By-pass cap calculator for Jfet, Mosfet & BJT

[WGTP]

I found this cool calculator for by-pass caps for commonly used tubes.  I'm wondering how it can be manipulated to work with our favorite J201, BS170 or 2N5089 transistors  There are other calculators that deal with various tone shaping circuits that play a major role in the various amp, sim type distortions, Dr. Boogie, BSIAB, etc., in addition to the standard TMB tonestack.  Any ideas?

http://www.ampbooks.com/home/amplifier-calculators/cathode-capacitor/

[PRR]

It's not that simple.

In tubes at audio frequency the cathode network has almost no effect on the input grid.

Tubes have a lot of dumb resistance. You can get a good idea of cathode-network conditions with only a little info about voltage and other resistances.

Tubes generally work in a narrow range of currents.

BTW: That calculator gives wrong answers for EXtreme cases. (Vary RK from zero 10 Meg.... gain does not change, even though tube current must change by factor of 30 and internal cathode impedance must change at least factor of 5 {and useful output with RK=10Meg would be mighty darn small}).

While a tube is rarely run over more than a 10:1 spread of current (12AX7 0.2mA-2mA), BJTs offer a much wider range of happy working conditions, 0.02mA to 20+mA. And while tube impedances "tend" to vary as square-root of current (2:1 of current is only 0.7:1 of impedance change), in BJTs it is linear (0.02mA to 20mA is 1000:1 of impedance change).

BJTs have very significant emitter-base interaction. What happens in the emitter also affects what happens in the base coupling network. (This can be seen in some of Gus' circuit ideas- a bass-bump is possible, so is unexpected bass-slump.)

There's less than a dozen basic tube-types in common use. There are 987,654 different BJT type-numbers. Though for this purpose they are more alike than different, this is not user-friendly.

Whole books could be written on capacitors versus transistors. I am not aware of one. Here's what Cowles says: "...a common misconception that {the emitter capacitor's} size is determined simply by making it smaller than the emitter resistor... Actually the emitter capacitor depends more on the impedance of the base circuit. ... ... For a voltage source or for high Beta, Ce must bypass h_ib of the transistor!"

[WGTP]

Thanks, I guess I should have known.  Looks like it's good old trial and error.   

[PRR]

> trial and error

-OR-....

Estimate h_ib of the transistor using transistor current and Shockley's Law.

Estimate source impedance (resistance and capacitance) seen by transistor Base, divide by Beta.

[rring]

I am not sure I agree with this. Reactance is reactance. If you look at Re as simple two port series, series feedback, roughly gain will be Rc/Re. If you have a Re of 1K and an ideal capacitor of 10uF, the reactance at 15Hz is approx 1K ...so your (Re||Ce) is now 500 ohms @ 15Hz and you should see a 3dB increase in gain. I checked this in spice and it behaves exactly like this. Using two port parameters you also get the same result - what am I missing here?

I have been wrong in the past but for a capacitor (performing close to ideal, low esr,low freq, etc) - i just dont see how its not simply paralleled with Re

[PRR]

> a 3dB increase in gain.

Right.

But.... we also (and more usually) want to know the 3dB down point. What is the ultimate gain? (Where cap reactance approaches zero.) Where is the response 3dB down from there?

This requires knowing the "internal Re", internal to the transistor emitter.

And.... the effective internal Re is in series with whatever impedances are seen by the Base, divided by Beta. Since in a non-SPICE world this is some resistance and usually a capacitor, the situation is not simple.

[rring]

Yes I understand your point -  I am working from a perspective that getting within 1dB or so of predicted performace is sufficient. Bottom line: I guess I am too lazy to exact your level analysis because very basic calculations will get you within this margin of 1dB.

[PRR]

> exact your level analysis because very basic calculations will get you within this margin of 1dB.

One of us is missing the point.

You have computed point A, the 3dB UP point:



In audio, we more often want point B, the 3dB down (from flat-band) point.

This has almost nothing to do with the 1K resistor.