Bypass cap for cathode bias push pull output tubes

[sbirkenstock]

Hi again,

I tried to change the bypass cap for a cathode biased 6L6 push pull output stage.
It did not make any difference in sound to me.
So I measure with an oscilloscope and figured out that the voltage of the amp at the cathode stays very even.
(I was applying a 400Hz signal at 0.125 volts)
Guess the logic would be that while the one tube draws more current, the other draws (almost) the same amount less.

So I conclude that the bypass cap only matters if one tube goes into saturation (and therefore the change in current becomes different with the other tube)?
Is this correct?

Any hints?

Best regards,

Stephan

[Jana]

Mostly true.

Depending on where the bias is at and the impedance of the OT, there might be a period when it moves from Class A into AB1 but prior to clipping. In this case, the capacitor will have an effect.

[teemuk]

I would assume that for an output stage, you'd want to effectively bypass the entire audio band, perhaps a tad more. So within certain range of cathode bypass capacitance, all AC signals practically see just a short circuit from cathode to ground.

If they don't, it means greater signal at certain frequency and below is generated to the cathode (the cap stops bypassing due to it's increased reactance), the degenerative negative feedback increases at said frequencies, and you have lesser gain at said frequencies.

You changed the cap.... Begs a rhetoric question how much did the capacitance change in comparison to former capacitance in the first place? If you went from, say, 25uF to 20uF or even 100 uF it's way different change than, say, going from 20 uF to much lower values like 1uF.

Drastic changes should be audible as either enhacement or attenuation of lower frequencies, depending on whether you choose high or low values of capacitance. But beyond certain point towards each direction the difference will become inaudible because turnover frequencies shift away from the audible band.

[Ken Moon]

If you want to hear a difference, try removing the bypass cap entirely, then try it with a 1,000uF bypass cap.

With an unbypassed cap, you get a smoother, more compressed, lower volume sound, and it seems to help reduce blocking and/or crossover distortion somewhat.

With the 1,000uF cap, it sounds more like a fixed bias circuit, with more clarity and volume.

The "zener-assisted cathode bias" method puts a zener in series with the cathode resistor, (not sure how the bypass cap connects), but it sounds interesting and may be worth a try.

[SoulFetish]

sbirkenstock, try thinking about it this way...

Cathode bypass caps are essentially a high pass filter for AC signals shorting to ground. The value of capacitor (100uF, 60uF, etc) is chosen based on the value of the resistor used and the frequency you set at a cutoff of -3dB.

the formula used for determining the half boost frequency is (providing a given resistor and capacitor value):
fg = 1/(2*Pi*R*C)

and the formula used for determining the bypass capacitor value given resistor value and chosen cutoff frequency is:
Cathode bypass capacitor value = 1/(2*Pi*Resistor Value*Frequency)
Ck = 1/(2*Pi*R*F)

for instance lets say you have a shared cathode resistor of 160R and you want to bypass all frequencies down to 10Hz.
Ck= 1/(2*3.14159*160*10)
Ck=99.47uF nearest standard value is 100uF (giving you a -3dB of 9.95Hz)

now lets say we need to choose a bypass cap giving the same 10Hz cutoff frequency, but we have a 350R cathode resistor.
Ck= 1/(2*3.14159*350*10)
Ck=45uF(essentially)
easy peasy, my man. You dig?

Thats why swapping between the same values of capacitors may seem to make a large difference in some circuits, but not in others.
hope this helps

[pdf64]

The tube's internal cathode impedance is in parallel with the cathode resistor, as well as the cathode impedance/s of the other linked tubes.
Cathode impedance ~=1/gm
So this may be more significant with some tube types than others, but should be considered whatever.

When both tubes are conducting for the full waveform, with linked cathodes, the balanced signals will largely cancel out.

In this condition a bypass cap has very little effect, as there's little Vac to send to ground.

As the signal level increases past the point at which the tubes are in cut off for some portion of the wave, for the time they're in cut off the cancellation will stop and the cap will begin to make a difference.

So the only condition in which the cap is working all the time is when the amp is being heavily overdriven.

[SoulFetish]

During the part of the duty cycle when one side of the output stage goes into cutoff, the load impedance reflected on the conducting tube/s is 1/4 the plate to plate impedence causing a significant increase in current drawn. Couple that with bias shifts occuring when the grid begins to conduct current in AC coupled output stages, this is where most of the "problems" occur anyway.
Bypassing cathodes do not solve all design issues in an output stage anyway. But, they do affect the performance and should be considered in the decision on how to use them or whether to use them at all.