Does cap voltage affect tone?

[cerrem]

The ESR in a cap makes a time constant against the value of the Cap....This is well known term call ESR ZERO .... because it puts a ZERO in the AC current loop transfer function... Do the math and figure it to see for yourself..
The AC current loop would be through tube...through plate resistor...into DC bus...through de-coupling cap..to ground..then back through the cathode....
This AC current loop will have a -3dB point at the frequency the impedance of the cap = the ESR ... This is 45 degrees of phase shift at this frequency...
Even if this frequency POLE is above audio lest say 60kHz.... the phase shift at upper audio range is still significant and push the highs behind LAG the lower frequencies by 14 degrees at 10kHz... this is audible....
Speakers do have audible range above the data sheet.... the speaker turns into a "tweeter" at higher frequencies... The voice coil, the wire and glue will resonant ... There is a frequency the speaker will no longer move...called the turn-over frequency...due to mechanical compliance limit....the speaker cone doesn't travel anymore,,instead the voice coil resonates and acts like a tweeter and square wave from a cranked and can sound harsh or smooth depending on the materials of the speaker...aluminum voice coil former vs paper would be an example...
As for the 3577... It is an industry standard for many years and is solid... I bought 2 of them broken ...for $300 each.... I repaired one ...the second unit is sitting in storage have no time to fix it...Your welcome to it...I ll even supply the service manual...

[Malcolm Irving]

… the highs behind LAG the lower frequencies by 14 degrees at 10kHz... this is audible....
...

I don't think it would be audible to me, but maybe that's due to age and playing the guitar too loud! I think there is still a debate going on about the extent to which phase shift is audible. I suspect that where it is 'audible' it is probably only in the 'direction finding' sense rather than 'musical timbre' sense.

[Enzo]

I think it also relevant to mention we are discussing guitar amps and not hifi amps. What goes on at 10kHz in a stereo may well be audible, but in a guitar amp? Not so much. No one runs a sparkly tweeter in a guitar amp. Guitar speakers roll off an octave or more below that.

[Malcolm Irving]

The ESR in a cap makes a time constant against the value of the Cap....This is well known term call ESR ZERO .... because it puts a ZERO in the AC current loop transfer function... Do the math and figure it to see for yourself..
The AC current loop would be through tube...through plate resistor...into DC bus...through de-coupling cap..to ground..then back through the cathode....
This AC current loop will have a -3dB point at the frequency the impedance of the cap = the ESR …
...

I’m not sure if that is relevant to the triode circuit in question.

At some low frequency the impedance of the capacitance is high compared to both the cathode resistor (and its own ESR). The impedance between cathode and ground is given by the cathode resistor. The stage is ‘un-bypassed’ at this frequency and has a lower gain.

At some higher frequency the impedance of the capacitance is low compared to the cathode resistor but still higher than the ESR. The impedance between cathode and ground is somewhere between the value of the cathode resistor and the value of the ESR. We could call the stage partially bypassed at this frequency. The gain of the stage is now higher.

At even higher frequency the impedance of the capacitance is low compared to its ESR. The impedance between cathode and ground is the ESR. Gain of the stage is maximum and we could call the stage fully-bypassed at this frequency.

There is a point where the impedance of the capacitance is equal to its ESR, but nothing very drastic happens at that point.

Depending on the values of capacitance, ESR and cathode resistor, the frequencies mentioned above (in practice) are down in the 10Hz to 2kHz region. At frequencies above that it is 'plain sailing' with a fully bypassed gain stage.

[Pedro Vecino]

He's right about the high esr in some models though. especially some of the bullet shaped dry tants that I have used. They are indeed rated in tens of ohms (at least on the Mouser description). I suppose this anomaly doesn't apply to all tant caps, but I thought I'd point this out. I remember being surprised because I thought that by default I would get better esr with the tantalums. Turns out I have to pick through them. Not that it matters a whole lot with a cathode bypass cap. 20 ohms compared to 2 ohms across a 1500 ohm resistor!?! I'd need to hear it to believe there's an audible difference.

In many Mesa amplifiers you can see 100 Ohms resistors in series with tantalum tubular cathode capacitors. I have never worried about understanding what function they have. The reason: sometimes I use that system to reduce gain and I have never used values lower than 390 Ohms. I'm sure that below that value I do not appreciate its effect (at least with my Fender and Gibson signal generators).
But maybe it's related to what you comment.

