Mosfet gain stage source bypass capacitor

[Sheldon]

Hello,
can you please explain me the role of the source bypass capacitor and how to choose its value?
On the following :

I think it bypass the source resistance for AC signal so will it increase the gain?
it seems to also act as a high pass filter, and I'd like to decrease the cut frequency, but if I double or triple the capacitor value, I'm afraid it will also increase the gain and change how the gain stage reacts and sounds.
Or should I also double or triple R12 value to compensate?

[Plexi]

As I know, the Cap from Source, in parallel with R4, to ground, is what made more/less presence/fat tone.
I won't add gain, if the Mosfet is biased right from the Gate (R7).
Fixing that value, and play with the R4 from ~5k (BS170) to 0 (direct to ground) you'll get full distortion.
Take a look to the SHO or EQD Black Eye:

[PRR]

> if I double or triple the capacitor value, I'm afraid it will also increase the gain

No. There are two cases.

For super low frequency, gain is dominated by R4, 3.3K.

For high frequency, gain is dominated by R4||R12, 0.767K.

The change-over is between 22uFd+1K, 7Hz, and 22uFd+3.3K, 2Hz.

So *already* "ALL" audio frequencies are at the gain set by 1K||3.3K. Making the cap larger will not change response or gain in the audio band.

For *guitar* I could see making the cap *smaller*. 2uFd would just-shave the bottom note of guitar. For many situations we want to trim much of the bottom octave, so more like 1uFd. Marshall tube amps have similar values at cathode, and we find 0.68uFd in a lot of these.

> I'm afraid

Oh fooey. Just do it. Put 100uFd, 10uFd, 1uFd, 0.1uFd in there. Nothing bad will happen: no smoke no stain on the carpet. Values less than 1uFd can be "less muddy". If your system (guitar, amp, speaker, room) is prone to mud, then bass-shave may be good. Other systems and players may want God-bass, and up to 22uFd brings out all musical bass, larger values are just squandered pennies.

[antonis]

I think it bypass the source resistance for AC signal so will it increase the gain?
it seems to also act as a high pass filter, and I'd like to decrease the cut frequency, but if I double or triple the capacitor value, I'm afraid it will also increase the gain and change how the gain stage reacts and sounds.
Or should I also double or triple R12 value to compensate?

About possible gain increase, Paul said enough.. 

C1 + R12 indeed from a high pass filter shunting R4 and almost grounding MosFet Source..
(it actually lowers Source resistance by a factor of more than 4..)

Your query about "compensation" rises two issues:

1. Do you ρεαλλυ need any kind of compensation..??
(in the mean of you alter some item's value for some reason - compensation "neutralize" it..)

2. For any first order passive filter, cut-off frequency is conversely proportional to RC product so, for compensation purpose,  doubling one product factor needs halving the other one.. 
(if you tripple cap value and also tripple resistor value you result in a cut-off frequency of nine times lower..)

[Sheldon]

thank you for your answers.
In fact I try to make a design more suitable for bass guitars. The lows are dropped too soon under 100Hz for my taste.
I discovered my "problem" doesn't come from the first gain stage, but from the second just before the tone stack.
In this second gain stage there is no R12 resistor, the bypass cap (10uF) goes direct to the ground, which makes quite a noticeable high pass.
I think I'll have to increase a bit the bypass cap value and adapt a little bit the tone stack.

[antonis]

A 10μF cap has an impedance of 160 Ohms at 100Hz rising inversely proportional with frequency..
(200 Ohms at 80Hz, 266 Ohms at 60Hz e.t.c.)

That impedance is set on denominator of gain calculation fraction (R_Drain/R_Source) and it lowers the gain with frequency lowering..
(it's actually dominates Source resistor by setting itself in parallel ...)

You can try a bigger cap with a trimmer in series (like R12) to make adjustment of your taste..

[anotherjim]

It could be a cumulative effect. Each corner freq calculation is the -3dB point. Where is the -1dB point? Several high passes throughout the circuit mean those -1dB cuts add up. What are capacitor value tolerances? Murphy's law says those tolerance errors will add in the wrong direction and you could easily have less bass than you have on paper.

If you have a full set of tone controls, it could be said that it isn't any business of rest of the circuit to alter the tone. If you agree with that, then make sure all your interstage coupling and bypass caps are certain to pass the full range of any instrument you might plug in. There isn't much of a penalty for over-sizing them apart from cost, increased board area and maybe a few more milliseconds longer time on power up before things settle. Actually, cost/area doesn't always increase, 3n3 film cap hardly different from 10n film cap.

In the part circuit posted. Input of 3n3 is a touch tight for bass. 4n7 would be better, but why not 10n?
Electro cap tolerances are such that a particular 22uF cap might be significantly lower capacitance. 47uF would be safer.
can't see the rest of it, but is that 100nF output cap big enough?

No matter what you do with the tonestack values (assuming Fender/Marshall TMB type), you may find the bass control has to be close to full on to get fullest bass extension. See Ampeg's "James" type circuits.

[Sheldon]

thank you for your answers