Matching JFETs - low Vgs readings

[patricks]

Hi all,

I've just this second finished building RG's jfet matcher (http://www.geofex.com/article_folders/fetmatch/fetmatch.htm) using the stripboard layout at tagboard effects (http://tagboardeffects.blogspot.com/2012/07/jfet-matcher.html).
I tested 20 x 2n5952 jfets I got from Tayda to use in a Phase 90 clone. The spec sheet for the 2n5952 says that the Vgs(off) should be between -1.3 and -3.5V, but all of mine tested between -0.286 and -0.648V. I suppose it could be a dud batch, but it's more than likely because of something I did. I've gone over the board with a sharp knife to make sure there are no solder bridges, checked the polarity of the capacitor and made sure the cuts and links are in the right places.
So, should I:
1) change the value of Rset or the capacitor and re-measure
2) just accept the fact that Vmid is a bit off and pick the closest four jfets

Cheers
Patrick

[armdnrdy]

Check to make sure your pin out is correct because those numbers are way out of the ballpark.

Looking at the front of the tranny (flat side) with the legs facing down....it should be Gate, Source, Drain.

[patricks]

Thanks for the reply

Yup, checked the pinouts and I was definitely putting them in the right sockets - I tried them in different orientations (hope that won't damage them) and got the same readings each time for all the different jfets (3.5V). That suggests to me that the pinout is correct, since there's only one orientation that gives different readings for each jfet.
The label definitely says "2n5952" as well.

Does the difference between expected and observed Vgs measurements give me an idea of where in the circuit to start looking for faults (apart from the usual solder joints, accidental bridges, etc.)?

[pappasmurfsharem]

Thanks for the reply

Yup, checked the pinouts and I was definitely putting them in the right sockets - I tried them in different orientations (hope that won't damage them) and got the same readings each time for all the different jfets (3.5V). That suggests to me that the pinout is correct, since there's only one orientation that gives different readings for each jfet.
The label definitely says "2n5952" as well.

Does the difference between expected and observed Vgs measurements give me an idea of where in the circuit to start looking for faults (apart from the usual solder joints, accidental bridges, etc.)?

It's not just you! Mine did the same thing using the same layout, same readings as well...

I put together the phase 90 from Tagboard as well and it appears to work as expected.

I bought my 5952s from tayda

[patricks]

It's not just you! Mine did the same thing using the same layout, same readings as well...

Oh, thank goodness!
I'll put my Phase 90 clone together and see how it goes. I've got a bunch of SIP sockets, so I'll use them rather than soldering in the jfets. If it all turns to custard, I can swap them out easily

[R O Tiree]

This is because the JFET matcher doesn't measure V_GSoff.  It's measuring the voltage at which R_DS is 10k

[patricks]

Cool, thanks!

So, if I wanted to adjust the circuit to get the readings within the specs, should I change Rset or the capacitor, and in which direction (smaller or larger value)?
Or does it really matter, is the precision and accuracy of the measurements still going to be OK even though the tester is reading on the low side?

[PRR]

> spec sheet for the 2n5952 says that the Vgs(off) should be....
> RG's jfet matcher ... mine tested between -0.286 and -0.648V....

What Mike said. "the (R.G.) JFET matcher ... voltage at which RDS is 10k".  

Vgs(off) is nearly the voltage where RDS is infinite (dozens of Meg).

> adjust the circuit to get the readings within the specs

Huh? No.

The extra-work way to match JFETs for this purpose is to test *BOTH* Vgs(off) *and* Idss. The two extremes.

R.G. realized that we don't really want the extremes, we want the mid-point. 10K is a fine midpoint for most audio JFET uses. His tester finds the Vgs(10K) voltage. Select several JFETs with near-equal Vgs(10K), you will be fine.

> -0.286 and -0.648V

That's a pretty broad range. Can you find four which are closer, like 0.5V-0.6V?

[patricks]

Cool, thanks for that.
The -0.286V to -0.648V was the range, min to max. I've got four that sit at 0.368V, 0.373V, 0.377V and 0.379V, so I was planning on using those. When I graphed the values in excel, the line of best fit through these four was almost horizontal, they were the best matching four of the 20 I tested.
I've got another three matched pairs, too, which is handy

[PRR]

> 0.368V, 0.373V, 0.377V and 0.379V

0.379/0.368 is a 3% match which is fabulous.

I would not worry about 5%, and would sure try a 10% spread.

[R O Tiree]

The pairs would be fine for Phase45 builds.

If, for some reason, you actually want to find V_GSoff then you could change R_set for something like a 4M7 resistor - that will get you within a gnat's hair of it (I'm talking within µV of the "true" figure) and, given that your multi-meter will only display 3 or 4 significant figures, that's as close as you need to get.  If you ever build an Orange Squeezer, for example, you'll need to know that figure for one of the 2 JFETs to set it up accurately.

[patricks]

Brilliant! I'm pretty sure I've got another four that'll be within 10%, so I might experiment with them a bit.

Out of curiosity, what other projects use matched jfets apart from phasers and the Orange squeezer?

[R O Tiree]

You don't need matched pairs for an Orange Squeezer - You just need an accurate V_GSoff for one of them.

Offhand, I can't think of another FX circuit that uses matched FETs. There probably is one, but I can't think of one right now.

[patricks]

Cool, thanks!