[Helmholtz]

Out of curiosity I just measured the ESR@1kHz of a number of different E-cap types (Tantalum, Al-LV, Al-HV) with C values of 10, 15, 22µF.
All of them measured around or below 1 Ohm. There is no significant/typical difference at audio frequencies between Tantalum, low-voltage Al-caps and 450V Al-caps regarding ESR.
The ESR tends to decrease with increasing frequency for all types at least up to 100kHz.

Apart from these results I would even consider an ESR of 20 Ohms irrelevant for a cathode bypass cap, as the gain of the triode is determined by the ratio of the total load impedance at the plate to the external cathode impedance (cathode resistor in parallel with bypass cap) in series with the tube's internal cathode impedance. In other words, at high frequencies, in "fully bypassed" mode, the gain is given/limited by the ratio of total plate impedance to the internal cathode impedance + ESR. Now the internal cathode impedance typically is around 1k for a 12AX7 and the effect of the series ESR seems negligible.

[Chuck H]

The AC current loop would be through tube...through plate resistor...into DC bus...through de-coupling cap..to ground..then back through the cathode....
This AC current loop will have a -3dB point at the frequency the impedance of the cap = the ESR ... This is 45 degrees of phase shift at this frequency...
Even if this frequency POLE is above audio lest say 60kHz.... the phase shift at upper audio range is still significant

Ok, really? Just how much signal is left to loop after it passes through the decoupling cap? So it's not as if All the signal riding on the cathode caps ESR (already tiny) is looping. It's whatever is left over after that the decoupling caps ESR doesn't eliminate. So now we're getting REALLY small. And we haven't even accounted for the fact there is further cancellations happening on the DC supply rail by virtue of (most sane designs) having an out of phase gain stage at the same node. Add to that any risk of audible anomalies or instability is usually only significant in higher gain designs that include top end bleeder circuits that will be in this "loop". Then factor in the reality of inter-electrode capacitance in the tube itself taking off a chip at such high frequencies and THEN the guitar speakers limitations. This HF phase affect of the "loop" sounds more like an exercise in theory than a real world issue. I'm not a quantum physicist so I'm really not concerned with such minutia. Or, put another way, there's no way this "loop" effect is significant and I'll bet three beer you can't show significant measured affect in any guitar amp. Not saying it isn't real. Just saying that the actual dilution of the agents responsible for the phenomenon border of homeopathic.

JM2C

[cerrem]

Ok, really? Just how much signal is left to loop after it passes through the decoupling cap? So it's not as if All the signal riding on the cathode caps ESR (already tiny) is looping. It's whatever is left over after that the decoupling caps ESR doesn't eliminate. So now we're getting REALLY small. And we haven't even accounted for the fact there is further cancellations happening on the DC supply rail by virtue of (most sane designs) having an out of phase gain stage at the same node. Add to that any risk of audible anomalies or instability is usually only significant in higher gain designs that include top end bleeder circuits that will be in this "loop". Then factor in the reality of inter-electrode capacitance in the tube itself taking off a chip at such high frequencies and THEN the guitar speakers limitations. This HF phase affect of the "loop" sounds more like an exercise in theory than a real world issue. I'm not a quantum physicist so I'm really not concerned with such minutia. Or, put another way, there's no way this "loop" effect is significant and I'll bet three beer you can't show significant measured affect in any guitar amp. Not saying it isn't real. Just saying that the actual dilution of the agents responsible for the phenomenon border of homeopathic.

JM2C

The AC signal loop is critical.... It's this AC current that develops the AC voltage signal that comes off the plate resistor...
A ZERO in the current loop makes for a POLE in the output signal VOLTAGE .....
I should have made more clear than the ESR of the DC rail filter cap of that stage is most significant for high frequency roll-off...since that is in series with the AC current loop... Any impedance outside of the plate resistor takes away signal voltage from the plate resistor..

Here a real example.... 47uF filter cap for amplifier stage DC rail filter.... lets say it's a old crapy vintage cap with 500 mOhms of ESR....
This will put a -3dB POLE at 6.7kHz .....
Having a Network Analyzer does come in handy...you can sweep the cap impedance in the audio range and actually see there issues that can takes place in the audio band.... yes..if your playing sin waves through your guitar amps you wont see or notice squat.... playing guitar is all about transients and making square waves which makes a whole different scenario...

