Author Topic: Unstable current limiting  (Read 484 times)

0 Members and 1 Guest are viewing this topic.

Offline jtruc34Topic starter

  • Regular Contributor
  • *
  • Posts: 62
  • Country: ch
Unstable current limiting
« on: March 06, 2023, 06:05:54 pm »
Hi,

Attached below is a schematic of a kind of regulated 15V power supply (don't mind the use of the LM2931, it only serves as a low drop-out voltage reference and as a overtemperature limit). I added a current limiting circuit for overcurrent protection (mainly for Q5 and Q7).

If C7 is not present, it works as intended, but I want to allow for short current spikes which wouldn't damage Q5 and Q7. I thus thought of slowing down the feedback loop by adding R22 and C7. If I do that, the circuit becomes unstable (current oscillates between 3 and 10A for a load of 1.5 Ohms).

Before I do the full small signal loop analysis to see where the unstability is coming from and where to put R22/C7 instead, is there an obvious way to do what I'm trying to do while staying stable?

Thanks
 

Offline Vovk_Z

  • Super Contributor
  • ***
  • Posts: 1458
  • Country: ua
Re: Unstable current limiting
« Reply #1 on: March 06, 2023, 06:40:24 pm »
Is there really a need to pass current spikes through? What if they will be smoothed by a current limiting circuit?
 

Online Benta

  • Super Contributor
  • ***
  • Posts: 6257
  • Country: de
Re: Unstable current limiting
« Reply #2 on: March 06, 2023, 08:00:15 pm »
R12 seems very high to me, 10k...22k would be closer to the mark.
C7 adds a pole to your current feedback loop. What that does to stability is anyone's guess. I strongly suggest a simulation run.
 

Offline jtruc34Topic starter

  • Regular Contributor
  • *
  • Posts: 62
  • Country: ch
Re: Unstable current limiting
« Reply #3 on: March 06, 2023, 08:40:43 pm »
Is there really a need to pass current spikes through? What if they will be smoothed by a current limiting circuit?

It will serve for audio, and it may require big current spikes. They could be bypassed by C2 anyway, but I'd prefer to not rely on it.

R12 seems very high to me, 10k...22k would be closer to the mark.

Why do you think so? 220k is sufficiently low to fully saturate Q6. BC546B's have a minimum current gain of 200. Do you suggest lowering it to discharge Q8 gate capacitance more quickly? I don't particularly need a fast response, quite the opposite. Plus decreasing R12 would mean higher loop gain and potentially less stability.

C7 adds a pole to your current feedback loop. What that does to stability is anyone's guess. I strongly suggest a simulation run.

Adding that pole is kind of the goal, but maybe in a more controled manner. What do you suggest to look at in the simulation?

Thank you both for your help
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf