[danieleb]

in these days I was trying modifications to my JC18, having very large grid stoppers (100K) compared to the common ones on EL84, I tried to reduce the values ​​of the long tail resistors of the PI (r16 1R->550R and r17  27K->17K) and the sound change a lot, less compression and more attak at rehearsals levels.

Could be a trick (in opposite way) to have a brownish sound in an EL34 amp?

[Helmholtz]

Fine, if you like the result.
Technically your changes lower the PI headroom, causing early PI distortion and probably reduced clean output.

[Showitevent]

Hi,

for my understanding most EL84 amps dont work too well having more reasonable Gridstoppers like you would find in EL34 amps.

The compression you feel is most likely the loss of some treble content with high value Gridstoppers. The EL84 tend to overdrive relatively quick and harsh when they start to draw current (just my subjective feeling) and more or less sound broken easily than any usable - its a matter of finding the sweet spot here.

EDIT: In case of self biased amps this also depends on the loadline that the Output transformer draws to the Tubes which i would assume is nearly always relatively steep.

If another amp reacts the same way on this change, depends on its guts and strongly on the type of tubes, manufacturer, voltages, e.g....
A more brownish sound is also not necessarily a question of the Poweramp alone. There are various variables to consider from the input to negaitve feedback. But i find it true that "brown" mostly finds place in the backend of the amp. 

Geronimo

[Helmholtz]

The EL84 has an effective input capacitance (including Miller effect) around 20pF.
Even if we assume twice that value (taking into account socket and wiring capacitances), the treble loss with an 100k grid stopper will not start below 40kHz.
I rather think that compression is caused by the signal voltage drop across the grid stopper with beginning EL84 grid conduction.

[Showitevent]

The EL84 has an effective input capacitance (including Miller effect) around 20pF.
Even if we assume twice that value (taking into account socket and wiring capacitances), the treble loss with an 100k grid stopper will not start below 40kHz.
I rather think that compression is caused by the signal voltage drop across the grid stopper with beginning EL84 grid conduction.

Im not speaking about miller capacitance. I also said that there is a subjective feeling of losing treble content due to drive. This may be drive current or miller capacitance or sideeffects that come in when pushing the tubes harder.

[danieleb]

Fine, if you like the result.
Technically your changes lower the PI headroom, causing early PI distortion and probably reduced clean output.

yes, for now I like how the JC18 sounds, I felt it was too compressed at high volume, I had tried reducing the bypass cap on V1a and reducing the C10 and C13 coupling caps to 22n, but I was still not satisfied. Now I have restored 100n without losing dynamics and (with OX UAD) I can have a sound more similar to how I imagine a jmp master volume. I haven't tried doing the opposite thing on the 50w yet.

Hi,

for my understanding most EL84 amps dont work too well having more reasonable Gridstoppers like you would find in EL34 amps.

The compression you feel is most likely the loss of some treble content with high value Gridstoppers. The EL84 tend to overdrive relatively quick and harsh when they start to draw current (just my subjective feeling) and more or less sound broken easily than any usable - its a matter of finding the sweet spot here.

EDIT: In case of self biased amps this also depends on the loadline that the Output transformer draws to the Tubes which i would assume is nearly always relatively steep.

If another amp reacts the same way on this change, depends on its guts and strongly on the type of tubes, manufacturer, voltages, e.g....
A more brownish sound is also not necessarily a question of the Poweramp alone. There are various variables to consider from the input to negaitve feedback. But i find it true that "brown" mostly finds place in the backend of the amp. 

