THOR/jfet biasing advice needed.

[slashandburn]

I think I'm needing to swot up on JFET biasing.  I built a R.O.G Thor using the guassmarkov project files and pcb, dialed in the recommended Q1 and Q2 drain voltages with surprising ease and, well, it worked and sounded okay but had very low output on the volume pot  and the gain pot did practically nothing until the last 10% of it's travel.  (pot values and taper are correct as per schematics provided below).  So I had a go biasing it by ear instead and the pedal really came to life. After a bit of tweaking the trimmers, the volume and gain pots seemed to be much more usable throughout more of their sweep and I was happy as Larry. That was until I noticed my biasing by ear had left me with drain voltages on Q1&2 that seem further off from what is recommended than what I'd imagined.   At 9.6V Supply voltage my Q1 drain measures 4.2V and my Q2 drain 4V.

I (sort of) understand that these biasing voltages are usually "ballpark" figures and that part of the motivation for biasing jfets with trimmers is to compesate for (often massive) variance in characteristics (hfe?  Vgs?  This is where I fall down a bit) between jfets from the same reel.  I don't know how well I've put that, do we use the tolerance for fets?

Anyway, so I'm not hugely fussed that my voltages don't match whats recommended on the build docs. But it's still bugging me slightly, as Q1 is off by quite a bit and I also figured that in the case of Q2, where the drain is to be biased to ~4.5V, I figured the "ballpark" idea was less relevant and this should be taken to mean "set it to half the supply voltage, erring slightly on the lower side if you can't get it exact".  Anyway here are the relevant links to the build docs should they be required:

http://www.runoffgroove.com/thor.html


https://web.archive.org/web/20101226135107/http://gaussmarkov.net/layouts/thor/thor-project.pdf

Tl;Dr.......My Thor seems to sound better with Q1 biased to ~4.2V rather than the recommended 6.4V, and with Q2 biased slightly below the recommended half supply voltage.  Is this normal behavior or have I screwed up something somewhere?

[Frank_NH]

Your drain voltages look OK to me for the source resistors specified, and of course, if it sounds good, it IS good.  Welcome to the wonderful (and highly part-variable) world of JFETs. 

BTW, if you used sockets for your transistors, try substituting other JFETS of the same model (e.g. 2N5457 for Q1) and see if the Q1 drain biases to a higher voltage.

[Ben Lyman]

I've always been a little suspicious of those circuits that say to use 100k trimmers on the drains. I'm sure it's a fine way to go but when I was researching the Plexi-Drive, I came across some interesting info from Brian Wampler. He said to use 15k fixed resistors on all three drains, then start swapping J201's in and out until they all read close to 4v5.
So far, this has worked out great for me, although it does leave me with a pile of leftover J201's.

[Frank_NH]

What I have done in the past is to measure the Idss and Vp for a pile of JFETs (J201s), then label each one with a unique ID and record that on a sheet.  Then, when you're using them for a build, consult your sheet and choose the appropriate labelled parts for the circuit (you can calculate the required drain resistor with reasonable accuracy for simple JFET gain stages if you know Idss, Vp, circuit voltage and the source resistance).  A lot of trouble, yes, but these are JFETs after all...

[slashandburn]

Good to know, thanks!  Something still seems slightly off though. The first half of the pot sweep is pretty much spend trying to get up to unity. Similar story with the gain pot, there's not a whole lot happening for the first half.

I'll have another play around and perhaps swap in other jfets. If it's still bugging me I'll post full voltages.  I've only really had a chance to trial it at fairly low volumes. It sounds great but something tells me i can get more from it. I wonder if my issues could be solve by placing a boost up front. I'll report back in due course.