Cathode bypasses - partial bypass and combination bias

[andyjevans]

Generally cathode resistors are bypassed in SE stages, but what other options are there?

a) Partial bypass - split the cathode resistor into two resistors and bypass one of them. In this case the upper part or the lower? What ratio of bypassed/unbypassed?

b) Combination bias - combine grid bias with cathode bias (bypassed or not)

Could anyone talk about these options and give examples? I'm thinking about a 300b output in a SET but it's a general principle. I'm wondering if any of these options might have some advantage over a simple bypassed resistor.

Andy

[DF96]

In this case the upper part or the lower?

If no other connections (e.g. feedback to the cathode) it makes no difference. If there is feedback to the cathode, then it may matter. The LF behaviour will change, and any distortion from an electrolytic can be inside or outside the loop.

Combination bias

I have done this for a modified 5-20. OK, not quite the same as a 300B SET but some issues are the same. Combination or mixed bias has some of the advantages and disadvantages of both types. The reason I did it was that it allows the quiescent bias setting to be nearer the optimum bias for both small signals and full signals, and the valves can run cooler. This is because cathode bias always needs to run valves hot, so that the bias shift caused by second-order distortion from signals does not push them too cool. Mixed bias gives less shift, so they can start off cooler anyway.

In addition, you get the adjustability of fixed bias and the safety of cathode bias.

[andyjevans]

While we're on the subject, how do you calculate the value of a cathode bypass cap for a 46 where the plate (375v) is loaded with an interstage transformer (Hammond 126C) and there's 135v on the grid from the direct coupled previous stage? Cathode resistor is 14K. Current is around 10-12mA

andy

[DF96]

The LF point is set by the bypass capacitor reactance being equal to the parallel combination of the resistance looking into the cathode (roughly 1/gm but modified by the anode load for a triode) and the cathode resistor. In your case the resistor is so high that you can ignore it. Just work out the cathode impedance. I can't remember the exact formula but it is something like (1/gm) + (anode_load/mu).

[Miles Prower]

Originally Posted by andyjevans

Generally cathode resistors are bypassed in SE stages, but what other options are there?

a) Partial bypass - split the cathode resistor into two resistors and bypass one of them. In this case the upper part or the lower? What ratio of bypassed/unbypassed?

This is more useful with solid state, where partial emitter degeneration helps to compensate for the nonlinearity of re. Not so useful for hollow state because you don't have that highly nonlinear behavior, because unbypassed cathode resistance drives up plate resistance (triodes). Triodes don't have the gain margins to burn like transistors either.

For the most part, cathode degeneration resistors can be set to be equal to the reciprocal of yf, or determined empirically:

Set up the tube with the necessary cathode resistance to establish the Q-Point. Connect a variable resistance in series with a large capacitor, and parallel the RC series circuit with the cathode resistor. Adjust the variable resistor until you get the results you're looking for. Remove, and measure the resistance. The parallel combination of the cathode resistor and the settingof the variable resistor becomes the portion of the total cathode resistance you'd use for degeneration, the rest being bypassed.

b) Combination bias - combine grid bias with cathode bias (bypassed or not)

Could anyone talk about these options and give examples? I'm thinking about a 300b output in a SET but it's a general principle. I'm wondering if any of these options might have some advantage over a simple bypassed resistor.

Combined cathode and fixed bias offers the advantage of being able to adjust balance between phases in PP amps. It somewhat mitigates the effect of pure cathode bias, and may be useful if paralleling VTs, and/or for those types where fixed bias isn't recommended (e.g.: 6AS7)

[Eli Duttman]

Andy,

I'm a big fan of combination bias, especially in the case of the 300B. An unbypassed cathode resistor of 150 Ω does 3 things for you: a reasonably convenient test point for setting the "idle" current is available, some protection against runaway (a known issue with "fixed" bias DHTs) is provided, and (courtesy of degeneration) the tube is linearized. Remember, even Harvey Rosenberg was honest enough to state the 300B puts on a show. IMO, the price paid in reduced gain and increased O/P impedance is not excessive.

[andyjevans]

Hello Eli,

Very useful - did you work out 150 ohms empirically by listening or is it a convenient value? What kind of bias supply did you use?

I'd forgotten Ultrapath and wonder if I should also try that.

Andy