Capacitors in Parallel with a Vref resistor... What The Heck?

[Paul Marossy]

I have a schematic with a small value capacitor in parallel with a vref resistor (three total). They are on a dual opamp. One is on each inverting input, (attached to the feedback loop, drive control and clipping diodes) and one is located between the output of first opamp and non-inverting input of second opamp.

In my 17 or so years of building things and looking at lots of schematics I've never run across that. What would be the purpose of such an arrangement? Does this prevent some sort of oscillation from occurring or something like that?

[ElectricDruid]

Sorry Paul, I'm not seeing it from the description. Can you do us a little sketch or something?

T.

[Paul Marossy]

Sorry Paul, I'm not seeing it from the description. Can you do us a little sketch or something?

T.

https://i.postimg.cc/zB6S7XHX/Weird-Vref.jpg

[ElectricDruid]

Those are all there to limit the frequency response.

The caps down to Vref from the -ve input roll-off the low-end in much the same way as the ones across the diodes roll-off the highs. In fact, doesn't the Tubescreamer use a cap like this to limit the bottom end to 720Hz or so?

The one on the input is providing a lowpass function in association with that input resistor. The effect of the lowpass will be limited by the other resistor (it's a "shelving" response, not a solid lowpass), but these things all help shape the response in some way.

What's weird about that is that the "bias resistor" from +ve input to Vref is the wrong side of the electrolytic cap. I'd have expected that the other way around.

HTH,
Tom

[Paul Marossy]

Those are all there to limit the frequency response.

The caps down to Vref from the -ve input roll-off the low-end in much the same way as the ones across the diodes roll-off the highs. In fact, doesn't the Tubescreamer use a cap like this to limit the bottom end to 720Hz or so?

The one on the input is providing a lowpass function in association with that input resistor. The effect of the lowpass will be limited by the other resistor (it's a "shelving" response, not a solid lowpass), but these things all help shape the response in some way.

What's weird about that is that the "bias resistor" from +ve input to Vref is the wrong side of the electrolytic cap. I'd have expected that the other way around.

HTH,
Tom

I probably showed that one resistor in the wrong place. I have something like this on my bread board but the weird thing is that if I take any of those vref resistors after the input out of the circuit it stops working. I guess if I were to do that ALL vref would need to be removed?

I'm asking all this because I have a biasing problem that I am trying to figure out.

[Rob Strand]

I've never run across that. What would be the purpose of such an arrangement? Does this prevent some sort of oscillation from occurring or something like that?

The values of the caps might help guess the motive.   In this configuration it looks weird.

My first guess is there was some sort of problem and that solution somehow come about.  I wouldn't go thinking everything you see in a device is correct.  Sometimes it comes from screwed-up thinking which just happened to solve a particular problem.  eg.  RF interference or reducing mobile phone pips.

M next guess is they are just maxing out the opamp gain.

It was common to add caps across R5 this type of circuit to stop oscillation, particularly when R5 was large,

but that doesn't quite fit the pattern.

[Paul Marossy]

My first guess is there was some sort of problem and that solution somehow come about.  I wouldn't go thinking everything you see in a device is correct.  Sometimes it comes from screwed-up thinking which just happened to solve a particular problem. 

That's what I was thinking. Some people jump right to "it's a filter", but in this case I am not so sure about that. All those vref resistors have to be in place for the circuit to work. It doesn't seem to care about the caps, I think. There must be some reason for why they thought they were necessary though. 

EDIT: Caps on - inputs are .047uF and the on second stage + input is 0.22uF

[j_flanders]

In my 17 or so years of building things and looking at lots of schematics I've never run across that.

Really? There must be hundreds of threads about this. They all boil down to: do I connect an ac signal (be it through diodes, or a filter or whatever)  to ground or vref? And is there a difference?
Here's what R.G says about it:
https://www.diystompboxes.com/smfforum/index.php?topic=110242.0

One subtle issue with returning signal "grounding" to Vref is that Vref is connected to most of the *inputs* of circuits on the board. So if you're "grounding" anything with any significant current involved to Vref, that current times the impedance of the Vref point is sent to the other inputs. This may or may not be an issue, as the details of the other circuits, including how much gain they have, determines the effect.

Digital Larry sums it up in one sentence:

My engineering education left me with a very strong sense that for signals, any power rail is the same as ground because all power supplies have zero impedance, right?

https://www.google.be/search?newwindow=1&safe=off&ei=hbc3XIOaApHWwAL_6CM&q=vref+vs+ground+site%3Adiystompboxes.com&oq=vref+vs+ground+site%3Adiystompboxes.com&gs_l=psy-ab.3...40475.44019..45221...0.0..0.170.987.12j2......0....1..gws-wiz.pG5ENWCbnRY

I wouldn't call all those 'vref resistors' though
Two of them are simply Rg as in this schematic:

But connected to vref instead of ground (which is why there's no need for a dc blocking series cap)

[Rob Strand]

That's what I was thinking. Some people jump right to "it's a filter", but in this case I am not so sure about that. All those vref resistors have to be in place for the circuit to work. It doesn't seem to care about the caps, I think. There must be some reason for why they thought they were necessary though. 

EDIT: Caps on - inputs are .047uF and the on second stage + input is 0.22uF

With such large values it is definitely intentional.  It's like the 47nF on a tube screamer but without the series resistor.   At high freqs the gain keeps going up until it maxes out the opamp gain.    If the gain keeps going up beyond 20kHz it won't add to the tone and is likely to just cause trouble with RF and oscillations.

