Gallien-Krueger 210LC - odd use of jfets

[Umlaut]

I recently stumbled upon a couple of old schematics from GK.
This particular one, marked 1981 and labeled "Lead Preamp 210LC" caught my eye due to the unusual configuration of the distortion channel.
It seems to be the standard input stage, volume control (dual pot), two differential jfet stages and a simple fixed tonestack.
Yet I struggle to fully understand how it works. I have simulated it on LTspice, and despite a VERY boring tone shaping (low cut + high cut, just a standard dome) the way it shapes the wave seems interesting.
My questions are:

- What exactly is the role of the opamps after the differential fets?

- What are the benefits and drawbacks of this design? As opposed to, say, other GK configs in which they cascade jfet opamps, or as opposed to diodes in the feedback loop.
In short, is there anything special to this? I would assume the contrary, as I haven't seen that anywhere else, but just curious if anybody has any particular info to enlighten this.

Thanks in advance!

[antonis]

It mimics power amplifiers 1st & 2nd state design..

JFET differential pair (J1/J2 & J3/J4) is the transconductance stage followed by op-amp transadmittance stage (U1 & U2) (Voltage amplification), simulaneously applying NFB (R13/C6 & R18/C10)..

[Umlaut]

Thanks for the explanation! I will breadboard it and see how it sounds. Could be interesting as a build block, definitely.

[antonis]

I think it might worth to play with U1 & U2 gain..
(R14 & R19 values..)

[Umlaut]

I currently have a Musicman HD130 preamp on the breadboard, so I quickly tossed in the 1st one of those stages just after the volume control of the HD. Used MMBF5484 instead of J113, eyeballed a 10k resistor in lieu of the trimmer, biased the noninverting input at 12.4v (close enough) but kept all other values stock.
Well... as is, I like it better thant the original last gain stage of the HD130 ( crude LM1458 rail-to-rail clipping). Raw yet pleasant distortion, even outside its original circuit.
Observations:
- ofc, the diodes before the jfets help "focus" the distortion and prevent farting out, with the drawback that it sounds slightly static and boring.
- subbing the 1N4002 with red leds opens up the sound a bit, but leads to some burps and quirks at maximum gain settings.
- the circuit seems to be quite sensitive to lower frequencies. Turning the bass pot more than halfway on the HD preamp resulted in farting and blocking distortion.

Overall not bad for a quickie without tinkering too much and with less-than-ideal test environment.
I will play with the spice sims a bit more, to wrap my head around what each component does, and maybe breadboard the whole circuit later this week  thanks for the suggestion on the gain resistors!

[teemuk]

It's a simplified differential stage, and operated open loop* it will clip quite "softly". The opamp stage after FETs is current-voltage amplifier that converts differential's drain current to voltage output. DC feedback is applied back to differential, as in "servo" configuration for automatic biasing that highly variable FET characteristics probably require. *Any potential AC feedback is capacitively "shunted" to retain the differential in open loop condition.

Some of these G&K amps integrated the same soft clipping scheme to the power amp stage as well and had the opamp driving a common base voltage amp "interface". I think there were also some versions that employed very slight amounts of AC feedback as well.

But overall in summary, it's a "soft clipping" distortion circuit with a clever DC servo bias.

[antonis]

Quite right..

For effective AC feedback, it should need series resistors for C6 & C10..

[Umlaut]

Thank you, Teemu!
Further simulations and your explanations have helped me understand a bit more what makes it tick.
Soft clipping indeed - the differential stage starts clipping WAY before the diodes do. I assume they are there to avoid input clipping on the jfet.
As Antonis pointed out, R14/R19 control gain. Increasing headroom of the stage seems to be a matter of decreasing the resistors at drain/source.
Contrary to what I originally thought (oh, bias adjusting a la fetzer), the trimmers at drain only change the symmetry/asymmetry of the wave, but the voltage bias remains constant.
Reminds me a bit of the servo biased input stages of the Sunn Stinger & Stagemaster, which are on my "to tinker with" list.
By "capacitively shunted" do you mean by means of C6-C10?
Cheers to all again for the lessons! 

[antonis]

By "capacitively shunted" do you mean by means of C6-C10?

(on behalf of teemuk)

That's right..

R13 - R18 provide (actually connect respective op-amp outputs to differential pair inputs) both DC & AC feedback..
In case of resistors in series with caps, we have both kinds of feedback..
In the absence of those resistors (like here) we have DC feedback only 'cause AC is effectively grounded..

By raising U1 & U2 gain a lot (for reasons beyond the scope of present thread) and make C6 and C10 impedances comparable to R13 and R18 values respectively, you can make a frequency dependent gain circuit..
(approximately the ratio of resistor value and capacitor impedance)

[Umlaut]

I have tinkered with C6/R13 etc to change the freq response on the sims, looks quite versatile 

This afternoon I breadboarded the whole thing and spent the better part of two hours bent over it. I didn't have all the exact values so I had to get creative. Observations:
- 2N5484 & MMBF4393 - lotsa gain, ugly behaviour on the last 3trd of the gain pot rotation.
- J201 - nnno no no no. Yuck.
- Servo biasing on both stages needs to come from the SAME opamp as depicted on the factory schem. Splitting that over two opamps results in a weird blocking/motorboating hiss, whine and spurt.
- Drain bias at 12.6V is spot on, even changing the supply from 15v to 13.8v.
- 2SK246 - much less gain, but pleasant behaviour.
- 2SK30A-GR - less gain also, but closer to a "sweet spot".
- removing the first pair of diodes makes the distortion more lively and dynamic. Replacing them with leds decongests the sound a bit, but doesn't make heaps of difference.
- removing the second pair makes it come alive, rowdier arnd nicer. The fact that the 2SK30A has much more " input headroom" than the others might be helping.
- increasing the feedback resistor on the last stage (obviously) increases gain.
- not a fan of the voicing, definitely. A bit nasal ( "low nasal" as in Cyrano, not as in M.Jackson), especially when pushed with a TS.
- the double pot makes for such a beautiful, smooth transition from crystal clean to full gain.
- Speaking of full gain, this thing has 1st gen JCM levels of gain, especially with the 2SKs. It crunches but doesn't chug.
- The cleans are EXTREMELY NICE. Warm, full, etc.
Conclusion so far: as is, not super thrilled, but I will definitely spend some afternoons tweaking the frequency response and trying out different components, as it has a lot of potential I was looking for a nice preamp to replace the guts in my Frontman 212, which I loathe immensely, so this might be a nice challenge. Lots of fun.

[Jubz]

Hi

I really like these transimpedance gains stages and they became my favorite input stages. I didnt know the Gallien Kruger circuit looks really cool i will try it. If you want some more read on it :

https://sound-au.com/articles/jfet-design.htm

look at the 6.1 figure, its a single ended input but looks kinda similar

A while ago someone posted a dl link to the National Semiconductor FET Databook 1977. Go on it and page 230 you have a go on another variation (with few words on a differential inputs).

Its very versatile and you can use it with pseudo mu amp arrangement.

[teemuk]

IIRC, the original "Fetzer Valve" circuit (not the ROG contortion of it but the genuine one) featured the transimpedance scheme too.

[Umlaut]

Thanks for the links and info!
I was familiar with the sound-au site, but I hadn't read through that article. Interesting read indeed.
I downloaded the 1977 pdf, heaps of interesting stuff to try too.
Will potter around this kind of circuits in the coming weeks, see if something nice pops.