Mixer inputs, just one cap gone bad - in all 16 channels?

[anotherjim]

I got an Alto mix to fix. Crackles galore but all is not right with some of the caps. Some bloated, some leaked gunk...
Here's a snippet of my tracing (can't find any service sheet).


It's the 1000uF in the gain control. Rated 10v and an ordinary polarised alu' electro.
I'm not 100% I've traced it right, but it seems like it could work like that.
The only visibly bad caps are this one which all the mic channels use.
Even the channels that have caps that look ok have crackle with the gain controls.

Clearly, those caps need replacing, but can anyone think of a fault that could damage just these caps?
Can it just be that one value/type of cap is from a bad batch with a faulty electrolyte? The mixer was made around 2018 if I read the PCB codes right. Weirdly, the worse looking caps are the first 8 or so channels starting from channel 1 and the farthest away from the power supply end.

BTW, the opamps are really RC4580 - forgot to change the part number. Also didn't note the opamp supply is +/-15v.

[ElectricDruid]

I've seen a few circuits a bit like that, but they mostly had one input to +ve input, and other input to -ve input on the other op-amp, not both to the same polarity like you've drawn. Also I'd expect to see something to set the DC bias point on those +ve inputs. This kind of thing:





One thought that occurs is that if the most-damaged channels are 1-to-8, that suggests they were damaged by being used more, since you'd typically start plugging stuff in at channel one and work across.

[antonis]

Clearly, those caps need replacing, but can anyone think of a fault that could damage just these caps?

Higher voltage than the one caps are rated, perhaps..
(some kind of op-amp malfunction resulting into (+) or (-) 15V applied across those caps or a couple of reverse polarity VDC..)

[anotherjim]

Thanks, guys.
I did find one Alto mixer scheme from which I've cut one input channel...

This has the same cap + pot between phases but how it's used is quite different. My one doesn't have those BJT (cascode?) stages and it's added overvoltage clamping diode pairs to protect the opamp inputs.
Pretty sure the gain control is between the -inputs as I traced. The feedback resistors are the same at 5k11.
I think it ought to maintain an average ground on the -inputs and there should be no significant voltage across the 1000uF. The reason the cap is so large is to maintain AC negative feedback from the opposite phase when the Gain pot is zeroed?

Yeh, the first channels will get more use, especially as 1 to 8 only include compressors.

[amptramp]

You are going to get reverse bias half the time on the capacitor and electrolytics don't survive being reverse biased for all that long.  The usual cure for this is to have two capacitors back-to-back of twice the value, so if you replace the capacitor with a pair of 2200 µF caps (nearest standard size) connected either cathode to cathode or anode to anode, they should survive and function properly.

It looks like the circuit is designed to go to the maximum gain possible whereas a better idea would be to limit how low the pot resistance can go by having a resistor in series.  This in turn would cut down on the amount of capacitance needed for the electrolytic to maintain low frequency response.  If you get low enough (~10 µF), there are non-polar caps available that are internally a pair of capacitors connected back-to-back. 1000 ohms and 10 µF would give you a cutoff of 15.9 Hz.  A feedback resistor of 10 K would give you a gain of 21 since the gain is that of two non-inverting amplifiers with a 10 K feedback and a 500 ohm input resistor.

[Vivek]

I know nothing about this circuit

But 1000uf only reminds me of power supply caps or loudspeaker caps.

Why such a huge value in the middle of an audio path ? How does one calculate the min cap value at that location ?

[r080]

I know nothing about this circuit

But 1000uf only reminds me of power supply caps or loudspeaker caps.

Why such a huge value in the middle of an audio path ? How does one calculate the min cap value at that location ?

Probably big enough such that the low frequencies are not affected at minimum gain. F=1/(2*pi*R*C). Rearrange for C=1/(2*pi*F*R). Assume there is some resistance in the traces/wire/pot lugs - say 10 Ohms, and that we care about audio down to 20Hz, C=1/(2*pi*20 Hz*10 Ohms) ~ 800uF.

[anotherjim]

Yes, my first reaction was WTF is that 1000uF 10v cap doing in a mixer channel running +/-15v?

Space is tight for the 1000uF caps between the XLR and 1/4" jacks. I was thinking of higher-rated replacements (negligible cost increase) but 16v might be the best I can go to.

I don't like polarized caps in these situations either, but hey, it's lasted 4 years, another 4 and it'll be considered obsolete!

The way it's made, it's going to be very irksome to power up so I can probe any circuits. The only way I could do that is to remove all the knobs and jack fixings so as to lift the top panels off - from a 24-channel 4 bus mixer!

[Keppy]

Based on the trace in your first post, I don't understand why that cap is there. I mean, sure, AC coupling opamps is best practice, but unless the two opamps are biased to different voltages, then any DC offset is unintentional and has a 50% chance of reverse biasing the cap anyway. My first reaction was to think you should just short that cap and see if the channel works appropriately then.

I look forward to hearing why this is a bad idea.

[PRR]

The cap voltage can't be more than the microphone voltage.

Polar Al electros usually do very well here.

They usually DO have the stopper resistor as Ron says. They seem to be pretending the gain can go to infinity, which it can't; it will suck for gain of even 1,000. You do not note the series resistors values, but typical 5k, the pot may go to 33 Ohms usefully, so you want >250uFd for -3dB @ 16Hz, and so 1,000uFd gets half-dB @ 20Hz for specsmanship.

It is also entirely possible that they got a bad batch of 1,000uFd 10V, only. Or that the 1000/10 used elsewhere have steady DC on them.

Or that it was OK until the kids threw BIG levels into the inputs which the cap could not follow. (Has it been used as a fuzzbox?) All 16 channels? Yes, when one input craps-out you move to the next jack. Might not take all night to blow 16 channels, if they don't need other sources.

You can just short the cap but it will scratch and thump as the knob is turned. You can short it with a 22r to 47r resistor and get a defined max gain and minimum impedance, but it may still be wonky when turned. Do what is easy and watch for a deal on a newer board.

[anotherjim]

I've thought of a mode that might damage the caps over the long term. If you switch P48 on, there is going to be a positive pulse in the input lines. This is common-mode, so the cap doesn't worry. However, if you use unbalanced inputs, the cold is grounded and only the hot goes positive. The protection diodes clamp it to 15.7v, the opamp drives positive as high as it may and the 5k feedback attempts to charge the 10v cap to around 14v or so.
Is that a reasonable scenario?

[anotherjim]

Well, I've finished the repairs. On the way, I came across another problem.
3 channels were making very loud random crackles. That was traced to some protection diodes breaking down, although they tested fine with the DMM. Never had that before, but replacing the 4 diodes on the bad channels fixed them. I suppose some electrolyte from the bad caps got under the SMD diodes, but I thought I'd given it a good clean with solvent.