Old JFET in mic goes noisy?

[marz.tv]

Hi all, a question for the solid state experts:
I have some old (1970) 48v condenser microphones that all are a bit too noisy IMO. They're called DC-20 and DC-21, made by Pearl (later Milab), small diaphragm capsules apparently very much constructed like the Neumann KM-84. Great for piano and acoustic guitar.

They use the jfet 2n3819, a bunch of resistors, one or two small caps and an output transformer. I'm trying to pin down the source of the noise, and I was thinking of swapping out the fet. It is not hum or buzz, just something like white noise.

Any suggestions for modern replacements? Do you think it is the fet that is the source of the noise, or should I perhaps swap the resistors for modern metal film ones instead? If possible, I'd like to keep or slightly improve the basic sonic signature of the mikes.
Any help welcome.

 

[john curl]

Clean with pure alcohol, pressure air dry, before making changes. 2n3819 fets are easy to get and usually work well. Important resistors are VERY HIGH in value, not normally available.

 

[mrfeedback]

Fets going leaky can be a source of noise in condensor mics, and so can contamination (spit, dust etc) of the mic capsule.
Leaky fets usually cause a crackly, erratic popcorn sort of noise, but can give a sort of 'white/pink' noise.
Johns advice is good (use Isopropyl alcohol) but keep it away from the actual mic capsule (probably better to remove it first.

Eric.

 

[BrianL]

It is possible that moisture may be between the diaphragm and
the backplate. This might be fixed by a >>gentle<< bake in a
dessicated atmosphere.

There are some Toshiba FETs designed for condenser mikes
that are very low noise, but you want to make sure they
have low gate capacitance. These will probably be substantially
quieter than your 2N3819.

Noise can also be caused by surface contamination in the
electronics.

So there are multiple possible sources.

PML/Pearl are/were very nice mics.

Maybe David Josephson (www.josephson.com) could give
your some hints. If not, try Scott Dorsey.

 

[marz.tv]

Thank you all for your advice so far!
Pearl (still in business!) graciously sent me the schematics, so now know what I'm looking at. Apparently their 2n3819 was a specially selected 'silver' version, but I guess it could still be that a 35 year old transistor goes noisy, sure? Or maybe it was normal and within acceptable noise back then before digital, so perhaps an updated fet would be good to try.
I'm usually fiddling with tube circuits, so please give me some hints to what jfets I should look into. I don't know which types are good...

 

[mrfeedback]

Some TO-92 transistors had silver plated legs that go black.
If your (transistor) legs are black, this will very likely be the problem, and is a common noise fault in older gear.

Eric.

 

[marz.tv]

...sorry, typo. I meant 'Now I know...'
But here's the link so YOU know what's inside the mic too.

http://w1.314.telia.com/~u31427269/extra/New-2.jpg

 

[marz.tv]

mrfeedback, that was very interesting information! Yes, the legs are silver and look a bit blackened. Do you know what is the connection between silver oxide and noise? Some sort of 'skin effect' or what? And, more importantly, how do I clean that oxide away? I presume it is best to unsolder the fet first, an then dip ino some solution.

 

[john curl]

Stick with the 2N3819. I know from experience. Yes, there are 'better' fets, but I doubt that they will be functionally quieter than a selected 3819. Watch out for input capacitance. Low noise fets have more.
Ultimately, the VALUE of the input resistor(s) will set the noise of the mike. Higher values like 1 gig ohm are best. 1*10^9 ohms or higher are difficult to find.

 

[mrfeedback]

I am not fully sure of the maechanism, but variously I have been told that the oxiding silver migrates to the transistor chip and causes leakage/noise currents and I have also heard that the silver oxide migrates across the package and causes leakage/noise currents between the transistor pins.
A transistor in this condition is visually unmistakeable - very black legs all the way up to the package.
In servicing work, any and all transistors in this condition are changed as a precaution at least.

Eric.

 

[marz.tv]

Thanks John and Eric for the insight. No, the legs weren't all blackened, still I tried rubbing a little with a q-tip and isopropyl alcohol where I could reach, but there wasn't much immediate difference.

I got a really helpful mail from Bernt Malmqvist at Pearl (not many companys care to support 35 yr old products!).
He said that 2SK879 is what they use in their mikes nowadays, but that it is surface mount. He also recommended J201 as a good TO-92 fet in this application. Does anyone have experience with these parts?
He also said that the old tantalum caps can be a noise problem, so maybe I'll swap these parts first before going transistor shopping.

I noticed the noise suddenly dropped a little in one of the mikes after half an hour or so of warmup time. Would this be some old cap reforming, or perhaps breakdown of an oxide trace somewhere? (ok, this might be deep water I know

 

[Vek]

In my experience old electrolytic capacitors often become noisy, especially if they are in the path of very small signals (i.e. phantom power blocking caps with mic signals).
I found this in noisy channels of older mixing consoles.

I think replacing the old tantals would be a good start.