Re: tonight's SRPP JFET preamp gain stage

Hi teemukyttala,

On May 9, 10:00 am, teemukytt...@xxxxxxxxx wrote:
Just my two cents,

It seems that the circuit has way too much gain for any realistic
application. I think you can omit the cap bypassing the source
resistor and still have a circuit that provides enough of gain. If
not, then just leave the cap there but put some series resistance to
it. Perhaps a pot for gain control?


I appreciate your thoughtful advice. Like I mentioned before, I
desire a gain stage with a good PSRR. Not that it matters for a
single gain stage running off battery power - it totally doesn't - but
for cascaded gain stages and an AC power supply, good PSRR is
something I really want in the future.

I agree that this circuit does sound pretty bassy, but that cap has to
stay. That cap is the reason why this circuit has good PSRR. It
unbalances things just right! It doesn't have to be a cap at all - it
could be a couple dropper diodes, a zener, a battery, a DC bias supply
with (relatively) low output impedance - a cap is just the most
convenient thing to use. It's big for PSRR - I know I don't need a 7
Hz lower passband - that's not why it's so big.

Take the schematic and pull out the bypass cap. Imagine that the
JFETs are identical and that the trimpot is set at precisely 47K -
it's really not too far from the truth here anyway. Ground the input
signal, no input at all, shorting out the gate resistor. Now you've
got 2 identical circuits in series, stacked up between V+ and ground.
Each is naturally going to drop exactly half of V+, it's a voltage
divider. PSRR = 3 dB, which is pretty bad. Might as well be using a
drain resistor. It's like a stretchy string, everything wants to
move, the resistors and JFETs all want to drop more of the voltage.
Now put the cap back in. Blip a positive-going pulse onto V+.
Pulling on a string, voltages rise everywhere except - - - the drain
of the lower JFET is being held down. Now Vgs of the lower JFET is
greater than Vgs of the upper JFET, turning it 'more on' than the
upper JFET, this pulls the drain of the lower JFET right back down.
Run SPICE on it, you'll see what I mean, that cap is the key. I
imagine that were that cap to get TOO big, in real life as opposed to
in simulation, that the bias of the SRPP would wander up and down at
random. But that's just my usual paranoia.

I suspect that the bassiness is only an issue because I'm driving it
from a high-impedance source. If I had a buffer stage between the
guitar and the SRPP, the gate capacitance wouldn't be such a big
deal. Or even put a passive EQ network between that buffer stage and
the SRPP, that's a good possibility. Definitely a thing to be
tweaking.

A major part of the clipping in your circuit comes also from the
source follower. Such circuit has 100% degenerative feedback so the
clipping onset is quite abrupt and clipping is not very soft. That’s
not a bad thing, though. In fact, this characteristic is pretty widely
utilized in tube amps as well. The source follower also introduces
plenty of asymmetry – and therefore even harmonics. The SRPP by itself
would introduce less of those kinds of characteristics since it’s a
push-pull stage.


I was trying to bias the source follower so that it wouldn't clip
unless the SRPP was. But maybe I don't have it right. I suppose it's
all good.


What comes to “metal crunch”, that’s pretty much a result of careful
voicing. Take for example the typical FET circuitry from SS Randall
amplifiers: Two high gain common source circuits (latter one of them
directly coupled to a source follower) are coupled in a way that
reduces bass frequencies a lot – I mean a lot. Treble is pretty much
reserved throughout the circuit. Cutting the bass pre-clipping
preserves much of the “clarity” and “definition” because clipping of
low frequencies tends to “mask” the higher frequencies that ride at
the top of the LF wave. Reducing bass pre-clipping creates a less
“buzzy/fuzzy” tone. The approach is quite the opposite of what your
circuit seems to do: There’s not much bass roll off and the gate
capacitances reduce the treble.

Continuing on the Randall circuit, the FET stages have a huge output
voltage swing but they are followed by a normal shunting diode
clipper. This produces almost square-wavish overdrive, except that the
signal before this clipping stage is already softly “pre-limited” by
the FETs and clipped in asymmetric manner. This causes the diode
clipper to introduce a duty cycle modulation; which in turn enhances
even harmonic content. Finally, the whole stage is fed to an ordinary
“mid scooping” tone control that is voiced to reduce high frequencies
and enhance the bass frequencies. I can't remember what the scoop
frequency was but it has a big effect to overall tone. The bass
boosting pre-clipping creates the massive bottom end “chunk” metal
tones usually have.


I haven't really gotten to see any GOOD solid-state guitar amp
schematics. Are these Randall schematics available online? There's
about a billion crappy homemade op-amp with clipper diode circuits, I
think nobody needs to see any more of those!

Now, that’s just one example and its point is mainly to show that
often creating a good sounding overdrive is simply something that
cannot be done in a single circuit. Yes, SRPP stages overdrive nicely
but that’s just one part of the whole thing. The way how you process
the clipped signal frequency-wise is the other part and it often has a
way bigger effect. You could, for example, experiment with a graphic
EQ following that gain stage and see what you can do by varying the
frequency response. Try different HF and LF roll off frequencies and
introducing scoops and boosts at various bands. You might find some
interesting settings. Sometimes doing this voicing gradually, stage-by-
stage, creates a more "organic" tone. That's what most tube amps tend
to do anyways.

Also, expecting that overdrive circuits do clean tones well is pretty
pointless. That approach is always a big compromise between various
things. Just use different channels for clean and OD.

Yeah, I was sorta figuring that I'd switch gain stages in and out of
the same basic layout. That's how I did my last couple projects - one
was a clean multistage JFET amp with a switchable bipolar gain stage,
the other is a tube amp where you can switch one gain stage in and out
(to give you either 2 or 3 stages before the P.I.). But I could
definitely see doing things a more flexible way for a future project.

Thanks for your comments and advice, I do appreciate you taking the
time to let me know what's up with my circuit,

.