[cerrem]

Ok, really? Just how much signal is left to loop after it passes through the decoupling cap? So it's not as if All the signal riding on the cathode caps ESR (already tiny) is looping. It's whatever is left over after that the decoupling caps ESR doesn't eliminate. So now we're getting REALLY small. And we haven't even accounted for the fact there is further cancellations happening on the DC supply rail by virtue of (most sane designs) having an out of phase gain stage at the same node. Add to that any risk of audible anomalies or instability is usually only significant in higher gain designs that include top end bleeder circuits that will be in this "loop". Then factor in the reality of inter-electrode capacitance in the tube itself taking off a chip at such high frequencies and THEN the guitar speakers limitations. This HF phase affect of the "loop" sounds more like an exercise in theory than a real world issue. I'm not a quantum physicist so I'm really not concerned with such minutia. Or, put another way, there's no way this "loop" effect is significant and I'll bet three beer you can't show significant measured affect in any guitar amp. Not saying it isn't real. Just saying that the actual dilution of the agents responsible for the phenomenon border of homeopathic.

JM2C

The AC signal loop is critical.... It's this AC current that develops the AC voltage signal that comes off the plate resistor...
A ZERO in the current loop makes for a POLE in the output signal VOLTAGE .....
I should have made more clear than the ESR of the DC rail filter cap of that stage is most significant for high frequency roll-off...since that is in series with the AC current loop... Any impedance outside of the plate resistor takes away signal voltage from the plate resistor..

Here a real example.... 47uF filter cap for amplifier stage DC rail filter.... lets say it's a old crapy vintage cap with 500 mOhms of ESR....
This will put a -3dB POLE at 6.7kHz .....
Having a Network Analyzer does come in handy...you can sweep the cap impedance in the audio range and actually see there issues that can takes place in the audio band.... yes..if your playing sin waves through your guitar amps you wont see or notice squat.... playing guitar is all about transients and making square waves which makes a whole different scenario...

[g1]

500 milli Ohms? That's crappy?

[Malcolm Irving]

...
A ZERO in the current loop makes for a POLE in the output signal VOLTAGE .....
...

'Current loops' and 'signal voltages' do not have poles or zeros.
The things which have poles and/or zeros are transfer functions (in the complex frequency domain).
Could you explain it more carefully? Maybe even show us the maths?

[Malcolm Irving]

....
Here a real example.... 47uF filter cap for amplifier stage DC rail filter.... lets say it's a old crapy vintage cap with 500 mOhms of ESR....
This will put a -3dB POLE at 6.7kHz .....
...

To find the RC time constant associated with a capacitance you have to include all the resistance in the charge/discharge path. The time constant you get by only including the ESR would only apply if the capacitor is short-circuited.

[Malcolm Irving]

The AC signal loop is critical.... It's this AC current that develops the AC voltage signal that comes off the plate resistor...
...

If there is 0.1mA of AC signal current in that loop it would create 10 volts AC signal across a 100k plate resistor and 50 micro-volts of AC signal across a 500 mOhm ESR. How can that be significant?

[Helmholtz]

To find the RC time constant associated with a capacitance you have to include all the resistance in the charge/discharge path. The time constant you get by only including the ESR would only apply if the capacitor is short-circuited.

This short-circuit "corner frequency" is just the frequency at which the absolute voltages across the inner capacitive and resistive components become equal. But these voltage drops are ridiculously small for currents in the mA range and thus will not show in the frequency response of the circuit.
The effects of an increased ESR on frequency and square wave response can be studied by wiring a suitable external resistor in series with the cap - or by simulation of the partial circuit.

An ESR of 0.5Ohm is excellent.

[Helmholtz]

Having a Network Analyzer does come in handy...you can sweep the cap impedance in the audio range and actually see there issues that can takes place in the audio band...

When I worked in SMPS design I had access to one of these great machines:

http://www.testequipmentdepot.com/us...zers/4194a.htm

and I used it extensively to analyze all types of caps, chokes, transformers, filter circuits and even guitar pickups (in the afterhours) for impedance response, resonances, parasitic components etc.

What you get when you plot the frequency response of the impedance of an E-cap up to high enough frequencies looks like this:

ecapimp.pdf

There is always one more or less pronounced Z-minimum around the frequency where the reactances of C and ESL (equivalent series inductance or parasitic inductance) cancel, i.e. the caps self-resonance. This frequency is typically above the audio range. The minimum value of Z corresponds to the ESR.
I have yet to see an E-cap up to 100µF showing any anomalies in the audio range.