(SOLVED) Distorting Neumann KM84

Hello everybody!
I've just bought a Neumann KM84 dirt cheap that's not working and I would really really like to get it to work.
The issue with it is the it's got very low output and the sound is all distorted. Sounds like a transmission from a moon landing.

The guy I bought it from had swapped the transistor but that did not help.
When I got it, it was SUPER noisy, and I realized that a tantalum cap had a loose leg. So I soldered it in place and the noise disappeared but the sound was still very moon landing-like.

I'm gonna try my capsule in another km84 next week to see if it needs a new one.

Any other ideas what it might be? What should I look for when troubleshooting?

Any advice much appreciated!

This is how I approach damaged KM8x:
- Check capsule connection (the two wires on top of the PCB assembly that connect to the middle pin)
- Check for shorts caused by improper repair attempts.
- Check everything for bad solder joints
- If the input FET has been replaced, re-biasing is necessary (like U87 biasing, it is the same procedure).
- Replace tantalum transistors because of age.

Be careful around the transformer - the connection wires are very delicate.

Thanks a lot! That was actually a question I had, about the transistor. How much difference can biasing do? Could it be the difference between almost no sound and distorted to angel like heaven sound?

Biasing? How much difference does it make?

Potentially a lot. Most JFETs have a gate-source threshold voltage that can vary by a factor of 4. Some parts can be purchased pre-selected (e.g. into a 'blue' or 'green' grade etc.), and that will narrow the range to 2:1, but for some circuits, you either need to select a part with the 'right' Vgs threshold voltage or adjust the bias resistors to work properly.

So, yes, it can matter. Not sure what the original transistor did and what the new one does, but it's possible that it's biased way out of the linear region, causing low gain and excessive distortion.

I personally wouldn't trust tantalums as far as i can throw'em, especially ones as old as my dad. Those would be my prime suspects (barring capsule issues or intermittent contacts).

http://sound.whsites.net/articles/capacitors.htm#s12

The KM84 circuit is about as simple as it gets so it shouldn't be too hard to diagnose.


Start with checking voltages. Check for +48V on the outside of R1. Check that the JFET drain is ~10V.

It could be that something is leaking current and because the bias is applied through a 1G resistor it will never charge up [1]. That would give you low / distorted output. It could just be dirty. Any residue or flux on parts connected to the capsule wire can conduct enough current to crush the bias. Use a q-tip and some 90% isopropyl alcohol to clean all of the parts connected to that wire including the board itself. Then polish off the residue of that with a dry piece of a lens wipe. Anything in the vicinity of that capsule wire could drain the bias. Or maybe there's dirt around the capsule itself that could be a path of conduction.

If someone change the transistor, make sure they used the right transistor. It needs to be low-noise and high gain like 2N3819 which is the usual part available on Mouser. And of course you need to re-bias. Lot's of ways to do that. I like to look at the spectrum with the level at the threshold of clipping. Or just bias for 10V on the drain like most people probably do.

You might also check to see if the previous owner tried to mod it like some people do to eliminate the coupling cap.

[1] You can't measure that net with a meter of course because the impedance is so high the meter will crush the bias.

This is so great. Thanks a lot for your insight. I REALLY appreciate it!

There has definitely been some poking around in it from previous owner and it’s quite a mess in there.  I’ve done some tests however.

I tested the capsule in another km84 and it’s working so that’s nice.

The voltages seem fine (46v outside R1 and 8v at jfet drain)

I’ll try and clean it up in the and maybe reinstall the original transistor. If that doesn’t work I’ll replace the tantalums. Or should I maybe do that regardless? To electrolytes?
Could they be the reason for leaking dc?

Tantalums, especially old ones, are... questionable, to say the least.

http://sound.whsites.net/articles/capacitors.htm#s12
http://web.archive.org/web/20030807122631/http://capacitors.com/picking_capacitors/pickcap.htm#signal

Found the fault. It was one of the wires to the xlr-pins that was of.
I never thought of that since it had been with to techs before me. Now it works fine but I have yet to put it to test.