Allen & Heath PSU goes bonkers!

[nickb]

I'm stumped. Totally slamming head against a brick wall kind of stuff.

Here is the power supply from a mixer. It's about as simple as a flyback switcher as you can get. The problem is that the primary voltage regulation is not working as it should. In other words it produces outputs but they are about 20% lower than normal and they drop a bit further than I'd expect with increasing load. I've loaded the +15v outputs with 200 ohms each or 60mA a piece. This is very light but when installed in the mixer the loading is not an issue

The odd thing is the voltage at the sense pin VFB of U1 is rock steady at 2.5V, just as it should be. So it IS regulating after a fashion. I would not expect the secondaries to be well regulated but it should be better than it is. It seems that the front end is all working as it should. Since I'm getting all the outputs the secondary is working too - nothing is getting hot either. I note that the on-time of the FET does not seem to change as it should with increasing load. Energy balance is maintained by the drop in the output voltage instead of more energy going in as it should.

What about the transformer? If it had a shorted turn the primary current would be huge and so the current limiter would kick in. This isn't happening as (1) the 2.5 VFB voltage is correct and (2) the voltage on the ISEN pin is (well) under the 1V limit.

Snubbers? I checked D5 and D6 and I see the flyback voltage is clamped lower than the TVS diode's limit so they are not really doing anything under my bench test conditions. That leaves R9 and C1. R9 is a little discoloured but still functional. I haven't looked at C1 but I would expect to see ringing on the primary if it was not working and in any case it not obvious to me why they would produce the symptoms.

Controller chip? Has to be! Naw, I changed that.

So what is going on? Over to you, wiser denizens of the MEF...

[Enzo]

I haven't had a donut yet so...

My first reaction is the sense circuit, R10-11-12.

[nickb]

I haven't had a donut yet so...

My first reaction is the sense circuit, R10-11-12.

I haven't had my donut either so there is a clear correlation. Did you know you can't get donut holes here in the UK? Well not this part anyway.

Thanks Enzo. That was the first place I looked too, but the voltage is small, just as it should be. I'd expect it to be under 1V when operating flat out. With the load I have it about 50mV. The very first test I did on this was to add a 470pF across C24 in case it had gone open and was allowing switching spikes through and so causing a premature end to the switching pulse.

I'm thinking I may be looking under the wrong rock. Right now it's very lightly loaded ( about 3W) and that stops it from regulating although I would have expected the outputs to be high. I'm now thinking that the real issue might be the mixer is drawing too much current so causing the PSU to hit the current limit and lower the voltages to compensate. I'm going to load the PSU more and see where that gets me.

[km6xz]

Do you have a current regulated bipolar bench supply to run the board. I had one a few months ago with low voltage and it was a head scratcher until I concluded it was a cracked core of the transformer. There is so little current load from the board, and parts are impossible to get here from A&H that I whipped up a small linear supply while searching for the correct part.

[Enzo]

You are probably right, but I was imagining something like one of those half ohm resistors be open, so that voltage, small as it is, might be twice what it should be. Just a thought.

Load it more or connect the mixer and measure the draw.

Are you on an iso so you can ground a scope to L_GND?

What is VCC in the IC, maybe 15v? Clean?

No donut holes? What kind of place is that?

[nickb]

@Enzo: I jigged it up with a heaver load with no improvement - so that is ruled out. With this extra load I can see the auxiliary regulation is working as should and the pulse width being modulated in a stable fashion too as I vary the voltage in using a variac. Yes, I have an an iso in the path

@Stan. Thanks. I assume you mean to run the mixer off the bench PSU. Yes, I have one. I will try that.

I did wonder about a cracked core too but I would have expected the primary current to be excessive due to the lower inductance. Therefore I had put that low on the list of suspects. Anyway, based on what you said I took it out and cannot see any issue, but I imagine that heat stress might cause a fracture and the hottest part would be the centre leg which I cannot see without breaking it as it's a glued assembly.

I measured the primary inductance as 160uH using a meter, a far from perfect method. At 100KHz and 50% duty cycle that implies 10.5A peak current. The limit is set to kick in at 4A. So that is a little suspicious.

I'll have to try a substitution. As it happens A&H are about an hour's drive away from me if need be. Need anything while I'm there?