MXL 603 -> KM84 Replacement PCB

[Ricardus]

My vote: The second one marked KM-84 sounds bigger, livelier, more musically exciting - not a lot, just enough to notice. The KM-84 has more percussive presence, i.e., it has more punch and sizzle, e.g., picks striking strings are more crisp and articulate and percussive. I like the overall tonal balance - the mandolin sounds right and the banjo is sufficiently far back to avoid the usual harsh percussive whack banjos have up close. If you were producing my record, I would have you use the KM-84. While both are VERY close in overall sound, the KM-84 JUST slightly more energetic. And that is just MY take ... James

This is why they should have labeled them A and B.

 

[till]

Hi all! The fresh run of PCBs has arrived. I've sent a bunch of emails for those who have put their name down already, so if you haven't got a from me message yet or would like to add your name to the list please message me and I'll get you sorted.

Rock Onwards!

Hello Graeme,
I don't know how to write a PM (maybe because I am a new member?). I would like to buy two of your mxl-km84-pcb's. Can you send me a PM?
Thank you in advance.

 

[joulupukki]

I was just doing some sample recordings with the KM84 clone I built (from this PCB) going up against a Microphone Parts SDC-84 and a Neumann KM-184. The Neumann has significantly less self noise. The MP SDC-84 is a bit noisier (noticeable) and my KM84 clone with Graeme’s PCB is even noisier than that.

Any ideas of what things would contribute to more self noise vs. less? I’m wondering if there are things I could do to improve the self noise in my clone build.

 

[klem]

forgive the basic stuff here, but have you cleaned the capsule pin and high z section with repeated scrubs of isoprop and distilled / purified water? I've also found that removing any excess solder and reflowing what's existing at high z areas lowers noise, my hunch (and nothing more) is that this lowers the amount of flux chemicals in the area. Assuming components are relatively similar between MP and GW boards, and housing is the same, PCB layout and/or composition may be the biggest culprit. I'm interested in hearing/buying @jp8 's PCBs once they are ready!

 

[joulupukki]

forgive the basic stuff here, but have you cleaned the capsule pin and high z section with repeated scrubs of isoprop and distilled / purified water? I've also found that removing any excess solder and reflowing what's existing at high z areas lowers noise, my hunch (and nothing more) is that this lowers the amount of flux chemicals in the area. Assuming components are relatively similar between MP and GW boards, and housing is the same, PCB layout and/or composition may be the biggest culprit. I'm interested in hearing/buying @jp8 's PCBs once they are ready!

Thanks. I will go back on one of my mics and re-clean that high-z section and see if it improves.

 

[klem]

IMHO this is a nightmare for anyone with the slightest OCD tendencies. I've no way to see contaminants beyond a certain level, it all just feels like guess work, and so easy to re-contaminate at any point in build or cleaning - a lot many chemicals leave their own residue, and isoprop still requires some firm elbow grease to remove some things, just a PITA overall. It gets discussed here from time to time but I don't rmember taking away any significant newfound techniques or insights.
You've shown impressive diligence in your troubleshooting in this thread, so if you can't resolve this there's just no hope.

 

[jp8]

If brushing and the flushing the PCBA with IPA does not help to remove flux residues, then try dedicated flux remover. I use Chemtronics Flux-Off, part no ES835BE. If you are sure your solder ia a no-clean version, you may want to use another type. This cleaner is optimized for removal of rosin based fluxes.

Assuming a clean board, the next big contributor of noise are the G-Ohm resistors. You could try 2, 3 or 5G resistors.

JFET noise does not affect total noise as many people think they do. @Voyager10 proved that in his JFET shootouts. But allegedly, some people experience widely different noise levels from JFETs and select them for low noise. They could also simply be damaged, e.g. due to ESD.

@joulupukki : do all mics have the same gain? Or did you normalize the noise signal in your DAW? I know you have been experimenting with the drain-to-gate feedback capacitor, so they could have different gains. Then if you did not normalize the signal and kept preamp gain the same for all mics, they will for sure have different noise levels.

