2N3819 source

[JW]

FWIW, I bought some (from memory LM846 does that sound right?) transistors from NatSemi where the bias resistor measured closer to original in the KM84, than the new stock 2N3819, which was the original transistor. If you research this in KM84 threads, there is some info.

 

[RuudNL]

Some time ago I bought 50 pcs. from Ali(Express).
To my surprise all 2N3819's could be biased correctly in a U87 circuit.
(I tested them all in my 'FET tester'.)
No idea if they are 'real' 2N3819's! But they work and have low noise.

 

[Ricardus]

When I built my 87s I used NOS Fairchild that I bought from a reputable vendor. Biasing was no problem. I have yet to try a "new" 3819, nor will I need to since I have a bunch of NOS. But sheesh, if Mouser is selling them it seems like it should be legit.

 

[RuudNL]

They probably are legit, they are FETs, but they don't bias correctly in a U87 circuit.
(I even know about sellers who sell NPN transistors as 2N3819 Fets!)

 

[Rybow]

Whoever told you that didn't realize, or forgot to mention, that this is dependant on the particular DMM you're using.
It's basically not more than a continuity test.
Regarding fakes, they often pass the test and even more sophisticated ones.
The most common fake is passing factory rejects for good ones.
Quite often the difference is in excessive leakage or capacitance or noise, that even a curve tracer can harly diagnose.
The ultimate test is in the intended circuit.

That’s a good point. Probably best to build with a socket for quick swapping.
Many moons ago I built a km84 LDC and biased by setting drain voltage to 10V. Sounded pretty good. Recently I rechecked it with a 1khz tone and adjusted for max gain/symmetrical sine wave. My original bias wasn’t too far off, so I must have gotten lucky with the 2n3819 I had installed. Didn’t recheck the drain voltage though. For optimal results, are we looking for a certain drain value after biasing with our test tone?
My mic is low noise and sounds great, so I’m not overly worried about it.

Going through my parts I discovered some 2n3819’s from digikey that I ordered in 2011 when I built my mic. I’ll have to build up a circuit, and then test those against the ones from ebay

 

[RuudNL]

I usually bias a microphone circuit with a low distortion test tone and a (hardware) distortion analyzer.
You find two 'optimal' bias points: one sharp distortion dip at moderate levels, and another dip at maximum SPL, where clipping almost sets in. As everything in life, the setting is a compromise!

 

[Rybow]

I usually bias a microphone circuit with a low distortion test tone and a (hardware) distortion analyzer.
You find two 'optimal' bias points: one sharp distortion dip at moderate levels, and another dip at maximum SPL, where clipping almost sets in. As everything in life, the setting is a compromise!

Oh the wonders of compromise! That’s really good info. Thanks!
One thing I forgot to do was actually push the FET into distortion. My test setup is jury rigged, so not sure how accurate I can be. Test tone from Logic out of my 2i4 to the gate. Back into my 2i4 mic input to logic with Melda Moscilloscope plugin on the track. Seemed to work ok. It was kind of neat watching the sine wave change as I adjusted the bias.