Focusrite mic pre oscillation

[gurkan75]

While recapping a focusrite RED 1 mic preamplifier I noticed the first amplifying stage has a rather large 220pF compensation capacitor between pins 5&8 on the 5534 opamp. All other stages use a more common 22pF value.

In all schematics I can find (this design appears in many of the focusrite preamps) the value of this exact cap should either be 22pF or is not given so out of curiosity I changed it to 22pF. It works and sounds great up until the very last setting on the gain switch, 60dB, where it starts to distort (oscillate?) badly.

The only other variations in schematics I have found are the capacitor to ground in the shunt portion of feedback/gain switch which is either 100uf or 470uf and a very slight variation in the loading resistor of this stage, between 4k7 and 6k1.

Any Idea as to why this revision of the circuit needs more compensation? Poor layout? It has 10nF to ground on both rails right near the opamp.

 

[ccaudle]

It works and sounds great up until the very last setting on the gain switch, 60dB, where it starts to distort (oscillate?) badly.

That is strange, typically op-amps have oscillation problems at low gain, not high gain. Now you have sent me down a rabbit trail of looking up papers on op-amp compensation trying to figure out what might be going on.

 

[JohnRoberts]

There may be some stray feedback path related to physical layout. Have you experimented with >22pF but <220pF?

JR

 

[ccaudle]

There may be some stray feedback path related to physical layout

That occurred to me as a possibility. One way to address excessive capacitance at the inverting input is to add a feedback capacitor. I am not sure if increasing the compensation capacitor has the same effect. It seems that the mechanism would be different, increasing the capacitor in the feedback path adds a phase lead around the entire device to counteract the phase lag, increasing the compensation capacitor reduces the open loop bandwidth so that the phase lag is less of a problem.
I have trouble following the switch contact naming to figure out what the resistor configuration is for the 60dB gain tap, but I suspect it may be related to physical geometry as much or more than schematic visible configuration.

 

[gurkan75]

There may be some stray feedback path related to physical layout. Have you experimented with >22pF but <220pF?

JR

I did try with 2x22pF & 3x22pF, neither was enough to make the last gain position work.

I have trouble following the switch contact naming to figure out what the resistor configuration is for the 60dB gain tap, but I suspect it may be related to physical geometry as much or more than schematic visible configuration.

The gain switch is set up with the first 6 positions on one deck forming an attenuator between transformer and 1st 5534, positions 6-12 is a potential divider for the negative feedback between negative input and ground, on deck two. The series feedback portion in parallel with 220pF, and a 2x6k2 series coupled with a capacitor to ground in the middle. Not sure what this bit is meant to do.

What is puzzling me is that the relatively small change in this potential divider between steps 11 & 12 (series r goes from 6350 to 6406 ohms shunt goes from 112 to 56 ohms) causes the 5534 to go nuts without the big compensation cap.

I have three of these red boxes and they all have the 220pF on all 4 channels. Opamp drives 4k7 in parallel with another 5534 for output and a 5534/5532 for vu meter driver.

The actual switch is 12 position grayhill instead of 11 position Elma on older blue/gray models. The extra position is used at the beginning of rotation for a -6dB setting, there is no regular pad. The switch has two tabs that are connected to audio ground, I presume they are there to support the switch physically to the pcb.

Does anyone know how 220pF vs 22pF affects bandwidth & slew rate for a 5534? Maybe I should just the put the 220pF back and be done with it. I just know that other takes on this circuit seem to work fine with the smaller compensation cap.

Thanks for having a look!

Johan

 

[Brian Roth]

This may be off the wall and have no bearing on the situation...

But, I have seen capacitor markings like 220 which means no zero after the "22" and thus 22pF.

221 would be 220 pF..."22" followed by one zero.

222 would be 2200 pF...."22" followed by two zeros.

I've seen this on dipped micas and probably ceramic discs. I think those caps were "mil spec".

"221" and "222" are pretty obvious markings. "220" is ambiguous.

Just a thought!

Bri

 

[gurkan75]

"221" and "222" are pretty obvious markings. "220" is ambiguous.

A good thought! It was the physical size that caught my attention so I did desolder and measure, just to be sure. Sure seems to be a 220pF, measures a bit higher.

 

[JohnRoberts]

The obvious question is how does it work? Is frequency response and linearity (distortion) good? Then no problem.

JR

 

[gurkan75]

For reference, I found an alternative fix for the instability at highest gain. I experimented with a higher resistance r107, between gain ”pot” and ground. I used a trimmer in parallell and found that at a sharply defined point, 64 ohms instead of 56, the oscillations disappear. I used 82ohms.

It now seems stable even without compensation, but used 22pF as in schematic.

I can measure no loss of gain.

Interestingly, rev 1 board is fine with no modification. Rev 2 & 3 have almost the same component layout, only the ground traces have much larger surface area and they seem to need either the 220pF cc or the larger resistor.

 

[user 37518]

The 22pF cap is required for low gains in the 5534, 22pF is the minimum value but I've definitely seen higher values, MCI used 47pF. 220pF seems excessive though, but maybe it was used as a brute force afterthought by the designers to stop OpAmp oscillation. Increasing the size of the cap will make the Opamp more stable at the expense of reduced slew rate. The obvious question is, what happens if you use a 220pF cap?

 

[gurkan75]

The obvious question is, what happens if you use a 220pF cap?

Nothing that I am able to measure. I have not been able to find a lot of data but it seems distortion should increase but stays ridiculously low. It seems it would affect rise time and, as you mentioned, slew rate the most.

 

[user 37518]

Nothing that I am able to measure. I have not been able to find a lot of data but it seems distortion should increase but stays ridiculously low. It seems it would affect rise time and, as you mentioned, slew rate the most.

If the voltage level at the output is low, a small SR isn't necessarily bad, in this case i think I would trust the Focusrite designers and use the 220pF cap.