Matching J-FETs

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.


Well-known member
Jun 19, 2004
georgia, USA
There was a technique if I recall correctly for matching fets. How are you guys doing it??? I am starting on the Hamptone J-fet line mixer.
I think this will sort specifically for the Hamptone (because the values look like Hamptone values):

I think this is an all-purpose FET sorter, drawn I think by Fred Forssell, with a small change proposed by myself for simpler operation at small loss of accuracy.
hi PRR, first,I want to thank you for all your great posts here.:thumb:

I used the circuit tat's suitable for matching fets for the hamptone .

I plan on trying a hamptone pre but what I'm matching for now is the
G1176 .Can you suggest a simple adjustment to the first circuit that suits the G1176 values?

> what I'm matching for now is the G1176

What is a 1176? Is that one of those FET limiters that is a 27K series resistor, a shunt JFET, and a bunch of other junk?

Why are you "matching"? Are you trying to lash two together for Stereo? Are you expecting them to track so well that there will be "no" image-shift while limiting? I suspect that this is extremely difficult; that you are fighting the normal variation from one FET to the next. My inclination would be to find out if there are any dual monolithic JFETs still available, and base your stereo design around the one part. FET parameters are so variable that any precision limiter will have enough adjustment to use about any FET. But doubling-up the adjustments to match two separate FETs gets harder to build and real hard to calibrate.

Set up a test rig similar to the actual circuit:

R1 should be equal to the series resistor used in the actual limiter.

V4 is a DC or AC source. It must be 100mV peak or less to stay within the FET's triode range. A 1.5V battery and a voltage divider to give 100mV will work. Or use an audio signal generator set for about 70mV RMS (100mV peak) output. Read the voltage from ground to the "V" symbol. For the DC source, you need a DC meter that will read 100mV down to a clear indication of 10mV; a 199mV digital voltmeter is barely good enough. For AC source you can read the output with a high impedance ACVM, or a high impedance audio amplifier (direct-box into a mike input) monitored with a VU or PPM meter. The audio technique can read in dB, which is intuitively what we want.

V1 is a 9V battery and a pot so we can put 0V to -9V DC on the Gate. Put a good DC voltmeter on the Gate.

Start without the FET. Trim your source so you get 100mV DC or 70mV AC at the "V" symbol. If you are using a VU meter, first use an ACVM to confirm about 70mV AC, then trim the mike amp so the VU meter reads 0 VU.

Put in an FET and trim the Gate voltage to -9V. You should still have the same voltage at "V". If not, check your connections.

Now trim the Gate voltage toward zero, slowly. Watch point "V". At some point it will drop 10%: 90mV DC, 63mV AC, or -1VU. Write down the Gate voltage needed to make this happen.

Now find the Gate voltage to make point "V" 50% or -6dB.

Now find the Gate voltage to make point "V" 10% or -20dB.

Insert another FET. Repeat.

The 90% or -1dB point is the 1dB Gain Reduction point. If you are just trimming a couple dB off the top, this is the most important test.

The 10% or -20dB point is of course 20dB GR. If you expect zero image shift in sustained heavy GR, this has to match very well.

For general use, all three points must match well. Also all points in between, but JFETs are predictable enough that if you match at -1dB, -6dB, and -20dB, it will probably be close everywhere else.

Yes this a fet limiter. I don't plan on lashing two together as a practice,
but I would have the option to do so only if I take the trouble to match fets .When I try stereo material I'll watch for image shift for sure.I've read that having matched fets in the unit itself also helps for dialing in the meter tracking.

Thank you for the test jig and clear guide through testing.

I have been using your jig to match some 80 fets all day. :roll:
My question is what is considered a close match? IE: what would be a 1% tolerence? This seems stupid, but my confusion is that I have two DMM's set up. One is measuring the mV. It can measure down to .01mv. I have it set for a range of 100.00mV and it sweeps down to 10.00mV for the 90% cut. The other DMM is measuring the 9V supply. It has a placement of 9.000V. Therefore, I'm getting readings at the 90% cut range of (for example) 1.332V on that meter and 10.00Mv on the other. So my question is should I be going out to that third decimal poing on the 9V DMM? Sorry this post is worded so confusingly. And yes, I am also trying to match up a stereo compressor.

