Source followers are pretty insensitive to changes in Rs (though Vgs will increase a wee bit as Rs goes up.) There should be no need to change Rs for these stages.
If you do want to tweak for maximum output headroom, the DC bias voltage at the gate is where the tweaking should be done. Not by changing Rs.
I typically keep power dissipation below 250 mW for anything in a TO-92 package. That's a conservative level that keeps them cool and happy. With a 12V supply, that suggests keeping Id below some 20 - 30 mA (allowing for the fact that the full 12V doesn't appear across the JFET.)
I don't think 2N3819s will ever flow that much current, even at zero Vgs, but some other types of JFETs with lower channel on-resistances and higher Idss specs will.
-Gnobuddy
To clarify, the above statement is specifically about source followers. (As is my entire previous post.)
-Gnobuddy
I didn't understand that this advice wasn't meant for the source-follower stages (doh!). I was attempting to put the gate voltage between the drain and source like in FET 1,2,4.
I'll switch those Rs back to 27K.
The output of the pre isn't super low, its better than it was, and I don't have the piano to test it out. I can tweak later if need be.
However, I still have the lone original K30A which came with the preamp. Maybe I'll put it back in to improve the gain since its still here - any suggestion of which position will yield the best results, 1, 2 or 4?
Last edited:
Well, you can tweak source-follower stages this way too, but you can't tweak away JFETs that want 6 volts of Vgs, instead of 1 or 2 volts! Only a JFET swap will fix that.
These are depletion-mode, N-channel JFETs; they bias up like a vacuum triode does, with the cathode (source) more *positive* than the grid (gate.)
So you can't put the gate voltage between drain and source; the source has to be more positive than the gate!
Now I'm confused which stages we're talking about. But to clarify: for the source-follower stages, I suggest going back to the original factory value of Rs (2.7k or whatever), and then tweaking the DC voltage at the gate, so that the source ends up roughly at half Vcc, or as near as is reasonably practical.
FET2. That's the only stage where the source resistor is fully bypassed with a capacitor. That means it was originally the stage with the highest voltage gain.
As mentioned previously, using a high Vgs JFET lowers the voltage gain even if the source resistor is fully bypassed - this happens because you end up with less DC voltage across the drain resistor, and that reduces the voltage gain.
So if you currently have a JFET with much greater Vgs_off in the FET2 position, it's producing less voltage gain than that stage was originally supposed to.
-Gnobuddy
Arrrgh! Looking back at post #23, there is still ongoing confusion, and miscommunication on my part, because my mental cogs slipped for a moment, and I interpreted "source-follower" as "common source", which is totally incorrect. 
To clarify and correct:
1) For a source-follower, leave Rs alone, tweak the DC bias voltage on the gate.
2) For a common-source gain stage, start by tweaking Rs. If Vgs_off isn't too far off spec, this will get the stage working. But if Vgs_off is way out of line, you might need to tweak the DC bias voltage at the gate so that Rs doesn't become ridiculously big in the effort to bias the JFET.
I hope that clears it all up!
-Gnobuddy
To clarify and correct:
1) For a source-follower, leave Rs alone, tweak the DC bias voltage on the gate.
2) For a common-source gain stage, start by tweaking Rs. If Vgs_off isn't too far off spec, this will get the stage working. But if Vgs_off is way out of line, you might need to tweak the DC bias voltage at the gate so that Rs doesn't become ridiculously big in the effort to bias the JFET.
I hope that clears it all up!
-Gnobuddy
Small Bear will sell you SK30 for a buck; however postal rules to Canada will make the shipping more than the parts. But maybe less than the value of torn-out hair?
Transistor FET 2SK30A-GR - Small Bear Electronics
Yes, it is not the right suffix. I still think it is more-right than '3819. You want a hot motorcycle engine, the '3819 is a medium sedan engine. Yes, I remember a Harley with a 283 Chevy in it, but it was not a good fit or ride, it was better with the original 45.
The J113 is still available at both Digikey and Mouser. It might make a good motorcycle engine. Vgs_off ranges from (-0.5V) to (-3V), and minimum Idss is only 2 mA (max Idss is not specified.) Those samples with Vgs_off and Idss near minimum spec are definitely "hot motorcycle engines".
Too bad the PF5102 has disappeared. It was solidly in the hot motorcycle engine category (small Vgs_off, small Idss.)
-Gnobuddy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.