BJT vs MOSFET in guitar amp

[EricKnabe]

Hi. I'm very much a beginner with these things so please bear with me. I'm planning on building a simple guitar amplifier using discrete transistor stages. I want it to somewhat emulate Marshall type of sound. My question is, which type of transistor should I use for my gain stages, BJTs or Mosfets?
From what I can tell for BJTs, it seems like they are simpler, more general purpose, and much easier to source. Mosfets seem to be more complicated, with more specialized uses and therefore much harder to source. Despite being the most common transistor. (Mosfets used more in ICs with BJTs having more discrete applications and better for hobbyists maybe?)
Then on the flip side, however, I'm concerned about a few things with BJTs. For one, could the lower input impedance of BJTs cause tone loss? And will the clipping of the BJTs be fuzzy or harsh? (Mosfets maybe more "tube" like so to speak?) And would the higher gain of the BJT cause too much distortion if biased similarly to a mosfet stage, or do the BJTs have higher headroom as well?

Finally, it seems like Mosfets are generally optimized for use in computers and not amplification. (Going back to being hard to source.) Would the threshold voltage of the Mosfet be a problem in the input stage of the guitar amp circuit, where a guitar's output can be lower than 1 volt?
And then would the threshold to turn on the BJT be any better?
Most SS guitar amps seem to be made with BJTs or Opamps that use mostly BJTs, and any one talking about electronics uses BJT as the default transistor despite Mosfets being much more common in the modern age. This confuses me. Does it go back to the general vs specialized purpose thing?
Which should I use in my design?
Thanks!

[GibsonGM]

Welcome to the forum!    I'm a Mosfet junkie.  There are many sources of Mosfets, they are not hard to come by at all where I am (USA).  I think you should do some more 'looking around', asking people about tone/sound quality, check out Youtube...there are a lot of good Mosfet based designs that DO, IMHO, grab that Marshall tone pretty darn well.  BS170/2N7000 - based designs are common.

Are you planning on building the WHOLE amp (power amp too) or just the preamp, AKA distortion circuit/boost....?    LOTS of info abounds on preamps.

Something like the "RAH" (Royal Albert Hall, tribute name to Led Zeppelin) may be a good place to start from. I like this one.  http://revolutiondeux.blogspot.com/2012/05/catalinbread-wiio-rah.html
This is simply several cascaded gain stages - pretty much how any amplifier preamp works.


Just to not play favorites...here is a very popular JFET based design.  There are many of THESE type, too: http://www.generalguitargadgets.com/pdf/ggg_bsiab2_sc.pdf


Do some snooping, you'll find something that attracts you, maybe build a couple of different ones to find out how they work, and pros and cons of them! 

[EricKnabe]

Thanks for your input. With an enhancement Mosfet like that, how do I ensure it's properly turned on? Guitar's output such a small voltage, they wouldn't come close to the required voltage on the gate. Or am I way off base in terms of how enhancement Mosfets work?

Also are jfets closer in practice to Mosfets or BJTs? Thanks.

[marcelomd]

AFAIK, BJTs are cleaner, less noisy. But MOSFETs distort in a pleasant way.

The guitar signal is strong enough, don't worry.

Look into the Zvex Box of Rock for a full MOSFET pedal. You can start with it as a preamp.

[EricKnabe]

So again, sorry if I'm not understanding something but what ensures the transistor (bjt or mosfet) is turned on? Are there circuit elements that do that or is the guitar's output just enough?

[ElectricDruid]

So again, sorry if I'm not understanding something but what ensures the transistor (bjt or mosfet) is turned on? Are there circuit elements that do that or is the guitar's output just enough?

There are circuit elements that do that. The magic word is "biasing". The transistors (of whatever type) are biased so that they start off from a position of being partly-on, so that the negative peaks of a signal make them a bit more turned off, and the positive peaks make them a bit more turned on. If you *don't* want distortion, you have to ensure that your biasing puts the signal in the middle of the most linear region of the transistors response curve. If you're aiming for some distortion, you have some more flexibility and can get a bit more creative!

