Fretting about FETs

R

rossbalch

Active member
*Renamed the thread as it is primarily about FET designs.

Hey folks, so I've been reading Merlin Blencowe's Designing Guitar Preamp book. One thing that has interested me is the effect of essentially a poor design leading to some current leaking into the grid of the follower and therefore leading to some signal compression.

This excerpt is especially interesting

Because the ECC83 is a high ra (anode resistance) valve, it is difficult to bias properly if the grid voltage is high and the cathode load resistor is not sufficiently large—as any classic textbook will indicate. If the supply voltages in the Fender Bassman had been much lower, or if the cathode load resistor had been properly chosen, this effect might not have been realized—and the amp may not have become the icon it is today. Similarly, if a lower ra valve were plugged into this same circuit, we probably wouldn't generate any quiescent grid current, and again, we would not obtain the characteristic compression effect. (Of course, this effect can be duplicated with any valve, provided the resistor values are suitably chosen.)
Click to expand...

He then goes on to mention that bootstrapping the stage as in the following example will exaggerate this effect (show at the end of the post). My question is, would it be possible at reasonable voltages (and by this I mean up to 24v I guess) to replicate this effect with a JFET or a Depletion MOSFET like the LND150? I think that would make a pretty nice addition to a "clean" boost that still helps control transients.

I did just get some 6n16b in the mail which I would be interested to in maybe using as the basis for a preamp. Maybe adapting the Vox Cathode follower stage to bootstrapping and using the 6n16b instead for lower overall gain. But then, how much is gain a factor in the amount of compression?




1750238803129.png

1750238941308.png
 
Last edited:
Just a hint: There is an old special tube called 12dw7, which one side 12at7 (ecc81) and the other side 12ax7 (ecc83).
Some years ago JJ released the new ecc823 and reversed version the ecc832. The latter one would be a great tube for a cathode follower circuit, which brings a ecc83 into a ecc81 (depending how the pins are wired in your amp. In most cases 6,7,8 are used for the first stage.)
 
Sounds like a good theory. IMHO definitely worthy of an LTSpice simulation or breadboard trial.

Be aware Q2 input is DC coupled from Q1 output. When Q1 is biased Q2 input bias voltage is also set.
 
I've breadboarded quite a few jfet circuits at 18v and 24v. JFet, BJT and Depletion Mosfet will all exhibit cathode follower compression/distortion, with some difference in regards to the tone/timbre. I found LND150 to sound a bit darker (or less harsh) than Jfet, with BJT being a nice middle ground. Cathode resistor value and the load that comes after will determine the degree of "compression". I did try bootstrapping at some point and didn't like the result, but I doubt I spent more than 10-15 min on it :P Mind the capacitor/resistor values, as you can end up with weird frequency boosts. As mentioned above, LTSpice is your friend!
 
Blackboarcult said:
I've breadboarded quite a few jfet circuits at 18v and 24v. JFet, BJT and Depletion Mosfet will all exhibit cathode follower compression/distortion, with some difference in regards to the tone/timbre. I found LND150 to sound a bit darker (or less harsh) than Jfet, with BJT being a nice middle ground. Cathode resistor value and the load that comes after will determine the degree of "compression". I did try bootstrapping at some point and didn't like the result, but I doubt I spent more than 10-15 min on it :P Mind the capacitor/resistor values, as you can end up with weird frequency boosts. As mentioned above, LTSpice is your friend!
Click to expand...
Have you ever done a whole preamp at high voltage, like a JCM800 or Rectifier type thing? I'm curious if there's benefits in running at higher voltage? It seems to me the initial work on these FET circuits was done at 9v just because that's the "pedal voltage", not necessarily because it was actually the best sounding or anything.
 
rossbalch said:
Have you ever done a whole preamp at high voltage, like a JCM800 or Rectifier type thing? I'm curious if there's benefits in running at higher voltage? It seems to me the initial work on these FET circuits was done at 9v just because that's the "pedal voltage", not necessarily because it was actually the best sounding or anything.
Click to expand...
Every single FET circuit I've ever played with has sounded a million times better at 18-24v.
 
