voltage drifting problems with j-fets

[kokodin]

hello, i wasn't sure where to put it so i decided to go back to basics because i am uneducated hack mechanic.

I am currently in process of redesigning of input stage, for a smal kit guitar amp. i promissed one kid from my school where i work that i build one, and i really didn't like the sound of it, so i decided to make it better. And now i am stuck with those j-fets.
it isn't really that complicated , but it kicks my back pretty hard because i don't know how to fight it and brute forcing a lot of biasing resistors might be the only option, or just a brainfart.

Two stages like this powered from 12V, separated with capacitors. 2nd stage has different resistor values to make it clip symetrically on both sides of a sine wave. But the darn thing runs away and swing the
output so it start to clip asymetrically in unpredictable way. Some place of the circuit is drifting . it is really a shame because it clips so nice and soft like a real vacum tube when it works fine, no sharp edges of noise.
I might work with capacitor values or double cap the stages and bias the mid point to a known state with high value resistors but i am not sure if that a good idea, and how would that affect the sound quality


The whole j-fet adventure is a thing for 2 reasons
1 end amplifier in bridge mode does not like to go into distortion so i would like to cap it at a known maximum input level for no distortion at the end, but handle the distortion before the signal hit the amp so it could play quiet but with overdrive already in it. or handle stompboxes if needed
2 i want to build a high gain stage with gain controll as a tube preamp simulation so it actually does have warm overdrive build in

i thought of a simple topology
buffer>>gain stage>>pasive tone controll>>level matching buffer>>vilume pot>>power amp

[TimFox]

The resistors in the circuit drawn should keep the DC operating point reasonably stable.
1.  Is there leakage (capacitor or whatever) into the gate terminal?
2.  Are you using electrolytic capacitors between stages?  If so, could their leakage into a high gate pull-down resistor cause the drift?

[jimmc]

If you're driving the second stage into clipping, then it is likely that the gate source diode is being driven into conduction on one half cycle.
This will alter the charge on the coupling capacitor and move the bias point as the signal varies.
Not sure how to get round this, maybe try a MOSFET instead?

Jim

[TimFox]

If you drive the gate-source voltage on the second JFET to forward bias, then the effect on the gate voltage waveform is to charge the coupling capacitor to clamp the positive excursion of the gate voltage to the source voltage (plus diode drop).  A misnomer for this effect is "DC restoration".

[kokodin]

jimmc might be up to something, i was driving both stages into cliping on the negative side pretty much equaly. 2nd stage being unity gain, first i think was amplifying. signal swing on the input was around 3V rms if i remember corectly (my analog scope does not have meacurement i used multimeter) and yes it was going off center much more  the more overdriven it got, then swing back to center, but not always, might be the breadboard thing.

I might overreacted though what i used to test it was way too hot generator signal out of another breadboard monster. it did use separate power suply but they might afect themeselves in some ways
i didn't try it with all the stages yet but with guitar instead of a generator it behaves more stable, even though voltage gain was done with npn transistor stage that was clipping itself  slightly.

Besides this is a budget fast prototyping, i just throw stuff i have at hand at it, not really wiling to buy different components that may or may not fix problem i created.
i have bucket of old parts from way back in highschool and i am trying to make use of them  in the best possible  way

If you drive the gate-source voltage on the second JFET to forward bias, then the effect on the gate voltage waveform is to charge the coupling capacitor to clamp the positive excursion of the gate voltage to the source voltage (plus diode drop).  A misnomer for this effect is "DC restoration".

I feel like i need to read that multiple times and still not know what i just read. Sorry english is not my first language and  textbook term only make me google more stuff into rabit holes. Could you explain it like to a particulary dumb person who don't understand complicated words? , joking asidedc restoration is a term i never learn anything about, i might used it without knowing about though.

the idea was that jfet would close  when it get  the signal that  bias the gate on negative swing below 0 then signal will get fliped upside down and the other stage would do the same on the popsite side

[kokodin]

I played a bit more with resistors  on a breadboard torturing single jfet, and  well amplification is about 1,1, but it clips simetrically inside just one transistor, and the best part is i am dc coupled  with the scope and my sine stay in the middle of scope screen
for bf257a jfet that is 12v power suply source at 2,3v, drain at 7,6v and since this j fet is a max 20mA drain current it will like working with 1,6mA quite happyly
values for the resistors were picked at random untill i was happy but i based my search of a tube simulating preamp from rugoffgruves.com so i started with 1,5k source resistor and worked my way down with gate resistor to around 2,4k. it is not a front end stage mind you just a distortion stage since it requires input voltage of 1,9Vrms (2,7 Vp if my math is right)to start cliping  and guitar gives 10 times less.
In perfect word i would just have to build a buffer and npn stage as a drive section and feed it into that, with overal gain of around 20 times the input, and i might clamp it with zeners at around 6Vpp, anyway Max swing from npn common emiter would be around 7,5V (2,65Vrms) with 12V power suply so i think it will be doable.

i think it will get more complicated that it needs to be  but i might actually make it work :]

edit
I cobled together an intermediate stage to produce voltage gain, as stated before i used npn transistor. Initially bc547c but i found some bc550c so i switched to lower noise version. diference is minimal.
After playing with biasing this stage i got pretty crazy looking voltage gain, and i am still a bit shocked how much this transistor can swing before it start distorting. because it produced 4Vrms from 12V power suply, it almost swing rail to rail. IT was biased so it  touches both sides of cliping  at the same time, the only problem with npn is, it want to go beyound that 4 volts in rather wierd way, so after hitting the wall signal changes into 50% duty cycle pwm signal, where any aditional input signal strength makes it asymetrical like changing the duty cycle. Still it looks awesome on a scope screen and even the kit power amp sounds good.
Doesn't go much into overdrive with guitar signal though, need to tweak front end to more than 1.1 gain and it be perfect (well as much as something like that can be)
i also find out why kit amp sounded so distorted. in it's preamp stage they used npn bc550 like i did only it was either typo or bad design because in the same topology they biased base to ground with the 470k resistor, for me it ended up being 47k exactly all the rest of components being the same.

[kokodin]

hello again
i kind of starting to finalise my design and it seems like i started with one project  http://www.runoffgroove.com/uno.html made my own changes and ended up pretty much with a different very similar schematic.https://www.open-electronics.org/guest_projects/1wamp-open-hardware-1-watt-guitar-amplifier/
That being said they all use j201 jfets and 9V and i ended up using bf245's and 12V and my bias voltges are pretty much proportional  (to uno's) so the gain might or might be the same as the runofgrooves uno project
it sounds good enough anyway after lowering the gain of the final stage

the question that i have now is why in the uno schematic drive potentiometer is wired "backward"? wouldn't that make a low impedance or short way to ground making the transistor basically bypased at the very end?
i kind of intuitively wired it conventional voltage divider way and that make more sense to me , even that 220k resistor make sense because it being in paralell to the pot  regardless where the potentiometer is transistor driving it should see 220k-1m load all the time with pretty much 200k input impedance of the next stage all the way down to last quoter turn

so why is the uno drive  potentiometer wired this way?

anyway i ended up with 2 jfet stages separated with gain atenuator between them and tone controll between the amp and preamp
amplification on jfets being pretty much 12 times on each stage runing them from rail to rail with strong guitar signal.
first stage cliping minimal amount of the peaks, 2nd stage pretty much going into full overdrive while power amp being overdriven only slightly at the very end  making up for volume loss if you turn all the tone to bass all the way (marshal ms2 tone stack kind of cut it at 80hz so all the top end is gone when turned to all bass and treble side bring it all back)