Strange behavior of Voltcraft DPS 4005 DC power supply

[Momchilo]

Hi all,
I am repairing a Voltcraft DPS 4005 DC lab power supply. It is old, quite cheap and nothing special. I already fixed a lot of stuff a year ago and it is almost working as intended. Now I have the ambition to fix it completely.
There were some burnt components on the secondary side of the auxiliary power supply. I replaced them and all capacitors in this power supply as well.

But there is one big issue left. I connect a programmable load (constant current) to the power supply, measured the voltage output with a differential probe and the current with a current clamp. The measurement setup is like this:




For exactly this scenario the voltage and current look like this:


Something is wrong.

If instead I connect the ground of the load not to the negative banana post of the power supply but internally to the diode D5 that is connected to ground like this:




The voltage and current output look like this:


So this is fine. In my opinion there must be some interference, but I couldn't find the source or any hint. I already soldered two short wires from the pcb to the banana posts instead of the connector and the long wires. But nothing changed. I connected different points of ground together, ground to earth and used a passive probe without any improvements. I have attached the schematics for this power supply.
Do you have any idea? Because at this point I really don't know what to do.

Thanks already for the help.

Best regards
Momchilo

[inse]

How can your electronic load cause such an oscillation?
If it is set to 1A but is drawing a peak current of 8A?!?
Do you have a power resistor or halogen lamp as pure resistive load at hand?
Did you check the connection from the diode to the post?

[Momchilo]

How can your electronic load cause such an oscillation?
If it is set to 1A but is drawing a peak current of 8A?!?
Do you have a power resistor or halogen lamp as pure resistive load at hand?

I don't know. Maybe because the pulses are too short to detect? It is power supply just with the ability to work as an electronic load (GW Instek GPP-4323).
If I use a 5.6Ohm resistor as a load the output looks different but also not good as you can see in the attached picture.
As soon as the power supply is in constant current mode (limit current of PS to 1A) the voltage and current look good again.

Did you check the connection from the diode to the post?

Yes of course. I also added another wire without any improvements.

I tried a lot of things yesterday. I tried also heated up and freezed the area without any changes as long as the banana post is used. I desoldered the blue wire from the banana post and then sometimes the output is good and sometimes not. In this case freezing the area helps. I thought maybe there is an issue with the banana post. But I can't see what should be wrong there. The resistance is like 0.01ohm between the soldering tab and the end of a plugged in 1m wire.

Thank you for your help so far inse

[inse]

What’s happening when your minus connection is at the diode and not at the post: there is a common mode choke between the board and the posts, which has no effective inductance. When you bypass the negative winding, the compensation is gone and you have an inductance in the circuit.
This obviously dampens the oscillation issue.
Also your electronic load is not so good…
There seems to be an issue in the feedback loop of the power supply if it also oscillates with a pure resistive load.

[Momchilo]

What’s happening when your minus connection is at the diode and not at the post: there is a common mode choke between the board and the posts, which has no effective inductance. When you bypass the negative winding, the compensation is gone and you have an inductance in the circuit.
This obviously dampens the oscillation issue.

I have also removed the core and it even looked a little better without it. When I desoldered the two wires of the banana posts and moved them around with the attached load sometimes the output was good and sometimes it was bad.

I connected another power supply between the diode and the banana post with 1A and measured the voltage drop. It is ~20mV without any additional wires. Looks good for me.

[inse]

The schematic is 💩, I have no idea what’s going on on the output board sheet…

[Momchilo]

The schematic is 💩, I have no idea what’s going on on the output board sheet…

Yes I added some lines in the schematic. Maybe it helps. The whole power supply is not worth it. On the one hand I have the ambition to find the problem and on the other hand I am about to give it away.
It is built very simple and I have the schematics. So it should be possible to fix it. But I've already put more time into this thing than I care to admit
Every measurement made looks wrong. You have this oscillation everywhere when you see it on the output.