Never tried higher voltages, too much faff. As drew points out, 18-24v sounds great :) And yes, 9v was initially done to keep the same "standard voltages" for 99.9% of pedals. Incidentally, J201s are one of the few jfets that sound great at 9v. 2SK208 (as seen in AMT circuits) also sound really decent at 9v, but I've tried them at 18v and they sound even better. Re-bias and overhauling the voicing components is a must, though.
 
Blackboarcult said:
Never tried higher voltages, too much faff. As drew points out, 18-24v sounds great :) And yes, 9v was initially done to keep the same "standard voltages" for 99.9% of pedals. Incidentally, J201s are one of the few jfets that sound great at 9v. 2SK208 (as seen in AMT circuits) also sound really decent at 9v, but I've tried them at 18v and they sound even better. Re-bias and overhauling the voicing components is a must, though.
Click to expand...

Do you have to measure the FETs IDSS and Vgs Off manually at the target voltage? I see the data sheets rarely specify much data after about 15v?
 
Also, I found this design interesting...

Mixing JFETs and MOSFETs. On the thread different circuits actually place the FETs and MOS in different locations in the circuit, though always a MOSFET into a JFET in a coupled "cathode" / source follower arrangement.

Why might you choose MOSFETs in this case? Is it purely texture, or are there also current / buffering considerations too do we think? There seems to be some attempt to simulate Miller capacitance here which is interesting.

Obviously this circuit has a lot more going on than the typical "conversions". But I'd love to understand more of the principles of what's going on here. For instance, the MOSFETs are Vr ground biased, but the JFETs aren't. Q3 appears to be biased in a cold clipper configuration just like V2b on a real Soldano, but I don't know how the match works out on the JFET end versus the AX7 math.

(https://www.guitarristas.info/foros/muestras-sonido-messa-soldano-krank-engl-vht/88517/pagina2)
1750567306285.gif
 
I tend to use jfets with tight specs (SMD J201, 2N5484, J113, 2SK208R, and the dmos LND150), and rarely bother checking their Vp or IDSS. By simulating on Ltspice, you can start with drain and source values that are in the ballpark, then check with the multimeter and do some small adjustments to get Vdrain between 1/2 and 3/4 of your voltage, depending on desired asymmetry. When dealing with cascaded gain stages, I found little difference in selecting gainier vs less gainy jfets.
I have seen that SLO emu before. For some reason I've never been very fond of N-channel mosfets for high gain, they sound noisy and rough to my ears. On the other hand, I have mixed jfets and depletion mosfets on several circuits successfully. Since the LND150 doesnt exhibit gate cutoff or blocking distortion like other fets and is not too gainy, it makes for a good first or last stage. For example, taking a Randall RG100 circuit, removing the cathode follower and replacing the clipping diodes with an LND gain stage biased more or less 1/2V has been an improvement over the original, at least to my ears.
Also, some jfets can be biased without source feedback (see some stages on the AMT Series 2), although even at 18/24v only a handful of models like J201 and 2SK208 allow it while maintaining decent gain and gate cutoff. To me, biasing it this way makes it sound closest to valves than anything else I've tried: harsh asymmetric clipping at gate, soft-then-hard clipping of positive lobe, and the negative lobe not quite reaching the rails due to the gate cutoff. Works beautifully with 2SK208 at 18v with a 10-12k drain resistor 🤘
 
Any idea why the MOSFETS have such high source resistors in this design? Is it to do with the Vr biasing and clipping point?
 
Do you mean the whole bootstrapped follower thing? I have never seen it in any pedal design, personally. There is an old thread on DIYSB about a similar idea using bjt, but I don't recall seeing it in any "complete" circuit.
I've ltspiced it with fets, it does increase gain and it does exhibit a different cutoff to standard follower (more progressive/softer), but then you kinda drift away from the sound we are used to (harsh clipping on negative lobe). Better or worse, that'll be subjective.
For "regular" topology with jfets, AMT M2 and S2 circuits, for example, feature a really well implemented simulation of the cathode topology from the original amps.
And yes, jfet into LND gain stages and viceversa sound good and I have no qualms mixing both in a circuit. I can share homebrewed stuff if there's any interest.
 
