Cheap bench supplies enter into self-oscillation when used with non-linear load.

[LinuxHata]

This is very common and happens with most of inexpensive chinese bench power supplies and modules, such as DOYUI, WANPTEK, XY6012, XY6020L and so on.
The symptoms are the following.
Say, I have a led that needs 10V @ 1A.
Set output voltage to 10V and current limit to 0.5A

I connect the led and it starts to flicker.
When measured with scope, I see the output voltage swing from 0 to certain voltage (until current limit kicks in), with sawtooth waveform of relatively low frequency (1-200-300Hz). That frequency is dependent on the current limit - say, if led working current is 1A and I set limit to 0.1A, flicker will be about 10hz, but if I set current limit for same led to say, 0.9A, flicker will have higher frequency - about 30-40Hz. The flicker frequency is also depends on the maximum output voltage - say if I set output voltage to 30 volts instead of 10 volts, it will have approximately 3x times lower frequency. Flicker will be there, as long as CC mode is enabled. If I go to CV mode, there is no flicker.

Adding beefy capacitor at output (even via resistor), does not help at all - of course, it reduces the flickering, because voltage now won't sag that much, but problem is still there.

This is not an issue of particular unit - I tested a whole lot of them, and most of them have these issues. Even modules claimed to be CC led drivers, are having that issue - this one, as example: https://www.ebay.com/itm/166506482076

So is there any way to fix these issues?
I googled for circuits and found this schematic of inexpensive CV/CC module, which has same issues.
Anything can be fixed here?

[Doctorandus_P]

Current limiting on this kind of power supply is more an overload protection then a real current limit. If you want to drive LED's, then use a dedicated current-mode SMPS IC.

[LinuxHata]

Well, Korad, Etommens or Riden are all fine.
I'm interested what's wrong with others and whenever it can be fixed.

[Kleinstein]

It is tricky to get a good current regulation from that type of switching regulator.
It would need quite some experimentation and with the relatively large capacitance at the output the current regulation will never be really fast. A switched mode regulator is also more tricky as it is relatively slow to start with.

One could try replacing the OP-amp with a slightly faster type (e.g. TLC272, MCP6002), to avoid the somewhat nasty delays of the LM358. Another point can be a resistor (~10 K range) in series to the capacitor at the LM358. This is the step from an I type regulator to an PI regulator.

[LinuxHata]

Well, the regulator above I posted the link uses 3843 and 358, and has same issues.
XY6020L and others use totally different components but having same issues.....

[iMo]

Try to disconnect the LED diode coming from the 358 output to -OUT.
It could be it creates a positive feedback when the LED current adds up to the load current..

[calzap]

What models of Wanptek did you use?  We have a couple on our shop bench and have done some LED work with them.  Have had no problems, but mostly used CV mode.  On the electronics bench, I use HP and Instek PSUs.

Mike

[mikerj]

Try to disconnect the LED diode coming from the 358 output to -OUT.
It could be it creates a positive feedback when the LED current adds up to the load current..

I agree, that LED should be returned to the other side of the current sense resistor.