Making Jfets more 'tube-y'

[Alpha579]

hey all,
the ROG emulaters have had great feedback, but most think there is something missing...and someone (i think it was RG) pointed out these sound similar becasue of the frequency response, not dist characteristics...

so, my question to all those people well versed in electronics...what are the main differences between tubes and jfets? can we do anything to the jfets to make them act more like tubes?

thanx,

Alex

[JimRayden]

By understanding how tubes work, it's pretty logic that it also "loses" some of the electrons. That causes certain imperfections and it, well, sounds like a tube.

What about adding a heater to FETs? Just to make it heat up like a tube...

Ok, ok.  

---------
Jimbo

[PaulC]

The obvious problem with the jfet circuits is that a jfet is not a tube.  You can't just sub it in and try to get a certain voltage set with a drain pot.  What you'd want to do is try to emulate the wave form, and headroom of the tube version.  If you looked at the JCM800 style preamp the 2nd stage has a large amount of asy clipping going on that hits the 3rd stage which has a good symmetry with it's clipping.  So you'd want to get those wave shapes out of the biasing of the jfets.  Having the source resistors/caps the same as the tube version wont get it.  Also you have to time the clipping between the stages.  The voltage dividers used in the tube circuit set the balance between 3rd stage and 2nd stage clipping.  If you used the same values with the fet circuits you might find that one stage is being puked out before the other stage does it's thing.  The best way is to have the amp infront of you so you can look at the waves from each stage, and then try to get something that looks the same with the fets.

The other problem I see is with the use of the tone controls from the tube preamps.  This works pretty good if you're going into a pwr amp, but feeding these same types of eq circuits into each other doesn't really work that well.  There's a pretty good mid dip going on with the standard ladder type EQ.  Stacking that dip to me always has a sort of death metal scooped sound.  You want an EQ circuit that makes the pedal sound good when fed into the front end of an amp.  One trick is to have the stock JCM800 tone stack so you can shape the fet wave, and then have a bax tone stack after that to help shape it into the amp.  This idea was used in the 3+ bradshaw preamp to help match it up with different pwramps/speakers.

Trying to get some bias shift, and sag going helps to.  Big pwr amps compress when slammed.  Getting that into the fet circuit helps give it a more real feel.

Later, PaulC

[Transmogrifox]

Something I just can't divorce from the JFET and MOSFET is the "wooly" intermodulation characteristic.  I believe there are two factors in the semiconductor devices that cause this:

1.  They are nonlinear enough for the small signal characteristic that the distortion added before the limiting region happens somewhat muds up the sound.

2.  They have an inherent noise characteristic that is not found in tubes, namely, shot noise and 1/f noise, the latter I think more unfavorable than the former.


I believe the 1/f noise is modulated with the signal therefore generating the solid state wooly characteristic in the clipping.

Two things I have tried with some succes, though limited:

Design for lower noise high fidelity circuits that just happen to clip.
Use a feedback loop that keeps the signal linear up until the threshold where you want it to clip, then the feedback circuit is defeated at the threshold of clipping, yet some forward gain allows the FET to impose its nonlinear characteristic on the circuit until the FET itself clips.

Basically, you want a transfer curve that stays very linear until you get near the limiting region, then you enter a soft curve, then the soft curve turns very abrupt.  A circuit I breadboarded at least got the tube "splat" in there, though I couldn't get away from the wooly sound.

[rubberlips]

What about adding a cap to maybe slow down the FET (effectively decreasing the slew rate I guess) at higher levels. I guess it would be working something like when you're trying to make a sine wave from a square wave with caps, it's just smoothing the edges.
What effect might that have??

The circuit I was thinking about was from a 555 timer which goes through a 100n cap, then 47n cap to ground, 10K series resistor to 47n cap to ground, another 10K resistor in series, another 47n cap to ground another 10K resistor in series which goes to the base of an NPN transistor. The transistor has a 1M and 10n cap across the base and collector. Each stage is trimming off a little of the square wave to make a sine wave.

Of course you wouldn't want to go back completely to a sine wave, but maybe one or two stages would knock off enough to make it look like it was compressing. Not too sure what it would do to the frequency response though! But it might be fun to play around with

Pete

[Rob]

A JFET circuit I don't believe simulates grid conduction like a tube does.  A way to do this is to simply connect a diode from gate to source with the anode connected to the grid.  Numerous people have done this... Peavey, Eric pritchard etc.  It really helps the circuit breath.    You'll have to work the gain and Vcc of the circuit to make it play all together.