[swamphorn]

The goal of biasing is to ensure that the transistor is turned on over the entire range of the input signal. For most circuits this means maintaining the output voltage at approximately half-supply when the input voltage is zero. The "Transistors" chapter in The Art of Electronics is a must-read¹. Analog Devices also has an excellent series of articles on electronics theory; here is their chapter on single-transistor amplifier stages².

I haven't had the chance to work with discrete FETs, but one of my favorite chips to design around is the CD4007. It has three complementary pairs of MOSFETs. All three may be used as unbuffered inverters, and the first and second as common-source/drain amplifiers. There are other creative ways to use the '4007 (some pedals use the chip as an all-in-one soft bypass solution, even); I can share my findings if there's interest.

The '4007 does have several disadvantages. First, the '4007 isn't specified for linear applications; the individual FETs are very poorly specified, in fact³ and are probably noisier than discrete devices would be. Second, the individual FETs have relatively terrible (and again, unpredictable) gain. A common-source amplifier with no emitter degeneration and a 10 kΩ resistor only gave me approximately 25 x gain. The maximum gain is about 100 x with a load of about 1 MΩ (that said, I've read that dismal gain figures are pretty typical of FETs). Finally, the internal connections of the FETs may be excessively constraining.

That said, the '4007 has the potential to make an excellent-sounding guitar preamp. For example, chip would be well suited for an architecture like that of the Alembic Tube Preamp: two gain stages with an equalizer network in between (the inverter could be used as an active Baxandall, for example). There are some other niche roles the chip might fill well too; I'd expect the FETs to be matched well enough to use as the variable resistance elements in a phaser, and the substrate can be biased such that the body diodes wouldn't be a concern.

¹ https://www.pearl-hifi.com/06_Lit_Archive/14_Books_Tech_Papers/Horowitz-Hill/The_Art_of_Electronics.pdf
² https://wiki.analog.com/university/courses/electronics/text/chapter-9
³ The SPICE models I've found online give threshold voltages of 2.0 and -1.7 for the P- and N-channel FETs, respectively, which matches my observations--at least for chips made by TI.

[thetragichero]

my lab series l3 used a mixture of 4007s and bjt op amps (although i swapped them with tl062s because they distorted better) and i dig the sound. probably more complex than you're looking for
any reason you're going discrete vs ic?

[antonis]

I'm a Mosfet junkie.

What a pity..!!   
(I'd just about grown to like you, Sir..)

@EricKnabe: A mixture or BJTs (for any stage other than last one) & MosFets (for last Source follower stage) should be interesting..
(amplified signal, up to distortion, could be more easily manipulated for BJTs rather than MosFets - you can take the advandages of the later for power stage..)

Douglas Self (among others) deals in depth, though with remarkable simplicity, with that task:
https://epdf.pub/small-signal-audio-design.html
https://epdf.pub/audio-power-amplifier-design-handbook58585b1e8538616db8ff33e83647d0aa26964.html

[GibsonGM]

I'm a Mosfet junkie.

I like them all, Antonis!  As a device snob, I feel that BJTs work well for a classic, 'woody' LPB-1 type boost or if overdriven, fuzzier distortion.  For just a simple boost for guitar, the truth is that I prefer BJT, I like the way it colors the sound...mosfet can sound sterile and cold!

Center-biased Mosfets in cascaded stages seem to me to distort more like triodes, but that is only my opinion - there is little data to back that up.  They remind me of solid-state analogues of triodes and I don't have to come up with 300V and 12V to use them...

What is done to shape the tone within, and following,  a gain stage may be much more important than what actual device is chosen, and how the device is biased is going to make a huge difference on distortion characteristics   

YMMV

[antonis]

@EricKnabe: As you might have suspect, your query isn't of the single thread page settlement kind..

[Steben]

I think of 3 elements.

1. linearity

MOSfets tend to be more linear than BJT well below the clipping treshold in any non-feedback circuit.

In fact, BJT are the least linear of the three semiconductors. In a single simple transistor stage they add the most harmonics (think of the rangemaster for example). Yet they are the highest gain devices, which means they are the number 1 building block of opamps. Opamps need a high open loop gain which get "tamed" by feedback, which lowers the non-linearity etc.
In essence, BJT are preferred to build linear circuits, but they are very non-linear components.

