FET fans..read this...

[dschwartz]

john..in my case i only do this at the most clipping stages..the later stages i keep the larger values to reatain more bass signal..

i actually make the presence cap bigger to lower the LPF corner freq and keep the bass response acceptable...

and the most important..most fetr preamps are PREAMPS and sound best, by far, using them at the power amp input,..using it at the guitar amplifier normal input will make the sound too bright..rolling down the presence knob helps a little, but alters the high end texture..guitar preamps clean channels ussually are very bright...

in the dr boogie case, keep in mind that the power amp section  of a mesa DR is pretty dark, keeping the freq leveled...

about tornado..maybe that low value caps helps a lot with the responsiveness....

[aron]

Simple mods - secret of stacked stages:

http://www.diystompboxes.com/wiki/index.php?title=Simplemods#Simple.2C_Easy_Mods.2FTips_.26_Techniques

[markm]

I may be completely wrong about this because I haven't even looked to check but,
didn't ROG cover some of this in the Fetzer Valve info?

guess not.

[dschwartz]

some clips i made:
some nice stuff (sorry for the lousy improv):
http://diynoise.googlepages.com/DRU.mp3

metal (metallica wannabe):
http://diynoise.googlepages.com/DRU_metal.mp3

as you hear, is pretty articulate and dynamic

[B Tremblay]

I may be completely wrong about this because I haven't even looked to check but,
didn't ROG cover some of this in the Fetzer Valve info?

guess not.

We did mention the frequency response differences, but mostly in reference to Miller capacitance: http://runoffgroove.com/fetzervalve.html#3

[dschwartz]

I found these 2 mentions about input impedance..maybe i´m not that wrong....

AT: http://www.tpub.com/neets/book7/26i.htm
sez (about  FETS):
"With a VGG of 1 volt, the microammeter reads .5 microamps. Applying Ohm's law (1V ¸ .5mA) illustrates that this very small amount of current flow results in a very high input impedance (about 2 megohms)."

and at:

http://music-electronics-forum.com/archive/index.php/t-300.html
sez:

"A grounded-cathode triode gain stage (I assume you meant this - the typical guitar-amp GS - rather than the common-cathode circuit shown below) has an extremely high input resistance at DC, but the effective AC input impedance is much lower due to the tube's interelectrode capacitances combined with the Miller effect. FWIW, here are some ballpark Zin's @ 20kHz, using a 100K Rp and unbypassed 1.5K Rk:

12AX7 - @ 90K ohms
12AT7 - @ 110K ohms
12AU7 - @ 260K ohms

The input impedances will be correspondingly higher at the lower bandwidths typical of most guitar amps."

so.. i may not be that wrong...
aron:
i know those stacked gain stages tricks...the thing is that most FET EMU´S schems, a big coupling cap is used, identical to the original diagram, but that´s not right, fet emu´s builders should know that 22nF for a coupling stage is too big if youre looking for clear sounding clipping.. most of us just follow the schem and dont question why this or that value.. i wanted to know why the original and the emu had different clipping characteristics.. and i think i founded it...

[markm]

I may be completely wrong about this because I haven't even looked to check but,
didn't ROG cover some of this in the Fetzer Valve info?

guess not.

We did mention the frequency response differences, but mostly in reference to Miller capacitance: http://runoffgroove.com/fetzervalve.html#3

Ah yessss.
I see.

[John Lyons]

Dschwartz
Yes, it seems that you are correct. I never doubted that, pretty nice discovery.
Glad it worked out for you.

The samples sound very nice! Nice playing as well.  "am I evil...!"
The palm muting seems articulate and the low end is cleaned up a bit.

John

[Bucksears]

Has anyone else tried this method with any of the ROG emus? I'm wondering how it might help out the 'flabbiness' that I've heard with the Umble, Thunderchief and 18.

Thanks,
Buck

[R.G.]

There are some bits here that need elaboration.

The input Z of JFETs and triodes are both orders of magnitude higher than the gate-gorund or grid-ground resistors. To all intents and purposes they are infinite.

Except...

when you drive the grid more positive than the cathode or the gate more positive than the source. This is an old, old problem in tube amps. If you drive a signal big enough to make the grid go positive with respect to the source, the grid starts conducting. It's nonlinear, but a 12AX7 grid drops to well under 10K when you do this. If the stage is capacitor coupled from a preceeding stage, the grid conduction builds up electrons on the coupling cap, charging it negative, and when the signal then goes negative, the charge stays on the coupling cap until it's bled away by the grid-ground resistor. That time constant can be long. So on especially low frequency (long time) and high amplitude signals, the tube's grid self biases colder and colder. This leads to a fizzy distortion or blocking distortion. The commonest term for this is farting out on bass notes.

JFETs do it too. Only the gate has to be one diode above the source to do the conduction, so it has a little more immunity than a tube does, since the tube starts conduction actually a few millivolts before it gets to Vg=Vk.

Lowering the coupling cap reduces the problem because it makes the problem go away faster (smaller cap for the same grid-leak resistor) and it limits bass to later stages. If the first coupling cap cuts the bass down below that which will overdrive the grid in later stages, the problem goes away for the amp. It's worst on the stages just before the power amp, as the signal is biggest there.

Power tubes do it too. A recently popularized trick to keep it from happening on power tubes is to reverse clamp the grid to ground with a zener which conducts in the reverse direction at about the same voltage as the grid conducts in the positive direction. This lets the overdriven signal suck out its excess charge as well as placing it, so it cannot build up to a bias shift.

