Working on a circuit that uses JFETs and am trying to convert to MOSFETs

[bushidov]

So I have this circuit I am working on:


Nothing special, but something I am trying to figure out on a breadboard. So, firstly, this is based off another circuit but Q2 and Q3 are JFETs, specifically, 2N5457s and with them, it sounds pretty good. So, my first move was to simply change out the 2N5457s and hear what it sounds like with BS170 MOSFETs. It doesn't sound "bad", but the gain dropped quite a bit and it sounds like they are not "biased" properly. Like, it sounds like a circuit with JFETs where you would have trim pots and it sounded like the trim pots were not set correctly. So, it "works" but not works well. Which is why I am guessing that it isn't biased properly?

Am I wrong?

Also, if I am correct, which resistors in this schematic need to be changed out for trim pots? I would guess R11 and R12, but I am not sure.

[Rob Strand]

The biggest difference between MOSFETs and JFETs is JFETs require the gate to be more negative than the source whereas MOSFETs require the gate to be more positive than the source.

JFETs can self-bias using source resistors and connecting the gate to ground.  That won't work at all for MOSFETs.

As far as the circuit "shape" goes  the gate bias for MOSFETs looks a lot the BJT base bias circuit.  The only difference is the MOSFET doesn't pull any current on the gate.

So for your circuit, the top MOSFET will probably work but the output swing might be helped by reducing R8 (or increasing R9) to raise the gate about 2V above Vcc/2.  For the lower MOSFET, following BJT biasing pattern,  add a resistor from where R6 and R7 join to Vcc.  You'll probably need something like 1.5MEG.

Feedback biasing often works best for MOSFETs but it's probably going to mess things up on that circuit.

You can play around the bias point on both MOSFET to suit your ears.    MOSFETs will probably sound a little darker than JFETs so you could compensate by reducing the 33k in the gate of the lower MOSFET.

[bushidov]

So, something like this?


I reduced R7 from 33K to 22K to compensate for the dark sound. I may go lower to around 10K. I will try this out tomorrow morning. Makes sense and seems simple enough.

[Rob Strand]

So, something like this?

Yes, spot on!

[Rob Strand]

Oh, you need to flip the polarity of C7 now.

[bushidov]

Oh, you need to flip the polarity of C7 now.

Yep, got that now that the voltage divider favors in that direction.


This sounds pretty good and close (not exact) to the JFET version. I also reduced R7 down to a 10K. It seemed to sound a little better. I also had some 1.2M and 470K resistors which got me a lot closer to that ((VCC/2) + 2V) mark.

So, the last part is I added an end stage to this circuit


This again was supposed to be a 2N5457 JFET that I just swapped out for a BS170 MOSFET. It seems a little "sputtery". Do I need to change the R25 and R26 to a ((VCC/2) + 2V) voltage divider?

[antonis]

Do I need to change the R25 and R26 to a ((VCC/2) + 2V) voltage divider?

http://www.geofex.com/article_folders/mosboost/mosboost.htm
(in the middle of the page, R.G. deals with MosFet Source followers..)

P.S.
IMHO, Q6 Source should be biased at a point higher than Vcc/2, due to limited current sinking capability in relation with current sourcing one..
Particular bias point could be estimated by voltage dividing action of R7 & whatever load Q6 drives..
(in case of load value higher than 10 X R7 say or your are not a perfectionist, just ignore the above said..)

[Rob Strand]

This sounds pretty good and close (not exact) to the JFET version. I also reduced R7 down to a 10K. It seemed to sound a little better. I also had some 1.2M and 470K resistors which got me a lot closer to that ((VCC/2) + 2V) mark.

Cool.    Since you ended up with R6 and C10 R10 and C6 in there, you might want to try Feedback-bias after all.  What you would do is,
- Remove the added R13
- Remove C10C6, and replace with a link
- increase (or adjust) R6 R10 to get the same drain voltage on Q3 that you have now.

It achieves a similar goal but saves a cap and a resistor.  Also, the feedback bias is a little more predictable than the current method.

It's probably fine like it is but I thought I would mention it if you want to play.

