Boss PH-1R - JFETs

[Bucksears]

I'm working out a new PCB for a PH-1R project. It omits the buffer/switching portion.
It uses 4 JFETs (2SK30A-GR) that have been matched (apparently), but I'd like to know if I can substitute a different JFET matched-quad (2N5485?) for the same effect?

I know the pinout will be different and I will adjust the PCB accordingly; just curious if I can use other JFETs.

[Bucksears]

Guessing I could just use these with the stock layout:

http://smallbear-electronics.mybigcommerce.com/matched-jfets-for-phasers-gsd-pinout/

[Mark Hammer]

Regardless of what you use, you're going to have to bias the FETs anyway.  But how applicable are FETs from a Phase 90 to a PH-1r?

Both units use a 10k input and feedback resistor for each phase-shift stage, but the P90 uses a .047uf cap, while the PH-1r uses .01.  The P90 uses either 24k or 22k (depending on issue) resistors in parallel with the FETs, while the PH-1r uses 100k.

The lowest point in the sweep is given by the cap value and highest possible resistance to the reference-point, which is going to be set by the fixed resistor.  As the drain-source resistance for each stage gets smaller, that combined resistance starts to go below the value of the fixed resistor, and the point where the max phase-shift is applied gets higher.

So, with .047uf and 22k in parallel with the FET, the point where maximum phase-shift is applied is going to be 154hz.  Anything below that will have <90 degrees shift.   Were the FET+resistor to have an effective value of half that (11k), the phase shift would only reach that max for content above 308hz. 

For the PH-1r, .01uf and 100k gives us a lowest point of max phase-shift at 159hz, or pretty close to the P90.  Drop that 100k resistance to half (50k) and max phase shift doesn't start until 318hz; again, pretty close.

The punch line is that you can probably use other FETs, but the drain-source resistance of the FETs used may not necessarily complement the fixed resistor values in a manner to produce the phase shift at the point in the spectrum where you want it.

HOWEVER...

As the calculations indicate, you can fix that by simply changing the value of the caps, OR changing the value of the fixed resistors, OR both.  What you need, to start out with, is some idea of the drain-source resistance change for whatever FETs you end up using.  K30s are actually reasonably easy to get, although the GR subtype may be a little harder.  If you can find K30s cheap enough, though, you buy a bunch, find yourself a matched set and plug'em in.

[Bucksears]

Thanks Mark.