Yes, lowering the coupling caps does the trick as long as you still get the frequencies you want. The negative clamp trick can also be used to good advantage, as can putting a big stopper resistor in series with the grid.

[John Lyons]

I haven't tried this in the DB etc. But I know for a fact that it will clear up and slop in the low end.
I would make these changes in the early part(s) of the circuit and see how it changes the sound.
Rolling off bass in the early stages will clean up the sloppy bass and give you more articulated picking and a faster response.

You may not have to change all the coupling caps to get the effect you are looking for.
If you change out all the caps to these lower values you might loose to much low end.
Cutting some bass at the early stages and then cutting treble at the later stages or the end of the circuit will give you a full sound that remains clear  and articulate.

John

[dschwartz]

yes john, in fact i only do this for the first stage coupling cap.. the latter caps are the same or even i get them bigger

[MR COFFEE]

RG,

If the stage is capacitor coupled from a preceeding stage, the grid conduction builds up electrons on the coupling cap, charging it negative, and when the signal then goes negative, the charge stays on the coupling cap until it's bled away by the grid-ground resistor.

Doesn't it temporarily raise the bias on the grid bias bypass cap as well if the driving impedance is fairly low?

[Uma Floresta]

There are some bits here that need elaboration.

The input Z of JFETs and triodes are both orders of magnitude higher than the gate-gorund or grid-ground resistors. To all intents and purposes they are infinite.

Except...

when you drive the grid more positive than the cathode or the gate more positive than the source. This is an old, old problem in tube amps. If you drive a signal big enough to make the grid go positive with respect to the source, the grid starts conducting. It's nonlinear, but a 12AX7 grid drops to well under 10K when you do this. If the stage is capacitor coupled from a preceeding stage, the grid conduction builds up electrons on the coupling cap, charging it negative, and when the signal then goes negative, the charge stays on the coupling cap until it's bled away by the grid-ground resistor. That time constant can be long. So on especially low frequency (long time) and high amplitude signals, the tube's grid self biases colder and colder. This leads to a fizzy distortion or blocking distortion. The commonest term for this is farting out on bass notes.

JFETs do it too. Only the gate has to be one diode above the source to do the conduction, so it has a little more immunity than a tube does, since the tube starts conduction actually a few millivolts before it gets to Vg=Vk.

Lowering the coupling cap reduces the problem because it makes the problem go away faster (smaller cap for the same grid-leak resistor) and it limits bass to later stages. If the first coupling cap cuts the bass down below that which will overdrive the grid in later stages, the problem goes away for the amp. It's worst on the stages just before the power amp, as the signal is biggest there.

Power tubes do it too. A recently popularized trick to keep it from happening on power tubes is to reverse clamp the grid to ground with a zener which conducts in the reverse direction at about the same voltage as the grid conducts in the positive direction. This lets the overdriven signal suck out its excess charge as well as placing it, so it cannot build up to a bias shift.

Yes, lowering the coupling caps does the trick as long as you still get the frequencies you want. The negative clamp trick can also be used to good advantage, as can putting a big stopper resistor in series with the grid.

I may have experienced this bias shift problem with my Umble pedal - sometimes it sounds great, full but not flabby, and sometimes it sounds very flabby and unfocused. I was attributing this to ear fatigue, but maybe this is the reason for it.

The zener diode solution - would this involve running a diode to ground from the gate?

[R.G.]

Doesn't it temporarily raise the bias on the grid bias bypass cap as well if the driving impedance is fairly low?

It does, but the grid bias bypass cap is a much larger capacitance, and is isolated by a big grid leak resistor, so the change is very small. Almost all of the change is kept by the previous stage plate capacitor; all of it in overdriven signal triode stages that are self biased and have not bias bypass cap.

[R.G.]

I may have experienced this bias shift problem with my Umble pedal - sometimes it sounds great, full but not flabby, and sometimes it sounds very flabby and unfocused. I was attributing this to ear fatigue, but maybe this is the reason for it.
The zener diode solution - would this involve running a diode to ground from the gate?

Yep. The idea is to give the presumably equal positive/negative peaks a chance to both load up the cap (through grid conduction) and also unload it (through zener conduction. You have to use a standard diode in series with the zener to keep it only conducting the correct direction.

[MR COFFEE]

RG,

So applying this to a N-channel JFET, the zener value would be equal to the Source to ground voltage of the stage, as in

Vz + 0.6 volts (Si diode to block reverse conduction) = Vg-s + 0.6 volts (JFET junction diode) 

Or maybe a diode string in low level stages since low voltage zeners below 3 volts are kinda strange birds ?

Thanks for this stimulating discussion !

[DougH]

Yes, lowering the coupling caps does the trick as long as you still get the frequencies you want. The negative clamp trick can also be used to good advantage, as can putting a big stopper resistor in series with the grid.

You can also throw away the coupling caps period and use DC-drive. This eliminates the blocking distortion problem altogether.

See ax84 for examples. I bet this approach could be adapted for JFETs as well.

[DougH]

some clips i made:
some nice stuff (sorry for the lousy improv):
http://diynoise.googlepages.com/DRU.mp3

metal (metallica wannabe):
http://diynoise.googlepages.com/DRU_metal.mp3

as you hear, is pretty articulate and dynamic

The clips sound very good!

[dschwartz]

yeah! if you could only hear it live!!!