Geronimo

The JC18 is fixed bias, sure 100Ks are darker than the marshall stock, could be a way the design eliminated some of the treble from the JCM800, I have also the pink taco and Friedman used a lot of cap to low pass the signal, but with a "standard" marshall el84 power section.
The JC18  made me think that it was not necessary to saturate the PI too much, I'm not sure mathematically, but in the JC18 I felt a lot of PI distortion, cool at low levels, but too much over 4 MV. It actually seems cleaner with decreased values , however, counter-reaction could also be involved, I'm not shuer it is 100% headroom.
The reverse modification on a 50 or 100w, without modifying the grid stoppers, would be convenient to mount on a switch with resistors in parallel to have the 2 options. If it works, because if it is the negative feedback, on an amp with 100k NFR on the 4Ohm, it could already be too little.
Friedman increases the resistance on the anode side to reduce headroom and uses bypass cap to cut hi end, but there the voltages are much higher to use a small switch. I never see mods on the long tail to simulate a "variac" the doubt that comes to me is that it isn't there because it doesn't work 

[danieleb]

The EL84 has an effective input capacitance (including Miller effect) around 20pF.
Even if we assume twice that value (taking into account socket and wiring capacitances), the treble loss with an 100k grid stopper will not start below 40kHz.
I rather think that compression is caused by the signal voltage drop across the grid stopper with beginning EL84 grid conduction.

I only read it now, so darker, but far from audible frequencies.
I think it is the compression of the PI which, with the same MV, with the colder bias, tends to cut the attacks. With the long tail changed, raising the bias, they probably pass more dynamically to the EL84s which are still far from distortion thanks to the 100k resistors, but here there is also the negative fedback which I don't know how it is modified by the long tail

[Showitevent]

Hi there,

im not quite happy with my answer above hence this BS post All just pure theory, not anything of this is proven, nor calculated!!!

1.) Correction
I know there is huge effect on a EL84 Push Pull amp with larger Gridblockers - all just subjectively based on own tests with some amps.
With large Gridblockers i tend to say that if you drive the amp harder, it doesnt have that unpleasant harshness in tone as it has with no or low value gridblockers - spoken for EL84 only here as i never tested this for other tubes aswell. Hence you can state that it feels like compression in lowend depending on when this effect would take place (clipping for example).

I also investigated that fixed bias amps react a little different here than Cathode Biased amps, regardless of the fact that both amps (that i tested with) had nearly equal B+ (from memory) but very different output transformers. So i had a typo in my above post, i was referring to Fixed Bias where i wanted to say Cathode Biased.

That must of course not be the case in any situation and there is of course always a subjective feeling of difference when you focus on a change you made. I did not do any measurements for that and its been a while.


2.) Miller Capacitance - Just pure theory in my brain!

I am not entirely sure about miller here, hence i didnt mention any miller capacitance in my inital post.
In Push Pull Amplifiers we almost never can rely on the idea of it having just a gain of 1. A gain of 1 would indeed imply that a 100K Resistor wouldnt have any effect in the hearing range.

I would assume that you can safely add up capacitance x 2 because we have two systems building one signal (bare theory) and if i have calculated correctly you already took that to consideration, hence 40KHz. Then its also questionable if we also can add up the series resistance we have from our gridblocker - my brain cant put that together at the moment. 

But now we still have the amplification factor which is X to us. We simply dont know how often we have to multiply the miller capacitance in this system.

If i remember correctly ( im sure Aiken will cover somewhere )
CIn = CgK + CgA * A , where A is the amplification factor - i think you even have to add 1 to A for the sake of correctness.

Plus other variables:
- We are possibly in a closed loop with Negative feedback (how much impact this has on linearity correction is a variable)
- We need to consider that if we push the tubes to B+, we cant build up any more amplitude but we still have nearly same amplification factor (depending on breakdown of B+). This of course applies to the full frequency range but i would assume that lower mud overrides higher crap.

Then one last thing to mention is that:
Lets say you have a typical EL34 or 6L6 Amp. Their Gridblockers usually are way lower in the 1 to 10k range. If you consider that their CgA and CgK are near identical (lower) than for an EL84 and a low value gridblocker is enough to stop this from oscillation, then my theory might not be far off from reality. There will be hughe impact of miller, due to amplification factor.


So in my opionion, there is only the chance of doing a frequency analysis via Spectrum analyzer. And since each amp is different (Transformers, B+, etc.) you may not end up with a reliable statement that large Gridblockers have an effect on X.

Cheers,

Geronimo