I’m going to guess you are contemplating removal of C3 and C4 in the OP schematic of the link you provided in the post above. And then you would use a DC servo to drive ouput offset to 0 ?
This shouldn’t be too onerous. But I will remain agnostic on whether the two-path scheme will control bias as you hope.
I think the servo you proposed in:
https://www.diyaudio.com/community/...ant-a-relentless-analysis.419418/post-7843243 has appropriate configuration. Add a resistor in series with C7 to provide a zero to provide possible compensation if you wish to retain C3,C4. Shunt this added resistor with a small cap (eg 22pF) to ensure local stability at U1.
This shouldn’t be too onerous. But I will remain agnostic on whether the two-path scheme will control bias as you hope.
I think the servo you proposed in:
https://www.diyaudio.com/community/...ant-a-relentless-analysis.419418/post-7843243 has appropriate configuration. Add a resistor in series with C7 to provide a zero to provide possible compensation if you wish to retain C3,C4. Shunt this added resistor with a small cap (eg 22pF) to ensure local stability at U1.
Doesn't matter if it is BJT of Fet differential.
You just set the gain of the amplifier with normal everyday AC coupling capacitors.
Passing additional DC through a amplifier that is biased with DC and needs to have stable DC for stability
is just a plain stupid audiophile myth. The capacitors don't ruin the sound
They block DC from input and feedback path. So it doesn't disrupt the designed DC
for normal operating points.
The so called extra pole of the high pass filter will be at the same old 1 to 2 Hz
And the compensation at high frequency will be the same old 500 kHz
And the input Radio Frequency filter that so many people swear to the world is for stability.
Is not. They usually set it to 150 kHz to 250 kHz which finally fixes the phase margin they were unable
to correct at 500 kHz. Because wild guessing is the usual design approach
You just set the gain of the amplifier with normal everyday AC coupling capacitors.
Passing additional DC through a amplifier that is biased with DC and needs to have stable DC for stability
is just a plain stupid audiophile myth. The capacitors don't ruin the sound
They block DC from input and feedback path. So it doesn't disrupt the designed DC
for normal operating points.
The so called extra pole of the high pass filter will be at the same old 1 to 2 Hz
And the compensation at high frequency will be the same old 500 kHz
And the input Radio Frequency filter that so many people swear to the world is for stability.
Is not. They usually set it to 150 kHz to 250 kHz which finally fixes the phase margin they were unable
to correct at 500 kHz. Because wild guessing is the usual design approach
Last edited:
That depends. Two poles within a negative feedback loop without any other frequency compensation is a recipe for oscillation. But if one of those poles is outside the feedback loop (C1, R3 for example) there's no issue.
Tom
What RF filter is that?
RFI/EMI filters are placed on the input to prevent RF from getting into the amp. As you point out they have no effect on stability. That's because they're outside the feedback loop.
Tom
RF filters at the input can also have an effect on stability. See https://www.diyaudio.com/community/...split-from-opa1656-thread.377331/post-6785883 and https://www.diyaudio.com/community/...split-from-opa1656-thread.377331/post-6790995 for an example.
There would be no effect if the amplifier had an infinite open-loop input impedance from 0 Hz up to the highest possible frequency of oscillation, but usually they don't.
There would be no effect if the amplifier had an infinite open-loop input impedance from 0 Hz up to the highest possible frequency of oscillation, but usually they don't.
Correct hence my point, it is udder basic stuff.
AC amplifier has AC coupling.
You dont have to re invent the wheel or worry about extra poles.
AC amplifier has AC coupling.
You dont have to re invent the wheel or worry about extra poles.
In the thread I linked to, an RC network from the positive op-amp input to ground was used to make the amplifier stable. That seems to contradict what was written in posts #22 and #24.
Because it is all you wanna do is contradict everything.
It Is actually unstable for many reasons.
If I wasted my time to fix it, then remove the filter you claim is needed for stability. It would be stable without it.
So again I dont care about designs based on wild guessing. Which makes the exact point in post #22
RF filter removes RF and should have been there anyways. Being a basic opamp circuit with very high gain
for a phono cartridge. Completely in another world regardless to a lower gain power amp.
In actual operation it could very well still be unstable. Even with the filter because the actual issue wasn't fixed.
It Is actually unstable for many reasons.
If I wasted my time to fix it, then remove the filter you claim is needed for stability. It would be stable without it.
So again I dont care about designs based on wild guessing. Which makes the exact point in post #22
RF filter removes RF and should have been there anyways. Being a basic opamp circuit with very high gain
for a phono cartridge. Completely in another world regardless to a lower gain power amp.
In actual operation it could very well still be unstable. Even with the filter because the actual issue wasn't fixed.
