function/purpose of capacitor from anode to cathode

[mikepukmel]

in the 2 'over drive' gain stages, there is a 270pf capacitor between the anode and cathode:

dumble_124_schematic_172.pdf

Anyone know what the purpose of those caps are? I poked through some schematics I have for other amps, and haven't seen anything like that.

Is this correct:
It looks like 'local negative feedback', since the output signal from the anode has DC stripped off, and is fed to the cathode, that has in phase signal compared to the input.

[eschertron]

Taking the anode bypass cap to the cathode is "almost as good" as taking it directly to ground. The signal through the 220p caps aren't going to notice the 4.7u caps at all. Some designers take the anode bypass to B+ (another short and easy soldering route) but Merlin and others warn of noise on the B+ getting into the signal path that way. I'll have to look at what the anode bypass actually does in engineering terms, but high-gain stages tend to have them. Tames the gain on high frequencies.

edit: V1b does have local FB, with the string of 10M resistors and the .05u cap. I'd think there is some inductive stuff going on in that loop, too, with all that lead wire hanging into space!

edit: 220k/40M is not much feedback any way, so why even have it in? Something to squelch parasitic oscillations in a particular build, and then slavishly copied into amps that might not even need it? Ah, mojo

[Enzo]

I'd also vote for solder. Bringing it to cathode instead of ground means it can mount right on the socket.

[trobbins]

The output signal from V2B triode can be thought of as a voltage supply with an internal plate resistance. The plate resistance is then the upper half of a voltage divider, with the lower half being the effective impedance of all the circuit paths connecting to the plate. For most circuits within the audio range of frequencies, that external impedance is just the anode 100k resistor paralleled with the 350k Level pot circuit resistance (100k + 250k), so about 78k (the 0.01uF coupling cap is assumed to be zero impedance).

Frequencies above a few kHz (8kHz corner) will start to see the 270pF also being an external parallel path for the signal voltage, so the voltage divider ratio lowers, and hence less signal reaches the level pot.

[mikepukmel]

So, this is more a parasitic oscillation fixer upper thing, rather than a tone thing? (sorry for the low tech terms)

[trobbins]

No I reckon it is very much an overdrive distortion tone control measure. Overdrive of any stage causes substantial generation of higher order harmonics, so the shunt capacitor is effectively tone-controlling those distortion harmonics.

[Leo_Gnardo]

No I reckon it is very much an overdrive distortion tone control measure. Overdrive of any stage causes substantial generation of higher order harmonics, so the shunt capacitor is effectively tone-controlling those distortion harmonics.

I take Dumble's hi freq filters to be there for both reasons. Some band limiting for guitar signals is a good thing at both ends of the spectrum. Old audio engineer adage "the wider you open the window, the more crap flies in."

But I have to wonder what he's doing with 4 x 10Meg resistors. Heck I got resistors that would replace that kludge, I'm sure he can get 'em too. One might guess the master builder is using his setup as a radio frequency "gimmick" besides its obvious filtering function? In any case Mike, gazillions of perfectly good Fenders & other amps were built without this Dumble twist. I'd concentrate on building a good amp first. Once it's proven working then try this if you dare.

[nickb]

The effective plate load impedance is about 40K ohms so a 270pF capacitor will form a low pass filter which will be -3dB down at about 15KhHz (=1/2/pi/R/C). A guitar speaker won't reproduce any signals above about 5 KHz so I must conclude that it's was put there to primamrily to prevent parasitics.

[trobbins]

I think the 270pf corner frequency of concern is when it equates to the low frequency ('mid band') impedance seen by the plate (ie. circa 78kohm).

It would effect the harmonics of the fundamental.

[nickb]

I think the 270pf corner frequency of concern is when it equates to the low frequency ('mid band') impedance seen by the plate (ie. circa 78kohm).

It would effect the harmonics of the fundamental.

Doing things with a little more precision, all the various plate impedeance appear in parallel so taking
Plates resistor= 100K
Plate internal impedance = 60k
Load impedance= 250K

==> effective impedance = 32K ( I had guessed at 40K, not too shoody, huh )

==>freq = 18Khz

Agreed this would be harmonics of the note but nothing was coming out of the speaker anyway so sonically it's not doing anything. There night be a case to argue that it changes the intermodulation products that result in audible beat frequencies but my hunch is the amplitude of those due to 18Khz+ harmonics was so low anyway that we can ignore them. I therefore conclude the primary purpose is to prevent parasitics.

[trobbins]

I should have started with the voltage gain expression for a simple bypassed common cathode triode: Av = u. R/(R=r) , where R is effective anode load impedance and r is plate resistance.

Assuming R is 78k, and r=60k, then voltage gain falls -3dB when R is reduced from 78k to about 40kohm.

A capacitive reactance of about 83k in parallel with 78k resistance achieves 40k impedance. A 270pF parallel capacitor has 83k reactance at about 7kHz corner frequency.

[nickb]

[QUOTE=trobbins;454094]I should have started with the voltage gain expression for a simple bypassed common cathode triode: Av = u. R/(R=r) , where R is effective anode load impedance and r is plate resistance.

I take it the "R=" in the formula was a typo and you meant to write the correct formula Av= u * R/(r+R)

The amplification factor is rather meaningless derivative quantity u=gm * rp IMHO so I prefer to use gm. So, the gain is

Av = gm * Zeff ; where Zeff is the effective total plate impedance

Ignoring the capacitor for the moment, Zeff consists of three components in parallel viz.:
1) Rp the internal plate resistance which we'll take as 60k
2) RL1 the plate dropper resistor, 100K
3) RL2 the loading due to the next stage, about 250K
so, Zeff = 250//100//60 ( where a//b is short for 'in parallel with' i.e. a*b/(a+b) ) = 32.6K

The voltage gain falls by -3dB when Zc =Zeff =32.6K (not 78k above) i.e. f=1/2/3.14/32.6k/270pF ~=18kHz.

To back this up here are the sim results showing 19.5kHz



Also, out of interest, the 3 x 10Megs reduce then gain by about -3dB. Why he did it that way we can only speculate. Just to be different, I suspect.

[trobbins]

Sorry, but can't agree. The simple triode equivalent circuit is a voltage source with 'r' series resistance - the external circuit is the anode load etc. Perhaps some references can help:
Page 131 of 1952 Army theory doc. http://www.tubebooks.org/Books/army_theory.pdf
Morgan Jones and Merlin has similar descriptions in their docs.

[mikepukmel]

Sorry, but can't agree. The simple triode equivalent circuit is a voltage source with 'r' series resistance - the external circuit is the anode load etc. Perhaps some references can help:
Page 131 of 1952 Army theory doc. http://www.tubebooks.org/Books/army_theory.pdf
Morgan Jones and Merlin has similar descriptions in their docs.

Great book link, thanks. Much better than the one I picked to read.
For the calculations, I get stuck on figuring out the load of the next stage.

[nickb]

Sorry, but can't agree. The simple triode equivalent circuit is a voltage source with 'r' series resistance - the external circuit is the anode load etc. Perhaps some references can help:
Page 131 of 1952 Army theory doc. http://www.tubebooks.org/Books/army_theory.pdf
Morgan Jones and Merlin has similar descriptions in their docs.

What is it you don't agree with? Is it the replacement of a voltage source with a current source to make the math easier? The transformation is explained, for example, here https://en.wikipedia.org/wiki/Source_transformation

If it's still not clear I do it again using a voltage source. It's just a bit longer...