@joulupukki tested my first KM84 build and confirmed its RFI robustness. I made some minor changes and also worked on other KM84 style circuits, which I hope to present soon on my website.

Jan

 

[joulupukki]

If brushing and the flushing the PCBA with IPA does not help to remove flux residues, then try dedicated flux remover. I use Chemtronics Flux-Off, part no ES835BE. If you are sure your solder ia a no-clean version, you may want to use another type. This cleaner is optimized for removal of rosin based fluxes.

I used Kester leaded solder (same stuff I've used for building tube amps). After making my post I compared my two KM84 clones and one of them was giving louder self noise than the other. After cleaning it up a bit they're now both on the same page. So maybe what I need to do once I decide I've got them both in a good spot is to make sure they're clean and then use the conformal coating I've got.

Assuming a clean board, the next big contributor of noise are the G-Ohm resistors. You could try 2, 3 or 5G resistors.

That's interesting. I could potentially experiment with that at some point.

JFET noise does not affect total noise as many people think they do. @Voyager10 proved that in his JFET shootouts. But allegedly, some people experience widely different noise levels from JFETs and select them for low noise. They could also simply be damaged, e.g. due to ESD.

@joulupukki : do all mics have the same gain? Or did you normalize the noise signal in your DAW? I know you have been experimenting with the drain-to-gate feedback capacitor, so they could have different gains. Then if you did not normalize the signal and kept preamp gain the same for all mics, they will for sure have different noise levels.

Yes, they've got the same gain. I think I was dealing with a mic that may have gotten dirty/etc. I'll have to re-do some of my comparison recordings with it and see if it's improved.

@joulupukki tested my first KM84 build and confirmed its RFI robustness. I made some minor changes and also worked on other KM84 style circuits, which I hope to present soon on my website.

Jan

I can confirm that Jan's builds are definite improvements in both RFI rejection and a large reduction in self-noise ... pretty much on par with the low self noise of a Neumann KM-184 I tested with yesterday. Right now, if I were to pick the "best" performing mic that is most likely closest to the sound of a KM84, it'd be Jan's KM84 clone board that he's built for the Takstar CM-63 body. The one for the CM-60 is a very close second.

IMHO this is a nightmare for anyone with the slightest OCD tendencies. I've no way to see contaminants beyond a certain level, it all just feels like guess work, and so easy to re-contaminate at any point in build or cleaning - a lot many chemicals leave their own residue, and isoprop still requires some firm elbow grease to remove some things, just a PITA overall. It gets discussed here from time to time but I don't rmember taking away any significant newfound techniques or insights.
You've shown impressive diligence in your troubleshooting in this thread, so if you can't resolve this there's just no hope.

Cleaning is definitely a good start and I think that was the main issue. I'll keep an eye on it and if that was it, perfect – relatively easy to solve!

 

[Voyager10]

JFET noise does not affect total noise as many people think they do. @Voyager10 proved that in his JFET shootouts. But allegedly, some people experience widely different noise levels from JFETs and select them for low noise. They could also simply be damaged, e.g. due to ESD.

Right - so what I measured for a 'charge amplifier' circuit like the KM84 is here. The FETs do show some differences above 3KHz or so, but below that the noise level is mostly determined by capsule capacitance and the bias (1G) resistor value - it's the laws of physics at play here.

I have found occasional outliers (the 2SK117 in this particular circuit, and a batch of 2SK170 of dubious origin from eBay) - I wouldn't say "there's no such thing as a bad FET", and you might just have got unlucky.

If you're thinking it might be the FET, try a BF256B or J113 from a reputable source (in the UK I mainly use CPC/Farnell or RS), they're just a few pence each.

Edit: I should add - I'm not obsessive about cleaning, my test boards were cleaned with IPA so no visible flux remained, but that's all. Also, the FETs were inserted into sockets, not 'air wired'.