'plate capacitor' JTM45 long tail pi

[makepeace]

Can anybody explain what the 47pF capacitor connecting the plates in the JTM45 long tail pair phase inverter is for? I assume it has something to do with balancing the amplitudes of the normal and inverted outputs? Good idea/bad idea? How did they choose that value?

Thanks

[jrfrond]

A small capacitor, anywhere from 47-100pF, is often placed across the two plates to squelch any spurious oscillation that might occur by suppressing ultra-high frequencies. Since the outputs of the PI are out of phase, once you've reached the corner frequency of the cap, the output are summed out of phase and cancel. It's common practice, as is a 47pF cap at the input of the first stage. It's just a way of stabilizing high-gain circuits.

[jmaf]

That capacitor cancels high frequencies between the two plates. It was added late in the Tweed era to stabilize the schmitt PI. JTM45 is nothing but a copy of 5F6-A, upside down.

The balancing of the outputs is done via the plate resistors and the negative feedback.

Edit: Opened this thread and left it there, then when I answered JRFrond had already given his reply. Didn't mean to sound redundant.

[J M Fahey]

Right.
I guess the value was experimental rather than calculated.
Anyway the idea is a balance between "small enough to avoid killing sparkle" and "large enough to be effective."
If you want to calculate it anyway, consider itīs an RC network, where C is already known (47 or, in some cases, 100pF) and R is the the impedance "seen" by each plate, multiplied by 2 .
Each plate sees 3 impedances which are in parallel:
1) Rp, the internal plate impedance. For a 12AX7 in those conditions itīs traditionally given as around 68K
2) Rl, the actual plate load . 82K or 100K. Or 47K on some Silverfaces.
3) Rg, the following grid resistor: usually 220K, also 150K and 100K are found many times.
So for a 68K//100K//220K (// meaning parallel) we get= around 35K which multiplied by 2 gives around 70K
Now: 70K+47pF gives around 48 KHz.
And popular value 100pF would give around 24KHz ; still way above the audio range.
That makes me think that beyond simple "cutting highs" (which anyway would not be reproduced by any guitar speaker) it probably cures or corrects some funky high frequency phase shift which causes unstablity.
Just a guess, of course.
Because, truth be said, it works.

[Chuck H]

Right. Absolutely. I always include a 47pf cap in ANY amp just 'cause. You don't hear it. The funny thing is you will the effect of a 100pf cap (even though, as noted, it's above the audible range)

Considering that there can be instabilities WAYYY up into ultrasonic frequencies I would say just include a 47pf cap (or 100pf where applicable for tone) on any build.

[Zozobra]

Merlins preamp book deals with the purpose of this cap in the feedback chapter, which I have a tenuous grasp on. The idea is to improve the stability of the feedback loop so its less likely to produce positive feedback by slugging the poles. I'd be doing everyone a great disservice in trying to explain what this means though

[jmaf]

I made an experimental tube preamp with a 47pF coupling cap on the first gain stage, just to make it scream with distortion on a later gain stage. I gotta say, a lot goes through a 47pF, though I didn't like the tone one bit. The string screeches from sliding finger hurt my ears.

IIRC Bassman's up until 5E6 left out that cap, I recall having read somewhere why Leo added it but I tried to find it now and couldn't. There's a lot of subtle things that Fender did which were copied forever on almost every tube amp out there, that capacitor is one of them.