4.5v at J-FET drain with 8k ohms--how?

[twabelljr]

    I am about to move a working "Hot Harmonics" from breadboard to perfboard. The schematic shows a trim-pot to be adjusted to obtain 4.5v at the J-FET drain or to just use 8k. Also the "Three Legged Dog" shows an 8.2K resistor for the same purpose. For me to get 4.5v I need to use 1.8k ohms. 8k gives me 750mV. Is the idea to get 4.5v and not worry about the resistor value. If so, why is 8k somewhat of a "default" value. I am using a MPF102 J-FET. Thanks in advance. Tommy   

[MartyMart]

Yes, 4.5v is about where you want to be.
Jfet's can be very different, so the "nominal" 8 - 15k values almost never work for me !
I would always use a trim pot, 20k 50k 100k, whatever gets you close enough with that
particular Jfet.
You may well have "one" that will bias with an 8k2 or whatever resistor, or you could find
the value with a pot, set it and remove it, measure it and put that value "fixed resistor" on the board !

Marty.

[BrianJ]

You might also want to try a current source on the source of the jfet.  Design it for 1ma and a 4.53k(common metal film R value) on the jfet source and you'll 4.5 v every time- even with different jfets/ supply voltage. That might be overkill.  Jfets can be tricky to bias because of there Vgs/Ids spread.  Using a smaller resistor to acheive the bias voltage you need may cause signal loss due to the divider created with the jfet moderately high output impedance.  Of course, if youre cloning, this probably is overkill...

[lovekraft0]

You might also want to try a current source on the source of the jfet...

Could you explain that a little further, please, for those of us with limited math skills... or maybe post a schematic? I'm not sure about anybody else, but I'd be willing to go to heroic measures to stop having to fiddle with all the trimmers in those ROG-style amp sims! 

[BrianJ]

There a several ways to make current sources (BJTs, FETs,OpAmps..).   The way I do it this:  Instead of tying the 4.5K source resistor to ground, connect it to the collector of an NPN (if the supply is pos) BJT.  This transistor need not be stellar in many regards, nor of special sonic quality.  I use 2n2448 (PNP 2n4250).  Connect the base of the transistor to the supply through a big R (10k-100k) and to ground through two forward biased diodes (for a +V).  This makes a little voltage regulator that will hold the base at 2 diode drops (about 1.2v).  That will hold the emmitter at .6v.  Now you "program" the circuit with an emmitter resistor that is .6v/I large.  For a 1mA CS:  .6v/.001A= 600ohms.  If you wired it right, the circuit will try to force 1 mA of current through the drain source channel of the FET and create an apropriate drop across the source resistor: E=I*R.  If it works, bypass the current source (collector of the bjt, "under" the source R) to ground with a big cap (4.7uF -47uF).  This will make it look like ground to audio frequencies (essentially its identical to what you started with as far as signals are concerned).  If you do try it, I recommend some light reading on current source operation.  There are other benefits(and consequences) of using CS for biasing but nothing that needs explanation now.


This is what I do when I design around JFETs for followers.  It has served me well.

I hope it helps