Blackboarcult said:
Do you mean the whole bootstrapped follower thing? I have never seen it in any pedal design, personally. There is an old thread on DIYSB about a similar idea using bjt, but I don't recall seeing it in any "complete" circuit.
I've ltspiced it with fets, it does increase gain and it does exhibit a different cutoff to standard follower (more progressive/softer), but then you kinda drift away from the sound we are used to (harsh clipping on negative lobe). Better or worse, that'll be subjective.
For "regular" topology with jfets, AMT M2 and S2 circuits, for example, feature a really well implemented simulation of the cathode topology from the original amps.
And yes, jfet into LND gain stages and viceversa sound good and I have no qualms mixing both in a circuit. I can share homebrewed stuff if there's any interest.
Click to expand...
Definitely interested in the home brewed stuff. I'm more interested in the "compression" effect of the cathode follower. With or without bootstrapping. I've been playing my Victory Cooper preamp over the weekend, and IMO the vox circuit does that compression thing really nicely, even when clean.
 
Blackboarcult said:
I've ltspiced it with fets, it does increase gain and it does exhibit a different cutoff to standard follower (more progressive/softer), but then you kinda drift away from the sound we are used to (harsh clipping on negative lobe). Better or worse, that'll be subjective.
Click to expand...


I think there is a certain application where this is a desirable outcome, not in the "amp simulator" aspect perhaps, but I'm thinking as part of a boost circuit you may want a bit of extra compression without too much harmonic content.
 
I did some sims just to compare side to side, here's some observations.

We have those three circuits, left to right:
- First one is a fairly standard jfet gain stage, with "cathode follower".
- Second one is the same, but with a follower nicked from AMT L2 pedals (those that have it, e.g. S2, D2, M2...)
- Third one has a bootstrapped follower.
All are biased equally, and the follower has a 26k load (akin to the AMT tonestack with bass and treble at zero, mids fully open).
Circuits.png

Taking the signal at gate, it is the same for all three circuits, since they are biased the same, no resistor on source and 11.2k on drain, featuring positive lobe clipping.

Gate.png

Standard circuit: gain stage inverts signal, positive lobe bangs against the positive rail on drain, and negative does not quite reach the negative rail, merely amplifying the previous gate clipping. At the output of the follower, we have clipping on the negative side due to the 26k load + a fairly high value on the follower source (47k).

Standard.png

AMT circuit: same as above, except that the AMT diode-to-source contraption compresses and prevents the positive side of the signal from reaching the rail, clipping 2v before it. Not knowledgeable enough to explain the theory behind it, but it does a fairly good emulation of what happens on the last stage of, say, a JCM800.

AMT.png

Bootstrapped circuit: we have more gain due to the bootstrapping (see the Merlin Blencowe chapter on it), and negative side clipping much earlier than the others, that "pushes" the negative signal (effectively soft clipping). The follower output looks quite smoothly compressed, and overall different than what you would see on the above circuits or a standard valve stage + cathode follower.

Bootstrap.png

If we reduce the source resistor, or increase the load, the negative clipping on the follower will decrease (cleaner signal). If we replace the load with a tonestack, the knob positions will affect the amount of clipping.

As for circuits, this one here is not where I would like it yet, but getting there. Kinda hot rodded marshall tones. LND150 works well as an input stage, since it won't ever feature blocking distortion even when slamming it at capacity, and works nice as a last stage too, clipping a bit softer than jfets and without gate cutoff distortion. Can be biased to clip quite symmetrically, and does a reasonable "power stage" clipping thing.

Screenshot From 2025-07-01 17-10-03.png

Hope this helps 🤘
 
Blackboarcult said:
As for circuits, this one here is not where I would like it yet, but getting there. Kinda hot rodded marshall tones. LND150 works well as an input stage, since it won't ever feature blocking distortion even when slamming it at capacity, and works nice as a last stage too, clipping a bit softer than jfets and without gate cutoff distortion. Can be biased to clip quite symmetrically, and does a reasonable "power stage" clipping thing.
Click to expand...
Can you post frequency resposne of this circuit? Thx :)
 
You must log in or register to reply here.
Back
Top