2. clipping

MOSfets tend to clip very softly when biased in that aspect. That's why they are found in some distortion devices (CMOS buffer based, box of rock, ....).
Jfets clip softly. BJT tend to clip very asymmetrically, usually hard.

3. noise



MOSfets usually exhibit more flicker noise. This means they are LESS interesting for high gain / preamp circuitry (low input signal) than BJT or jfet because the ratio of noise vs signal is worse. Simply because the ideal configuration for low noise is having the highest possible gain in the beginning of the chain with the lowest noise content. Otherwise you amplify the accumulated noise along the path at the end.
They are very stable as power device. That's why they are / were used a lot in power amplification. This is / was mostly in class AB with source followers. This means the voltage gain is done before the MOSfet (higher input signal to noise) and then the MOSfets do the current gain.


Conclusion:
MOSFETs are very soft clippers, but are put in the shadow by jfets and BJT for low power, preamp circuits. Thermally more stable than BJT. Very sensitive to static electricity. Best used as power amplifier.
jFETs are soft clippers, but are put in the shadow by BJT and MOSFETs for power circuits. High input impedance. High tolerance, which means biasing is very component sensitive. Best used as input stage.
BJT are hard clippers, but they offer the best gain per device performance. Biasing is less component sensitive. Used both in preamp as power amp devices.
OPAMPS are "complex circuits in a small box" which can be used as almost ideal amplifier building block. They are very linear, can have low noise, high input impedance, can drive high loads, etc.... but they are far from ideal to have "tone" when driven hard.

[jonny.reckless]

I think BJTs, MOSFETs andf JFETs can all sound "good" in the right circuit. The trick to getting a great overdrive sound, in my experience, is to have 2, 3 or 4 cascaded gain stages, roll the low frequencies off early in the signal path (cutoff somewhere between 200Hz and 1kHz is typical), have asymmetric clipping, DC bias point shifts, and roll off the high frequencies late in the signal path (cutoff anywhere between 700Hz and 3kHz is typical). This can be achieved with many different types of circuit topology. Some examples of guitar preamp circuits I happen to like the sound of:

BJT: Peavey Bandit Transtube
Op-amp / LEDs: Marshall MG100HDFX, Guvnor pedals
JFETs: Boba FET, Rosie amps (shameless self promotion) 
https://www.diystompboxes.com/smfforum/index.php?topic=118627.msg1104856#msg1104856
https://www.diystompboxes.com/smfforum/index.php?topic=123572.msg1169169;topicseen#msg1169169

I've actually grown to prefer the sound of a JFET based preamp to many ECC83 tube based gain stages for high gain circuits. Tubes still sound best to me in the lower gain Plexi / JCM800 type amps which are actually pretty low gain at sensible volume levels. I'd much rather have a distortion pedal into an old 3 stage Marshall preamp than a modern high gain all tube amp sound like a 5150.

JFETs are easy to work with since they are depletion mode devices, you can self-bias them like a triode common cathode gain stage. They also operate on a square-law transfer characteristic (triode is 3 halves law) and only conduct gate current when you forward bias them, so they behave a lot like a triode in a typical guitar amp circuit. I've been using them in guitar amp designs for about 30 years with pretty good results. You just have to beware that VGSoff and transconductance varies a lot more than with tubes, and either select for these properties or design your circuit to accommodate them. My pet favorite for guitar amps is the J113. You need at least 18V supply rails (preferably 25V or more), and they sound great with 4k7 drain resistor and 1k source resistor. Make sure you have at least 100k gate stopper to stop farting out if you use a source bypass cap (I use 10uF).

ROG has loads of pedal designs using the JFET as triode approach, most of which sound pretty good to my ears. I really like their Thor: http://www.runoffgroove.com/thor.html

For a power amp, BJTs are hands down better than MOSFETs. I recommend Doug Self's audio power amp design handbook for a good explanation of why that is. That said, guitar amps are strange beasts, and if it sounds good, then it is good.