Author Topic: multi stage jfet amp, coupling caps range from 4.7n to 680n. Sloppy design?  (Read 357 times)

0 Members and 1 Guest are viewing this topic.

Boner

So Im in the process of designing a multi stage jfet preamp/buffer and in between each stage I have coupling caps..... of course.

Doing things in SPICE first, I trial and error'ed each stages cap for -3dB of gain at 20Hz or so one at a time and they range from 4.7n all the way to 680n. In all other designs Im seeing 4.7u to 47u

Am I doing something wrong?

*edit*

Do I have to answer how many knobs a certain guitar has or who makes what specific guitar everytime I post or edit? :icon_redface:
« Last Edit: December 04, 2017, 03:00:10 AM by Boner »

antonis

First of all, Wellcome..!!  :icon_lol:

Let's now try to understand what are you talking about..

Why you want cut-off points (-3db) at 20Hz and with what resistances are created those High Pass filters..??
(I presume HPF due to caps coupling placement..)

IF they create RC filters with JFETs input resistance, their capacitance value is actually set by JFETs biasing resistor(s) BUT you have to oversize them due to cascaded filters stage loading..



<Do I have to answer how many knobs a certain guitar has or who makes what specific guitar everytime I post or edit?>
You're doing something wrong with your log-in..
Tired of updating my account to enable 3rd party hosting..

duck_arse

also welcome. how many knobs does a certain guitar have? are you in Russia?

to your caps size question, as antonis is fond of saying:
Schematic???
".... the Victoria’s Secret of pedal underwear" - Phillip H
"Who?" - Steven Toast
"No way you cna go wrong" - Juan Wayne
is the duck allowed to walk along the top of the fence?

thermionix

<Do I have to answer how many knobs a certain guitar has or who makes what specific guitar everytime I post or edit?>
You're doing something wrong with your log-in..

I think it's just a first post thing, to combat spam.

Boner

Thanks for the welcome and I've attached a schem!

The first cap (C3) sets a nice -3dB point to block everything below 20ish Hz. The other coupling caps (C6, C9, C7) I made 47u to block DC just for now. These are the caps I was talking about from before.

To figure out what cap values these should be, I made all of them HUUGE (including C3) so they didnt have any influence at 20Hz. I then played around with each, one at a time to get -3dB at 20Hz at the final output again while all the others were huge (like 820u). Once it was figured out I set it back to a huge value then moved on to the next. So my plan was to set each cap to these -3dB values, but then increase them all by around 20% or so until I have a final output of -3dB at 20Hz....
I hope that makes sense

I figured this would be "best" to block low frequencies well

Or am I dumbass and should I set C3 to have -3dB at 20Hz and just make the rest 47u or so like in the schem? I thought if I did this, the circuit wouldnt perform as well since frequencies way below 20dB but slightly above DC wouldnt be blocked as well and maybe screw with any settling times or biases


*edit*

I still have to answer these verification questions!! ERrrggggg....
« Last Edit: December 04, 2017, 07:33:51 PM by Boner »

PRR

> verification questions!! ERrrggggg....

We had a BAD rash of spam recently. Admin may have set the bar higher.

47uFd will be an electrolytic which will be very leaky relative to a 6Meg resistor. Works in SPICE, not (for long) in real life.

You have four C-R networks. If all four are scaled to -3dB @ 20Hz, the result is -12dB @ 20Hz which is about -4dB @ 80Hz, low guitar.

Leave the first one near 20Hz. Set the others around 2Hz, you will be fine. C6 may as well be 5nFd like C3. C9 drives >1Meg when clean, but much less when diodes clip- this may have to be decided by ear. Output impedance should be "low" through the 50/60Hz power frequencies, but 10uFd would be fine at C7.

Oh, and C10, which sees 1.2K and 120r+rs, which may be ~~1K. A full-bass solution might be 10u or 20u. Marshall made a boatload of amps with a 0.68u in a similar position, to give a "shaved bass" which emphasizes guitar's tenor range rather than baritone.

Remove C5. C5+C2 split power supply crap in half. C2 alone would give tremendous reduction for  buzz. Put your C5 pennies at C8, and run ALL three stages from there, not raw power. Yes, source follower can have good PSRR, eXcept your bias strings bring power crap right to the gates.

AHH. You are using a floating battery with soft +/-4.5V rails. In guitar, fashion has turned to one 9V power supply for the whole pedal board, negative common with signal return, so circuits are designed around zero and +9V.

Boner

Thank you so much!!!


Quote
47uFd will be an electrolytic which will be very leaky relative to a 6Meg resistor. Works in SPICE, not (for long) in real life.

Would it help to make that resistor a 1 watt? Or is it best to reduce the 6M resistor to around 3? I dont see why reducing it to 3M would hurt.


Quote
Output impedance should be "low" through the 50/60Hz power frequencies, but 10uFd would be fine at C7.

If I understand correctly thats to help alleviate hum.... How low should the impedance be?


Quote
AHH. You are using a floating battery with soft +/-4.5V rails. In guitar, fashion has turned to one 9V power supply for the whole pedal board, negative common with signal return, so circuits are designed around zero and +9V.

I wired ground this way so that later on I can throw in a waveshaping circuit Ive been working on in between the first buffer and amp... This circuit was powered with -4.5 and 4.5 with a reference point of 0, went ahead and made "ground" 0

I dont see why I couldnt change it to be powered from 0 and 9 and made the reference 4.5

IIRC I read that op amps like to be powered with single pole power supplies anyways, like 0 and 9 and not -4.5 and 4.5. Is this true? I figured the op amp wouldnt care or even notice anyways but what the hell do I know

antonis

Would it help to make that resistor a 1 watt? Or is it best to reduce the 6M resistor to around 3? I dont see why reducing it to 3M would hurt
Resistor wattage has nothing to do with it..
If you want to keep the same RC time product, you have to lower cap value and raise resistor value..

Quote
If I understand correctly thats to help alleviate hum.... How low should the impedance be?
For cap impedance, use 1 / (6.28*C*f), here C in Farads & f in Hertz

Quote
IIRC I read that op amps like to be powered with single pole power supplies anyways, like 0 and 9 and not -4.5 and 4.5. Is this true? I figured the op amp wouldnt care or even notice anyways but what the hell do I know
In practice, many op-amps don't work well with such a low positive supply although they don't care much for their low negative one..
(where "low" are the margins from zero volts..)

P.S.
4n7 (C1) is rather big value for a feedback cap..
Tired of updating my account to enable 3rd party hosting..

PRR

> I figured the op amp wouldnt care or even notice

Generic opamps have no "ground" pin. How would they know where you think ground is.

In simple bare-device AC-coupled applications, single-supply is almost always easier to deal with, and "natural".