[R.G.]

One thing it for certain does - it helps couple any noise on the Vref into the amplifier. That way, if any hum/noise/RFI gets in on Vref, everybody gets it. And if any of the amplifiers oscillate, they can all sing the same song. 

Go ahead - ask me how I know this.     

[Paul Marossy]

That's what I was thinking. Some people jump right to "it's a filter", but in this case I am not so sure about that. All those vref resistors have to be in place for the circuit to work. It doesn't seem to care about the caps, I think. There must be some reason for why they thought they were necessary though. 

EDIT: Caps on - inputs are .047uF and the on second stage + input is 0.22uF

With such large values it is definitely intentional.  It's like the 47nF on a tube screamer but without the series resistor.   At high freqs the gain keeps going up until it maxes out the opamp gain.    If the gain keeps going up beyond 20kHz it won't add to the tone and is likely to just cause trouble with RF and oscillations.

Now that would make sense, especially knowing the people involved with the design of this thing.

[Paul Marossy]

One thing it for certain does - it helps couple any noise on the Vref into the amplifier. That way, if any hum/noise/RFI gets in on Vref, everybody gets it. And if any of the amplifiers oscillate, they can all sing the same song. 

Go ahead - ask me how I know this. 

That's not good. Can it also go into an inaudible oscillation and just stop working? Or do weird things with the opamp bias? That's the kind of problems I am having with this circuit, can't figure out why though. But it definitely seems to be a biasing issue of some sort.

[Paul Marossy]

I wouldn't call all those 'vref resistors' though
Two of them are simply Rg as in this schematic:

But connected to vref instead of ground (which is why there's no need for a dc blocking series cap)

I understand your point. I always associate voltage dividers that give half the supply voltage with biasing. In this case I still see it that way as well since there is a vref connection on the input of the circuit. So that must be biasing the input, obviously. The other ones are the odd ones that I don't recall seeing on any schematics, or maybe I just never noticed until I reverse engineered this thing I'm working on.

[jubal81]

I get the impression that it might have been a conversion from bipolar supply to single supply and the person who drew it up just replaced the GND with v/2.

[Paul Marossy]

I get the impression that it might have been a conversion from bipolar supply to single supply and the person who drew it up just replaced the GND with v/2.

Your guess is as good as mine!

[ElectricDruid]

Looking at this thing again, I don't think there's anything weird about it, really.

https://postimg.cc/YGNWq7xJ

If you think about the DC conditions, you can ignore the caps, in which case it all looks pretty normal. The second op-amp has its own DC biasing (I'm assuming that resistor to Vref should be the other side of the electro cap) which is unnecessary, since the signal is already correctly biased from the first op-amp, but you often see that in beginner's circuits where people have bolted two or three things together and not really looked at the whole picture.

The schematic hasn't got values on it, but I don't see any reason why (with reasonable values) it shouldn't work, aside from that one resistor to the wrong side of the electro cap.

Am I missing something? Can someone else point out what I'm not seeing here?

HTH,
Tom

[anotherjim]

Without seeing the cap values. The gain has a huge treble boost with bypass on Rin but countered by the cap across Rf. However, as Rf can be reduced to zero, then gain can be reduced toward unity anyway. As Rf increases, the low pass with the cap has a falling cut-off frequency which might either overlap & counter the treble boost of the Rin cap or allow a mid boost to exist.

[Rob Strand]

If you think about the DC conditions, you can ignore the caps, in which case it all looks pretty normal.

The schematic hasn't got values on it, but I don't see any reason why (with reasonable values) it shouldn't work, aside from that one resistor to the wrong side of the electro cap.

Am I missing something? Can someone else point out what I'm not seeing here?

There's two things which aren't normal:

Paul said the caps were 47n and 220nF.    So as far as the circuit working it's like tube-screamer with the 4k7 resistor shorted.  So the gain at high frequencies is really high.

As you said, the DC connection of those resistors from the opamp -input will work.   In a tube-screamer type circuit the gain tapers to unity at DC.  At low *frequencies* the clipper is disabled.  With the DC connected Vref resistors in place the DC gain is no longer unity and the clipper is never really disabled, it's only disabled by a low input level.   I can only guess the motive is to reduce the amount of "clean" signal that leaks through the clipper.

As you vary the gain on the first stage the DC offset of the opamp get amplified which is undesirable.  It can promote pot scratchyness.   The Rat circuit has a puts a cap in series with both gain resistors which stops that. 

[aion]

A more specific question about the parallel resistor and capacitor at the (+) input of the op amp, as shown:



I understand R4 is providing a bias voltage to the opamp. But is the C17 capacitor interacting with the resistor at all in the audio spectrum, or is it independent? Based on what R.G. said about potentially coupling Vref noise from elsewhere, I'm wondering if it'd be better for the capacitor to go to ground instead, if it's just acting as a lowpass filter. But if R4 is creating a shelving response along with C17 as Tom said then that perhaps wouldn't be a good idea and it should stay as-is.

[Rob Strand]

I understand R4 is providing a bias voltage to the opamp.

That part of the circuit looks 100% normal to me.  (Yes, R4 for bias.)

It's just a low pass filter.  The effective resistance is R3 in parallel with R4 which is quite close to R3.  The presence of R4 causes a small decrease in the gain   (R4 / (R3 + R4)).

The tube screamer is essentially R3=1k and R4=10k with no C5.  C5 separates the DC on each side of the circuit.