Unstable power supply output or normal?

[Fusion916]

Attached is the output of my power supply with about 1A load (ac coupled). It looks like the frequency is 120hz/60hz to the input ripple looks to be relevant.

This power supply is unstable with no cap out the input (oscillates at about 50khz), but adding at least 20uf output cap seems to stabilize it, just have the ripple attached.

Is this still unstable? I get this type of ripple no matter how much output cap I have above a certain threshold (about 20uF).

[Mooly]

Moved to power supplies

[Mooly]

What is the PSU and what type is it ?

[jan.didden]

It's not unstable; as you noted yourself, it has some residual mains breakthrough.

Based on the pulse shape it looks like it comes from the narrow, high current cap charging pulses. These occur when the secondary voltage in its cycle climbs above the cap voltage and the diodes start to conduct to top up the caps.

These current pulses can be quite high, especially if you use a lot of capacitance after the rectifier. Even a fractional ohm resistance in the ground or return line can cause some mV across them. A simple test could be to remove all load from the psu, that should also make those pulses disappear or greatly diminish.

Then again, you may be just having your scope probe ground lead at a bad point - try to measure directly at the supply output or directly across the capacitor.

jan

[DF96]

The output with a cap looks like a grounding problem - charging pulses are getting into a ground somewhere.

The oscillation without a cap is probably a separate issue; presumably an unstable design for a stabilised supply?

[Mooly]

Just adding to what Jan mentioned... try connecting your scope probe tip and its ground lead together. You should get a noise free trace of course. Now keeping them shorted connect the two to the same ground point you used before. If the trace now shows a problem, then that suggests a grounding issue via the scope.

[jan.didden]

Missed the oscillations bit. Most regulated supplies (with feedback) are unstable without an output cap. Data sheets will give min required cap - read them!
Most are OK with anything above a few uF. And don't use film or other boutique caps there, you need the ESR for stability.

jan

[gootee]

Maybe the ripple minima at the regulator input are simply getting too close to the reg's dropout region. Did you try adding capacitance before the regulator? Can we see the input voltage waveform?

Also, are your scope traces always that thick? You might still be seeing some HF oscillation. But it's difficult to tell if it's "real" or not.

Maybe you should also zoom in on the diode turn-off time, while probing across the bridge's input.

[Fusion916]

Originally Posted by jan.didden

It's not unstable; as you noted yourself, it has some residual mains breakthrough.

Based on the pulse shape it looks like it comes from the narrow, high current cap charging pulses. These occur when the secondary voltage in its cycle climbs above the cap voltage and the diodes start to conduct to top up the caps.

These current pulses can be quite high, especially if you use a lot of capacitance after the rectifier. Even a fractional ohm resistance in the ground or return line can cause some mV across them. A simple test could be to remove all load from the psu, that should also make those pulses disappear or greatly diminish.

Then again, you may be just having your scope probe ground lead at a bad point - try to measure directly at the supply output or directly across the capacitor.

jan

Yes there are no pulses with no load, and increase in amplitude with load.

Is there anyway to diminish or eliminate these charge pulses?

[magnoman]

Really would help to see the input to the regulator, your at 20mV peaks, thats 40db down from 2V not very good but maybe within the objective of the design.

Thanks
-Antonio