GK 1001rb Loud Pop on power down

[boxofknobs]

Just got a GK 1001rb as the title suggests. I plugged it in, turned it on and it sounds great. No noise or any apparent problems. On power down it made a really loud pop which actually sounds destructive. It goes away when I turn the woofer volume to 0, IE no BLAM pop. I took it apart and tested some of the power caps and smaller caps but I cant seem to find a problem.

Does anyone have any suggestion as to which part or section of the preamp (I assume)to test in order to kill the pop? Is it possible?

[Enzo]

Millions of solid state amps pop when you power off. Not a defect.

The pop is limited to the power rails inside, which means it can be no louder than th loudest amplified sound. I(f the speakers can handle the amp loud, the pop won;t kill them.

[Steve L]

Thump Fix

This may help.

Gallien Krueger 700RB, 1001RB Turn-Off Thump Modification.pdf

Steve

[boxofknobs]

Thanks for the link! That does help a lot. I appreciate it.

[woodyc]

It seems like the theory behind this fix is that a step doesn't sound as bad as a pulse of the same amplitude?

Because in either case when the supplies are shut down, as the current drops in the FET, the drain voltage is going to shoot up a couple of volts. The difference is that with the fix, it stays up instead of falling back down on the C14 discharge curve. (C14*R28 = 10ms)

[The fix being: make R28 a diode and make C14 470uF.]

I think the source keeps steady during all this because C16*R31 is about 400ms.

Is that what's going on, or am I missing something?

I redrew a part of the schematic while I was thinking about this...

[J M Fahey]

I see it as a power rail decoupling problem.
The Fet is an amplifying stage before the power amp.
The power amp influences the rail voltage by its own current consumption.
If it has a small power off thump (which is natural and mentioned in the service note) *and* it causes a glitch on the main Power Rail which causes a glitch on the Fet power rail which gets coupled to its drain which gets back into the Power amp input, it may have the proper phase and level to drive the power amp output hard against some rail, heard as a *huge* thump.
Clearly original R28+C14 decoupling (plus whatever decoupling appears when deriving 15V from the main rail) is not enough.
Diode + larger C14 means that when the main rail goes down quickly at turn off, this huge negative pulse gets decoupled from the Fet drain.

[Steve Conner]

Yes, it is a decoupling problem, the design of solid-state power amps is a trade-off between PSRR and thump size.

[woodyc]

I see it as a power rail decoupling problem...

Please tell me more about this glitch the power stage puts on the rails... Is it an over or under-voltage? Is it large or small compared to the supply value?

[woodyc]

Steve, I meant to direct that request for information at you too.

By the way, if there is a glitch this may be irrelevant... but I just modeled the thing I was talking about, and its exactly as I said. In this pic, I just shut off the supplies abruptly at 2ms into the sim.

[J M Fahey]

In due order:
1) after you turn the amp off, only way rails can go is "down" , meaning closer to ground.
The big capacitors will discharge at a relatively slow rate, given by the large capacitance and relatively small idle current , *but* if there signal and a load, the amp will feed it a quite larger amount of current, which will in due time discharge the filter caps faster.
In all capacitors, if you pull charge from them, the voltage lowers, so this sudden discharge I just mentioned will produce a voltage step ("down" or negative step, if we are studying the positive rail).
Since the Fet is fed fom the positive rail, its own 15V rail will receive that negative step from the main +V one, so its drain will also show such step.
Attenuated, of course, but *now* we have a "power amp made-signal" (the step) being fed into an earlier stage which then feeds it back into the power amp input.
Did I say "feed" and "back" in the same sentence?
If loop gain and phase are correct, you have an (unintended) closed loop ready to oscillate.
Not necessarily at an audio frequency, slamming the output hard against a rail is also some type of positive feedback.
Google "relaxation oscillator".
And the output will remain full (I guess positive but may be the exact opposite) until those filter caps discharge through the speaker VC.
The Mother of all thumps.
But .... which is the small signal which triggered this snow avalanche?
The original, small, unimportant "pop" or "thump".
And see that it only happens with certain control settings !!!
2) don't understand your simulation, can you please explain?

[woodyc]

2) About the simulation... its just what happens at the drain and source of the FET if the supplies are turned off. The exact circuit is the one I drew above. Its an interesting but fairly subtle thing -- it looks like a little pop, not a big one.

1) If I understand, the glitch you were talking about is just the sharp drop in supply voltage. But the pop is now an oscillation involving positive feedback? That seems more plausible since the GK fix does throw a wrench into a potential feedback path, but I find the rest confusing. For example, I've designed and built relaxation oscillators and I don't see the relevance. Maybe you could explain it more precisely? Also is this a theory, or do you have some evidence?

[J M Fahey]

2) About the simulation... its just what happens at the drain and source of the FET if the supplies are turned off. The exact circuit is the one I drew above. Its an interesting but fairly subtle thing -- it looks like a little pop, not a big one.

It's "a little pop" because :
a) you simulated just a switch opening on the +15V line.
There's *a lot* more going on.
Besides, you simulated an open loop path, I am talking about a closed one, very different thing.

1) If I understand, the glitch you were talking about is just the sharp drop in supply voltage. But the pop is now an oscillation involving positive feedback? That seems more plausible since the GK fix does throw a wrench into a potential feedback path, but I find the rest confusing.

Look at the full system and the interaction of its parts, not separate stages.

For example, I've designed and built relaxation oscillators and I don't see the relevance. Maybe you could explain it more precisely?

Try to understand the relaxation oscillator , not as "a transistor here, a resistor there, the formula was such and such" but as a functional block.
You have one there.

Also is this a theory, or do you have some evidence?

You supplied the evidence (the amp with a problem), I'm explaining it to you.

[woodyc]

I don't follow your "explanation" at all.

The only time I've seen an oscillatory component in these pops is when you have a speaker load -- they ring at the speaker resonant freq. With a resistive load it's just an odd-shaped pulse, not unlike the one in my model above (namely a pointy pulse with a sharper rise and a slower decay.)

Also the power rails don't actually shoot up and down all that fast, nor do they contain any "glitches" when the pops occur. I haven't spent a lot of time on this, but I have looked.

[J M Fahey]

OK, fine.