shared primary bridge rectifier and active PFC board. About two times a second the output voltage of one of the PSU's jumps a few 100mV, which over a few seconds might make the output drift with more than 500mV before returning to its set point. Over an hour the output might drift away more than 1V. The second PSU's output is rock solid.

The device has a 'Series tracking' mode where the second PSU takes over the voltage setting of the first PSU. When I enable this mode also the second PSU becomes unstable, exactly replicating the voltage drift of the first one. This makes me think there's something wrong with the voltage setting/reference circuit of the first PSU, however I don't understand how it's supposed to work Full schematics are linked at the bottom, this this a diagram I made that depicts the basic voltage setting/reference circuit:

On the right side we have the output of the regulated PSU. At the bottom are the coarse and fine output voltage setting potentiometers. The TL494 is fed by an auxiliary linear PSU with no direct connection to the regulated PSU. This linear PSU is stable, as is the 5V reference out of the TL 494.

Who can clarify the working principle of this circuit? More specifically:

• I expect pin 2 of the TL494 to get provided with a stable target PSU output voltage, obtained from pin 14, but how can this voltage be stable if the 2 potentiometers at the bottom have no connection to the TL494's ground? When I measure between the 5k trimmer and the 100k potentiometer or on pin 2 of the TL494 (with reference to PSU out -), I see the same voltage jumps as on the PSU output. No surprise the PSU is currently unstable if both pin 1 and 2 of the error amplifier are unstable
• What's the purpose of the circuit with the 2 zener diodes? It looks like a waveform clipper circuit, but again, how can it work if the PSU's outputs have no reference to the ground of the TL494, and with different resistors in series with the zeners which still allow voltage to rise?

Any help is appreciated as I have real difficulties troubleshooting this issue if I don't understand what makes the circuit stable or unstable. Feels like I'm chasing multiple moving targets at once.

Full schematics: https://asset.conrad.com/

 

[gvc]

After 2 days of troubleshooting I finally got the PSU fixed. The problem is almost to stupid to admit, but anyway.

The PSU's have external sense connectors at the back of the unit with, when not being used, the sense in and out's connected via bridges. I removed these bridges to check the voltage on the connectors, reinstalled them, and ... problem solved. I had something similar happen with a large studio mixing desk long ago. Something with oxidation.

That whole voltage sensing circuit remains very sensitive I guess, A serious length of wiring running through the device, bundled with wires carrying 420V not stabilized DC, multiple connectors in series and, looking at the schematics, extremely small currents flowing in the circuit.

So problem solved, but if anyone can give some insights on how the voltage regulation circuit above works I'm all ears.

 

[tepalia02]

It seems like the external sense connectors and their connections play a crucial role in stability. Removing and reinstalling the bridges resolved the issue. It suggests that there might have been some contact or oxidation issues affecting the voltage-sensing circuit.

 

[Tony Stewart]

Lambda once made the best linear supplies with remote V sensing and R controlled current limits and voltage control. (ohms/V) and was very stable under transient loads with no drift. I used them on long umbilical cables to charge silver oxide batteries on payloads.

This one compares voltage drop on 470 with the voltage drop on the 3k3 and 5K pot for negative feedback for a ratiometric gain on 2.5Vref as a comparator