[bushidov]

C10

I don't think I have a C10 on that schematic.

[Rob Strand]

I don't think I have a C10 on that schematic.

Sorry, it's R10 and C6  I was talking about.

[bushidov]

Alrighty. I had some time to go back to this project and expand one more step further with it.

- Remove the added R13
- Remove C6, and replace with a link
- increase (or adjust) R10 to get the same drain voltage on Q3 that you have now.

Did these things. R10 needed to be at 2.2M to have the same voltage, but it was sputtering terrible. Apparently 4.85V is not a good spot for that. Dialing it down to 1.2M, it was now at 4.45V and that sounded pretty good.

I changed R25 to 470K and R26 to 1.2M and that further improved the sound a lot.

Now for the last piece of my project, adding the middle MOSFET stage


I pretty much guessed that I needed to get R16 and R17 at the same values as my other biasing resistors, as the JFETs wanted a pair of 1M, but these MOSFETs want a 470K and a 1.2M. So I did that. To follow what Rob had me do on the first dual MOSFET stage, I would have "removed C6 and replaced with a link", but there wasn't an equivalent one to remove and replace with a link, so I ignored. There also wasn't an "R10" to change and adjust so I ignored that too.

I plugged it in, and without the Gain knob close to maxed, I don't get any sound. At maxed, it is very sputtery.

I am guessing I need to do something like the C6 and R10 fix to this second dual MOSFET stage?
I am guessing lowering the values of either R14 or R15 (or both)?
Are these dual MOSFET stages Mu Amps? For some reason I don't think they technically are.

[duck_arse]

you want that R10 in at least, to bias the lower of the added mosfets. and a cap, cause there is a DC path from gate thru gain to ground.

SRPP ?

[antonis]

SRPP ?

Plain or with Aikido Mojo..??

[bushidov]

Ooops, also forgot to mention that I flipped the polarity of C11 to be correct (and forgot to update my schematic, but C7 as well, per Rob)

you want that R10 in at least

I am not understanding. I kept R10 in as a 1.2M resistor. Or are you referring to the second dual MOSFET stage and putting in it's "R10" in there?

Originally, I didn't have a resistor in there, as my schematic shows, but in that case, it was working fine when Q4 and Q5 were 2N5457 JFETs. Obviously, JFETs do not equal MOSFETs, which is pretty much the exercise of this circuit for me in figuring out how to "translate it over".

If I am to add a resistor to that second set of MOSFETs, am I putting one between the where C11 meets the drain of Q5 over to where R15 meets the gate of Q5? What values would I be looking at? Also, someone mentioned an additional capacitor? Where and what value?

[Rob Strand]

Quote

    - Remove the added R13
    - Remove C6, and replace with a link
    - increase (or adjust) R10 to get the same drain voltage on Q3 that you have now.


Did these things. R10 needed to be at 2.2M to have the same voltage, but it was sputtering terrible. Apparently 4.85V is not a good spot for that. Dialing it down to 1.2M, it was now at 4.45V and that sounded pretty good.

I changed R25 to 470K and R26 to 1.2M and that further improved the sound a lot.

The things I suggested were more along the lines of something to try - upfront they weren't better or worse than what you had.   Like you found you might have to play around a bit to make both cases as good as they can be.   If there's only a fine window where the new circuit sounds good then maybe the biasing of the top MOSFET and bottom MOSFET are fighting each other, in which case it might not have been a good idea in the first place.   But if it sounds better then so be it.

I plugged it in, and without the Gain knob close to maxed, I don't get any sound. At maxed, it is very sputtery.

I am guessing I need to do something like the C6 and R10 fix to this second dual MOSFET stage?
I am guessing lowering the values of either R14 or R15 (or both)?

There's a number of problems to be solved on the second stage.

So the big problem is the gain pot is is stuffing up the DC bias on the gate by shorting it to ground.  That works on a JFET but it will kill a MOSFET circuit bias.   There's no option but to add a cap in series with R14, on the gain pot side.

