Then what can be done to reverse the situation? Is it all about the output cap?
Next week I will make measurements again. Until then I have to figure out what to try. I used a Panasonic FM 220uF as an output cap - maybe it's got low esr for the application? My first try would be to replace it with something else - a 47uF tantalum (which I would like to avoid) or a crappy 220uF cap.
But it seems weird that compensation caps did not do anything. Shouldn't it become stable if I reduced the bandwidth so drastically? Dominant pole compensation.
If it's stable in simulation and unstable on the lab bench, the thing to do is discover why the simulation is wrong.
If it's unstable in simulation and stable on the lab bench, the thing to do is discover why the simulation is wrong.
If it's unstable in simulation and unstable on the lab bench, you can choose whether to work on the simulation first or the physical lab prototype first.
If it's stable in simulation and stable on the lab bench, then you can move on to the next item on your checklist of validation / certification / verification.
You may discover that circuit design isn't trivially easy.
If it's unstable in simulation and stable on the lab bench, the thing to do is discover why the simulation is wrong.
If it's unstable in simulation and unstable on the lab bench, you can choose whether to work on the simulation first or the physical lab prototype first.
If it's stable in simulation and stable on the lab bench, then you can move on to the next item on your checklist of validation / certification / verification.
You may discover that circuit design isn't trivially easy.
On both occasions, I observed some quickly dying oscillations around the point of current load change, as expected. On scenario (1), it had a peak of about 1V, whereas on scenario (2), it had a peak of about 1,5V. In both cases, the period of oscillations was in the order of 20-50 ns, if I recall correctly. Sadly I forgot to take pictures - will do next time.
The oscillations died after about 100ns. But here is an important note: I never used decoupling caps close to the load, on purpose.
But when I did, the oscillations completely dissappeared.
If you are saying that that ringing is more of an artefact of the fast dV/dt of the test signal on the wiring rather than anything related to the regulator, I agree with you.
Jan
Yes, that is what I am implying. I think that a few nanoHenrys and few picofarads are possible to create MHz ringings, that need lots of resistance to be damped.
I think that a 20ns risetime pulse is way too fast for an audio regulator, and anyway such a load should have proper decoupling.
But to be sure, next week I will perform more measurements and take pictures, covering many cases.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.