> Sorry this post is worded so confusingly.

It does not make ANY sense unless you are doing something very wrong.

Why do you have a meter on the 9V battery????

Why does it say 1.332V????

V4 is a 100mV AC source. Measure the AC voltage across the FET.

This is, if you think about it, the same circuit as the limiter, except manually adjusted and with fixed input level, and all excess frills omitted.

If the AC signal is any higher than 100mV, the FET will distrot bad. It would be good to put a scope or amp/speaker across the FET to be sure you are not overdriving it.
Why do you have a meter on the 9V battery????

mbira said:
The other DMM is measuring the 9V supply.

This was incorrectly stated-sorry.

The meter is not on the battery. It is connected to ground and the Gate voltage. The reason I called it 9V is that that is what it measures when the pot is turned all the way up.

Why does it say 1.332V????

Because in that instance, I was trimming the pot until the other DMM read 50mV and that was what I see on the gate voltage.

V4 is a 100mV AC source.

I am using a 100mV DC source. I am never going higher than 100 mV.

I hope this clarifies things?

I apologize for my tone. My hay fever really kicked in today, and between that and the drugs I'm really grumpy.

And stupid. I did say "DC", didn't I?

> what is considered a close match?

Aye, there's the rub.

Let's call the meters "Gate bias" and "Output". The Gate bias is 0 to 9VDC. The Output ranges 100mV to 0mV.

If you follow my instructions, for each FET you have three numbers: -1dB, -6dB, -20dB. These are the numbers on the Gate Bias meter.

Find the best matched pair. Look closest at the gain-reduction range you plan to use. Myself, my usual goal is "verbatim", the limiter is only to save my butt on surprises. So I'd look most at the -1dB and -6dB numbers. If I were smacking-down persussive transient tracks to sit in a mix, I might be running 10-20dB compression most of the time, then I would look at those numbers for best match.

Now what you really want to know is: at the SAME Gate Bias (what will happen in the final limiter), what is the difference in Output for these two FETs? Say part A gives -1dB (90mV) Output at 1.332V Gate Bias. Don't touch the pot, put part B in the tester, at the same 1.332V Gate Bias. If the Output meter again says 90mV, the two FETs are perfectly matched at -1dB compression.

They won't be perfectly matched. So how much "off" can it be? For a stereo limiter to not give image-shift, even 1dB difference may be audible. Say 0.5dB goal. That's 6%. So the "90mV" could be 95mV to 85mV, and I would call it "perfect". The -6dB point, 50mV for part A, could be 47mV to 53mV on part B at the same Gate Bias. The -20dB point, 10mV, could be 9.4mV to 10.6mV.

If your two best FETs don't match this close, you either need another box of FETs, or a relaxed match specification. Frankly, 1dB stereo shift will generally be unnoticed, especially in panpot stereo. (I do mostly close-mike no-mixdown stereo, so I have an honest image and image-shift bothers me more than most folks.) That's 10% difference, say 81mV to 99mV (not 100mV) at the -1dB test point.
I apologize for my tone...

No need to apologise. My post was very poorly woorded and I'm pretty ignorant of all this stuff, so you're having to do your share of "mind reading"

Thanks for the clear answer!

Has anyone had any problems with oscillation when trying this jig to measure the the AC as you trim the gate voltage? I have remade this 3 times with the same thing. I am using a 70mv AC signal from a sig-gen, a 100k pot. Power is from a fresh 9 volt. Both meters to ground at the source. One meter measuring DC at the gate and the other measuring AC at the drain. When i trim down the voltage from 9v, the AC signal starts oscillating, and i can't get a solid measurement.
Just as another possible aside, both meters have good batteries
Any ideas?

Latest posts