The next problem is to fix the biasing of the lower MOSFET.   There's a two main options: 
- The first option is add a resistor across the drain and gate of the lower MOSFET making it like the first stage.
- The second is to add a resistor from the gate of the lower MOSFET to Vcc.    Essentially forming a divider on the gate with R15.

There's two remaining problems related to R14 being a large value.

The current circuit has a voltage divider constructed from R14 and R15.  When you add a resistor to R15 to fix the bias it's going to load the divider down a bit more.   However, unless you are doing some fine circuit tweaking it's not possible to say what's best.  So just add the resistor and don't worry about the loading for now.

The large value of R14 is going to make the MOSFET pedal sound a low darker than the JFET.    One solution would be to simply reduce R14 until it sounded good.

These fixes will definitely kick the circuit in the right direction.  The thing is if you wanted the MOSFET version to sound like the JFET version you would be up for a lot hours of tweaking.   You might even need to add a reverse diode to the gate and source of the MOSFET.   The other way to spin it is, treat the MOSFET version as a new beast and tweak it until it sounds good.

As far as tweaking goes, you might look at some circuits on the web which MOSFETs in high gain stages.  Particularly "amp emulation" pedals which have a gain pot with a following MOSFET stage.   There's been a lot of these over the years both DIY and commercial.

Are these dual MOSFET stages Mu Amps? For some reason I don't think they technically are.

Yes.

[Rob Strand]

I forgot to mention another way of counteracting darkness with large resistors in series with the gate is to add caps in parallel with it.

[bushidov]

The second is to add a resistor from the gate of the lower MOSFET to Vcc.    Essentially forming a divider on the gate with R15.

There's no option but to add a cap in series with R14, on the gain pot side.

The large value of R14 is going to make the MOSFET pedal sound a low darker than the JFET.    One solution would be to simply reduce R14 until it sounded good.

So, going that route, something like this?


Yeah, I will be making it a "different beast". The primary idea was to make a "preamp" to sound like Tony Iommi's RangeMaster into a Laney type sound. I've heard this setup with JFETs and it nailed the Master of Reality/Vol 4 sound, but with JFETs going the way of the dodo, I wanted to see if I could capture a similar sound, but with MOSFETs as they look like they will be around for a bit longer.

[Rob Strand]

So, going that route, something like this?

Yes, looks good.  You might find you need to raise R28.

Yeah, I will be making it a "different beast". The primary idea was to make a "preamp" to sound like Tony Iommi's RangeMaster into a Laney type sound. I've heard this setup with JFETs and it nailed the Master of Reality/Vol 4 sound, but with JFETs going the way of the dodo, I wanted to see if I could capture a similar sound, but with MOSFETs as they look like they will be around for a bit longer

It's a very tough task matching a sound.  You can spend many hours.   A lot of guys who have done this in the past tended to copy the whole signal chain except change the Tubes to JFETs (see JCM800 and Boogie emulators).   The only difference is you don't want to copy the tone controls (and some of the voicing) as your amp already has those built in.   So the tone controls are more about tweaking the sound and rolling the highs (like you have done).

Eg, but this one keeps the tone control,
https://guitarpedalbuilders.blogspot.com/2013/01/jcm-800-emulator-pedal-marshall-amp.html

I suspect the MOSFET thing will come-up more and more over time.   To me, the MOSFETs never quite sound like JFETs.   Mixing BJTs and MOSFETs might be a way to get closer since the BJTs are more fizzier and the MOSFETs more rounded.  The hope being you can balance things out a bit.

[PRR]

> MOSFETs never quite sound like JFETs

Grid conduction is part of most distortion. MOSFETs don't do that.

[POTL]

Sorry to speak off topic, but I want to insert my 2 cents.
I think it makes no sense to create a Mu stage based on two mosfets, but it is worth using a standard class-a amplifier circuit.
Several years ago I tried different combinations, including what you want to build, and I came to the conclusion that single transistors are better.
1) They are less noisy.
2) The gain is less and this allows you to use more stages and more accurately tune the sound (mu stages in the amount of three or more sound terrible).
3) Dynamics is better
4) The sound is more spacious and transparent