Amplifier startup problem

Are you trying to run this thing with a floating ground ? I just reread the OP and that is what it sounds like.

Well to do that the actual circuitry of the amp must be designed for it and it seems like this IC isn't.

There may be no solution except to use a true split supply.

Refer to the design of QSC commercial amplifiers for an idea how they must be designed. You can't just have resistors do it, the circuit establishes i ts ground and balances itself continually. There is something inherently di fferent about the architecture of the circuit. I looked at the QSC circuit and it was not blatantly obvious but I'll bet it has something to do with h ow the feedback is implemented. I haven't looked at it lately, and when I d id it was only enough to fix the thing. It MIGHT be that it is absolutely f ull complementary, which most circuits aren't. Even then only certain topol ogies will work reliably.

No, at low output all the IC sees is crap on the psu rails. Turn it up and it all goes wrong fast.

yep

no, just someone that overlooked something.

NT

On 18/09/2016 12:00 AM, Phil Allison wrote:

I don't know why you're willing to state as lies things that are readily checked.

Version 4 SHEET 1 1132 760 WIRE 192 -352 112 -352 WIRE 256 -352 192 -352 WIRE 320 -352 256 -352 WIRE 528 -352 320 -352 WIRE 576 -352 528 -352 WIRE 192 -320 192 -352 WIRE 256 -320 256 -352 WIRE 320 -320 320 -352 WIRE 192 -208 192 -240 WIRE 192 -208 176 -208 WIRE 256 -208 256 -256 WIRE 256 -208 192 -208 WIRE 320 -208 320 -256 WIRE 320 -208 256 -208 WIRE 432 -208 320 -208 WIRE 528 -144 528 -352 WIRE 496 -128 432 -128 WIRE 688 -112 560 -112 WIRE 816 -112 688 -112 WIRE 912 -112 816 -112 WIRE 288 -96 -48 -96 WIRE 320 -96 288 -96 WIRE 400 -96 384 -96 WIRE 496 -96 400 -96 WIRE 288 -80 288 -96 WIRE 400 -64 400 -96 WIRE 816 -48 816 -112 WIRE 912 -48 880 -48 WIRE 176 -16 176 -208 WIRE 176 -16 -48 -16 WIRE 176 32 176 -16 WIRE 192 32 176 32 WIRE 288 32 288 0 WIRE 288 32 192 32 WIRE 336 32 288 32 WIRE 384 32 336 32 WIRE 400 32 400 16 WIRE 400 32 384 32 WIRE 736 32 400 32 WIRE 192 64 192 32 WIRE 288 64 288 32 WIRE 336 64 336 32 WIRE 816 64 816 32 WIRE 112 160 112 -272 WIRE 192 160 192 144 WIRE 192 160 112 160 WIRE 288 160 288 128 WIRE 288 160 192 160 WIRE 336 160 336 128 WIRE 336 160 288 160 WIRE 528 160 528 -80 WIRE 528 160 336 160 WIRE 576 160 528 160 WIRE 736 160 736 32 WIRE 816 160 816 128 WIRE 816 160 736 160 WIRE 880 160 880 -48 WIRE 880 160 816 160 WIRE 384 224 384 32 WIRE 432 224 432 -128 WIRE 608 224 432 224 WIRE 688 224 688 -112 WIRE 432 272 432 224 WIRE 384 352 384 288 WIRE 432 352 384 352 FLAG 576 160 -VEE FLAG 576 -352 VCC FLAG 432 -208 0 FLAG 912 -48 Output_Ground FLAG 912 -112 Output SYMBOL cap 368 224 R0 WINDOW 0 -49 -4 Left 2 WINDOW 3 -61 36 Left 2 SYMATTR InstName C8

SYMBOL cap 800 64 R0 SYMATTR InstName C7

SYMBOL cap 320 64 R0 SYMATTR InstName C4

SYMBOL cap 240 -320 R0 SYMATTR InstName C3

SYMBOL cap 320 -80 R270 SYMATTR InstName C6

SYMBOL res 704 208 R90 WINDOW 0 40 53 Left 2 WINDOW 3 38 23 Left 2 SYMATTR InstName R1 SYMATTR Value 20K SYMBOL res 272 -96 R0 SYMATTR InstName R4 SYMATTR Value 1E6 SYMBOL res 800 -64 R0 SYMATTR InstName R5 SYMATTR Value 1 SYMBOL res 416 256 R0 SYMATTR InstName R2 SYMATTR Value 3.9K SYMBOL res 384 -80 R0 WINDOW 0 -21 47 Left 2 SYMATTR InstName R3 SYMATTR Value 22K SYMBOL cap 304 -320 R0 SYMATTR InstName C1

