Help fix amp - rail voltage gone crazy

[Eva]

The regulation control loop is compensated to be stable with the inductors in place (with a zero very close to LC filter resonance to provide some phase margin, see RC network before the pot). Removing the inductors moves the LC resonance to a much higher frequency (L changes from 135uH to a few uH from transformer leakage inductance) and is likely to cause oscillation, which may be fatal under load. Also, that kind of inductor-less regulation results in high stress on the push pull MOSFET, rectifiers and +B capacitors, and high EMI, I have analyzed it in detail on Alpine amplifiers.

Both rails are linked by duty cycle, they can't be regulated separately. The only way to get very dissimilar rail voltages on that SMPS topology is a faulty output inductor or capacitor, or a strongly unbalanced load. That's how I figured out that the inductors were the problem.

The feedback circuit just regulates the highest of the rails (at the expense of not controlling the other) but not in the same proportion, my math tells that the negative rail should be limited around 35V (R902) while the limit for the positive one is adjustable from 21V to 38V (VR951).

If it's not working that way, you should check all the feedback components (stuff around the green square) as Perry suggested. There may be an open resistor or solder joint on the negative rail sensing (plus the inductor problem).

[Perry Babin]

There are many millions of power supplies that don't have inductors before the capacitors. I understand that it could change the stability under some conditions, it's not going to cause problems at idle. In most amps (including this one, I believe), jumping the inductor is OK for troubleshooting a problem like the one he had.

Yes, inductors can reduce the stress on power supply components but at idle, it's not a concern.

[Eva]

In power supplies without output inductors the turns ratio of the transformer is substantially lower. In this power supply, due to the high turns ratio, something like 15-20V would be applied to transformer leakage inductance during on time, resulting in very high di/dt and EMI.

Imagine over 100A/us current slope on the primary side... Not healthy...

[Perry Babin]

The only thing I'd like to clarify is that jumping the inductors is not unsafe (especially on this type of power supply) and is not likely to cause any damage when testing.

I did this with several amps I have here. I shorted one inductor (when there were 2 individual inductors), both inductors and 1 or both halves of a common mode inductor. None had any problem that would cause any damage.

In PPI amps and some other amps, they use very large inductor cores. They tend to drop voltage (peak voltage from transformer, minus ringing, is significantly greater than the rail voltage at idle). These operate differently than most amp power supplies, in this regard. I jumped the PPI and the rail voltage was within 0.1v of the voltage without the jumpers. In an amp with this type of inductor and an un-regulated supply, the rail voltage would have increased if it were jumped.

The other amps that were tested used a much smaller common mode inductor or that had 2 individual inductors, the rail voltage was within ~1/2 volt of the peak voltage on the transformer's secondary both before and after the jumpers were connected. The jumpers made essentially no difference in the rail voltage.

In the amp with individual inductors, I shorted between adjacent windings, the regulation held and both rails remained equal.

Manina, I think you should try installing the inductor you pulled from this amp in the other pioneer amp to see if it produces the same results.

If you're concerned that the inductor is shorted or has worn insulation where the windings overlap, you can check it with your meter. Touch one probe to the tinned end of the inductor. Take the other probe and rub it over the area where you suspect that the insulation is worn through. The meter should never change from OL (assuming that it reads OL with open leads). The most likely place for it to short is the point where the terminal winding first meets the rest of the windings.

[manina]

In the W.E. I have checked the section Perry circled (well, squared) and all is OK.

I mounted the inductors I'm sure are OK and the finals. Voltage is OK. It's stable in the 35V for both rails. I'll check the other two inductors before placing them in the other amp that I use as spare.

I'll now change the rail caps and place new ones as, at least, one of the old is gone (it was on the negative rail).
Is it safe to use 35V caps (as the stock ones) or shall I upgrade to 50v? I have anyway to change them and, giving the operating voltage of the amp, the stock ones are borderline (IMHO)

Unfortunately, I have to wait for the IRFIZ44N as it takes some time to get them here (local shop run out easily) then the amp should be fully operational.

I have some IRFI 3205PBF in house - is it safe to use them or it's likely to oscillate?

[Perry Babin]

The drive circuit looks like it should be able to drive the 3205s. You may have to reduce the value of R978 and R979 to 1k or just a bit lower. If you power it up and the FETs run hot at idle and isn't drawing excessive current, change those resistors.

If you only have 35v rails, you don't need 50v caps. Unless you can find 3300uf caps rated for 50v that will fit, go back with the original 3300@35. 35v caps will operate at 35v with no problems.

[manina]

I placed the inductors in the working amp and I can confirm they were the cause of the problems.

With them in place, I ether get high voltage or the led blinks and the amp does not power up properly. I'll rewind both of them to be on the safe side.

As I have anyway ordered the 44n, I'll wait for them rather then to use the 3205 - just to be on the safe side.

So far, I think the amp is fixed.

Thanks for all your help.

[Eva]

Remember to make the inductors vibration-proof once you are going to leave them in the PCB...