OPA1664 oddness

[MrStab]

Here's a weird one! Does anyone have any idea why an OPA1664 op-amp would outright die in this active volume application, where TL074 and others handle it fine? Bit frustrating given the limited selection of TSSOP-14 quad op-amps.




This has taken place over 2 board revisions, with the 200R being added in a futile effort to prevent it. Can only guess it has some power-up requirement at this point. Maybe something buried deep in the datasheet which I can't figure out.

The other half of the op-amp, which handles different (also TOTALLY NORMAL!) functions, still works.

Non-inverting input of U9a gets pulled up to ~7.5V (despite 100k to 4.5V bias), which is presumably then inverted by U9b to 0V. Anarchistically shorting U9a input to output for a brief moment (IDGAF - it's dead anyway) can produce either 4.5V or 0V on U9a output pins, which come out U9b as 1.2V and 7.5V, respectively. Almost as though U9a input is floating.

Should I get the scope out? It's not a pressing issue, just perplexing.

[MrStab]

It actually seems to work fine. Bias is closer to 4.5V with vol pot in middle. I'm overlooking something. SPICE and common sense tell me DC level should be 4.5V! But then I don't have much of the latter in my life.

Headroom is ofc rubbish as a result.

No particular strangeness on scope when powering up.

Edit: Okay, bypassing the probably-superfluous 22u (as preceding output is biased at 4.5V anyway) has resolved things, albeit with a slight offset. Interesting.
Lol sorry, I'm always doing this. But it does help to A. Think out loud and B. Motivate myself with the panic of impending public shame.
Comments still welcome on disparity between op-amp types' behaviour, nonetheless.

[diffeq]

Headroom is ofc rubbish as a result.

Headroom is the last of an issue here - you are feeding a gain stage with its own output.

so how about doing this:

[ElectricDruid]

+1 agree. The problem is the circuit, not the op-amp. What's amazing is that *other amps survive*, not that this one dies.

[MrStab]

Isn't this a pretty commonplace Baxandall active volume control? I'm not sure I see a problem. I'm feeding a gain stage with an inverted copy of its own input.

Things were crapping out in certain scenarios at max volume (ie. zero resistance between lugs 2 & 3), but resolved by placing a 200R resistor in place of the 22u cap. No issues as far as I can tell.



U3 output in blue, U2 output in yellow.

I don't rush my simulation layouts! How dare you!

[ElectricDruid]

Isn't this a pretty commonplace Baxandall active volume control?

Yes, it is, my apologies. I had to look it up, but it's in Small Signal Audio Design. Page 344 in my 3rd Ed. Not so commonplace that I remember seeing it before, apparently. Thanks for the pointer.

So...back to the beginning...what's going on?

The DC conditions look ok. Both amps are biased ok. About the only thing Self says that might be relevant here is that its input impedance drops significantly. His example goes down to 1.87K or something - ok for a 5532, but a lot of op-amps would baulk at that. You won't see that in a sim. That might also explain why swapping the 22u for 200R helps - you're increasing the effective input impedance that the driving circuit sees and reducing the problem, perhaps to a level it can manage.

HTH

[MrStab]

So I suspect that in the first revision, there were two problems, one of which killed the op-amp (specifically an analogue switch IC input connected to the output, perhaps effectively shorting the output on power-up). In this revision, one of those problems was reinstated (the 22u cap), but this didn't kill the op-amp. Just caused biasing issues. Seems I was mistaken and had assumed issue A had returned.

Very helpful observation re. the input impedance - thanks! Just an odd quirk where a preceding gain setting caused latch-up. Makes sense, and may have prevented me from going down a supply voltage rabbit hole. Testing at 9V currently, expected final voltage will be 18-24VDC.

The rest of the circuit is quite dense and complicated in places, so of course it's the one thing you'd think would be simpler that goes awry!