SYMBOL cap 304 128 R180 WINDOW 0 24 61 Left 2 WINDOW 3 19 6 Left 2 SYMATTR InstName C2

SYMBOL res 176 -336 R0 SYMATTR InstName R6 SYMATTR Value 1K SYMBOL res 176 48 R0 WINDOW 0 -37 26 Left 2 WINDOW 3 -37 57 Left 2 SYMATTR InstName R7 SYMATTR Value 1K SYMBOL voltage 112 -368 R0 WINDOW 0 -77 2 Left 2 WINDOW 3 -64 68 Left 2 WINDOW 123 0 0 Left 2 WINDOW 39 -109 32 Left 2 SYMATTR InstName Power SYMATTR Value 27 SYMATTR SpiceLine Rser=2 SYMBOL voltage -48 -112 R0 WINDOW 0 -17 -11 Left 2 WINDOW 3 -102 126 Left 2 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName Input SYMATTR Value SINE(0 2 100 0.5 0 0 20) SYMBOL Opamps\\UniversalOpamp2 528 -112 R0 SYMATTR InstName U1 SYMATTR SpiceLine ilimit=1 rail=0 Vos=0 phimargin=45 TEXT 48 408 Left 2 !.tran 1 TEXT 56 376 Left 2 !.lib opamp.sub

As I realised last night, it's certainly true that once one amplifier has pulled the ground low, the other amplifiers are forced into the same state, so I cannot determine the culprit after the event. It appears that the only way I can discover whether there's a particular amplifier that's always creating this situation (when it occurs), is to remove each from the circuit in turn, and see whether the problem recurs. That's going to take a while, if I even bother.

Sylvia

In this, which is a rehashed version of the one I used, I've omitted the load, which in my case is an 8 ohm speaker. Also note that the power supply series resistance is unrealistic, and should be reduced.

Sylvia.

Have you run the simulation with an 8 ohm load?

Yes. It works OK provided the input isn't too high, for example 1 volt peak to peak. The ground does get pushed up and down relative to Vcc and

-Vee, so the amplitude of the output is constrained by the need not to approach the power rails given the skewed ground, but otherwise it works fine. This is supported by the fact that I've been using these amplifiers, and when they don't glitch on start-up they sound fine. Given that I never destroyed my ears by listening to disco music, I haven't been tempted to run them at levels that the simulation predicts would cause issues.

However, the sensitivity to startup transients is undesirable, and the simple fix is to connect the power supply ground (approximately midway between the rails) to the speaker output ground. Any ground loops that result do not contain any part of the input ground, so shouldn't be a problem, something that was born out when I tried it.

Why are you going through all this bullshit ? Use a chip that works or build it out of like eight transistors yourself. Fuck, I'll design to you for free.

On Sun, 18 Sep 2016 03:54:16 -0500 John S wrote in Message id: :

Hey John, by any chance were you the John S. that ended up with the HP

No, Sir.

The chip works fine.

Sylvia.

I think it probably does, too. Your simulation shows me, however, that you must reduce the input to .5V peak to keep distortion down. Of course, I don't know your particular thresholds/desires for distortion and amplitude.

That's only because I'd omitted the ground connection from the power supply. As other's noted, the 100uF bypass capacitors are too small to keep the ground line steady at lower frequencies.

Note that the simulation is not using a model of the specific IC, but is treating it as an op-amp with large output voltage and current capability.

Sylvia.

Take jurb up on his offer... ought to be hilarious to watch >:-} ...Jim Thompson

Do the speakers have leads? If so then disconnect each (-) lead from the pseudo-GND and put a 2200uF capacitor between it and the negative power supply plane- get the speaker ckts off the pseudo-GND IOW. That's a fairly minimal fix.

Oh, okay. Exceptions noted. I had hoped for a realistic simulation. You seem to have it solved anyway. Very good. Cheers.

You really think I can't do it ?

I think the simulation's probably quite realistic, since the amplifier is behaving as an op-amp, with a 90dB open-loop gain at DC, and a gain-bandwidth product of 5.5MHz. At audio frequencies and low gains, the relationship between the output and the input is almost entirely determined by the feedback network.

The only major thing not simulated is how close to the power rails the inputs can go.

Sylvia.

And I expect how low each rail can go before the amp drops out or locks up.

NT