Last edited:
An RC network from IN+ to ground does nothing for loop stability. Unless it somehow miraculously affected the loop gain through coupling between inputs. I suppose that's a very, very remote possibility.
If you're curious to learn more about stability analysis of opamp circuits, I highly recommend reading Ch. 8 in Sergio Franco, "Design With Operational Amplifiers and Analog Integrated Circuits". There are many other sources, but Franco explains well in plain language without drowning you in equations.
Tom
Marcel, as always an interesting analysis. Thank you for posting.
In the case of high input capacitance opamps - probably mosfet which you allude to in your analysis - would it not be easier to look at this analysis by considering the inverting input in conjunction with the feedback network as a bootstrapping arrangement wrt the non inverting input? For low Z sources the bootstrapping effect would be minimal, but with high Z sources (maybe for example a MM cart at HF?) it could be significant.
In the case of high input capacitance opamps - probably mosfet which you allude to in your analysis - would it not be easier to look at this analysis by considering the inverting input in conjunction with the feedback network as a bootstrapping arrangement wrt the non inverting input? For low Z sources the bootstrapping effect would be minimal, but with high Z sources (maybe for example a MM cart at HF?) it could be significant.
I appreciate your explanation, but you still weren't dealing with loop instability.
In your writeup you state that you suspected that the phono cable formed a quarter-wave resonator that acted as an oscillator. That's outside of the feedback loop, so not related to loop stability. You also didn't show anything to support your suspicion other than the notion that an RC snubber helped. Your scope pictures in the posts linked to above show a lot of switch bounce but no oscillation, so I'm curious why you'd bark up that tree.
Another possibility for the loud bang could be that the input stage had some DC offset that got nulled out by the mute switch. It wouldn't take much offset to create a loud bang given the 50-60 dB gain of a typical phono stage at LF.
It could also be that the mute switch is a double-throw switch that connects the phono cable to either the cartridge output or to ground. While it switches there'll be a brief moment where the input to your phono stage is floating.
The source impedance is not included in the loop gain. I agree that Zs reduces the signal at the input to the opamp, so V(a,b) in your writeup, but that has nothing to do with the loop gain. Also, the input impedance of an opamp is extremely high. For the OPA1656 the input impedance is 100 MΩ || 9.1 pF (differential) and 6 TΩ || 1.9 pF. Yes. Tera-ohm. 10^12 ohm. So the source impedance would have to be extremely high for it to have even a minuscule impact on the differential voltage experienced by the opamp.
You can see that pretty easily in your math:
Zin,ol >> Zs makes the last term equal to 1, i.e., the source impedance has no effect on the loop gain, even if it wasn't already excluded by definition.
I'm also not sure where you get ZLt from. Do you have a reference for this or is it your work? I mean... I can see it in your writeup just fine, but I'm not sure where it came from.
The reason capacitive loading of an opamp tends to lead to instability is that the load capacitance forms a pole with the output impedance of the opamp. That adds extra phase change in the feedback loop, which causes instability.
Tom
In your writeup you state that you suspected that the phono cable formed a quarter-wave resonator that acted as an oscillator. That's outside of the feedback loop, so not related to loop stability. You also didn't show anything to support your suspicion other than the notion that an RC snubber helped. Your scope pictures in the posts linked to above show a lot of switch bounce but no oscillation, so I'm curious why you'd bark up that tree.
Another possibility for the loud bang could be that the input stage had some DC offset that got nulled out by the mute switch. It wouldn't take much offset to create a loud bang given the 50-60 dB gain of a typical phono stage at LF.
It could also be that the mute switch is a double-throw switch that connects the phono cable to either the cartridge output or to ground. While it switches there'll be a brief moment where the input to your phono stage is floating.
The source impedance is not included in the loop gain. I agree that Zs reduces the signal at the input to the opamp, so V(a,b) in your writeup, but that has nothing to do with the loop gain. Also, the input impedance of an opamp is extremely high. For the OPA1656 the input impedance is 100 MΩ || 9.1 pF (differential) and 6 TΩ || 1.9 pF. Yes. Tera-ohm. 10^12 ohm. So the source impedance would have to be extremely high for it to have even a minuscule impact on the differential voltage experienced by the opamp.
You can see that pretty easily in your math:
Zin,ol >> Zs makes the last term equal to 1, i.e., the source impedance has no effect on the loop gain, even if it wasn't already excluded by definition.
I'm also not sure where you get ZLt from. Do you have a reference for this or is it your work? I mean... I can see it in your writeup just fine, but I'm not sure where it came from.
The reason capacitive loading of an opamp tends to lead to instability is that the load capacitance forms a pole with the output impedance of the opamp. That adds extra phase change in the feedback loop, which causes instability.
Tom
