Getting 6-8 fewer dB out of a new DIY KM84 clone in a stereo matched pair

[joulupukki]

R3 is supposed to be a trimmer pot instead of a resistor. Usually its a 10 turn bourns potentiometer.
This is a simple circuit. I wouldn't bother picking up a pre-made board unless its only a couple of bucks different from a proto board.

Yeah, I guess it depends on whose kit you're using to build. In the case of the micparts kit, it's not a trimmer but a specific resistor value.

On the board, I'm saying that ... if I wanted to build one from scratch, rather than order a pre-made board, if someone has Gerber files available, one could order themselves a board using the "prototype" pricing (where they build 5 boards and send them to you for cheap).

 

[drtechno]

Yeah, I guess it depends on whose kit you're using to build. In the case of the micparts kit, it's not a trimmer but a specific resistor value.

That's their fault they are not copying one of the copies and using the original design that you have to match everything for a mirrored pair. Other people that use that circuit use a trim potentiometer so they can match them.

 

[kingkorg]

Looking at the schematics, R3 should be adjusted to match gain.
Also, if you use a single turn potentiometer, make sure it goes to the other pin. Some will have resistance at the bottom of the wiper differently from each other. This should be a multi-turn potentiometer anyways.

R3 is for biasing, lowest possible THD and symmetrical clipping. It will somewhat affect gain, but not to the extent described in the OP. Judging by the voltage measurements, there was nothing wrong with that resistor.

 

[drtechno]

R3 is for biasing, lowest possible THD and symmetrical clipping. It will somewhat affect gain, but not to the extent described in the OP. Judging by the voltage measurements, there was nothing wrong with that resistor.

Hard to say. If I was into building them I would get a dozen or so of those fets to match them for a set. That is why I think some kits are not that great sounding compared to someone getting one from a boutique builder. Because I think everyone including me would use different parts than what microphoneparts.com does .

 

[joulupukki]

Hard to say. If I was into building them I would get a dozen or so of those fets to match them for a set. That is why I think some kits are not that great sounding compared to someone getting one from a boutique builder. Because I think everyone including me would use different parts than what microphoneparts.com does .

According to MP, the JFETs for these two microphones were matched and biased before they were sent to me (part of the extra charge to get a matched set). So ... probably just something I did wrong during installation. Strangely, the mic with lower output was the 2nd one I built and I was even more careful not to use too long of heat when soldering in the JFET. So I dunno what could have gone wrong. Maybe there's some other component that's not good. We should hopefully find out within a week.

 

[drtechno]

According to MP, the JFETs for these two microphones were matched and biased before they were sent to me (part of the extra charge to get a matched set). So ... probably just something I did wrong during installation. Strangely, the mic with lower output was the 2nd one I built and I was even more careful not to use too long of heat when soldering in the JFET. So I dunno what could have gone wrong. Maybe there's some other component that's not good. We should hopefully find out within a week.

JFETs can change when you solder them into the board. Because I've been there with JFETs before matching it for a 1178 build and tried using a test circuit to match them.

 

[Chrisfromthepast]

For fun I was putting together sourcing all the components to build a DIY from a donor mic body/etc. and for a pair of these it was still gonna be between $500 - $600. Crazy how it all adds up. That was if I were to use a 3U transformer and capsule and Dale CMF55 resistors & WIMA film caps from Mouser ... and would still need to source or build a PCB.

You could go much cheaper on the passives if you reference this bom from another KM84 project.

BOM: https://www.mouser.com/ProjectManager/ProjectDetail.aspx?AccessID=f5036488e2

Schematic: DIY KM84 - Schematic.pdf

You can substutute the xformer for NTE10:3 and save a tad as well.
You can use the cm60 and get a donor body and pretty good capsule for $25.
I still prefer the 3u stuff, but you can make a nice mic for less than $200.

 

[joulupukki]

CM60. Wow, yeah, pretty cheap. I’d also need to find / make a PCB.

 

[ChuckHanning]

I had a similar problem for two mic and mod km84s I bought. I was biasing to be close in voltage but the transconductance of the feta was different.
It was suggested to bias them a different way. Turn the resistor all the way down while feeding a 1k tone at -50 dB through the mic and monitor output. Bring the output level to the same the back down just a half turn. Do the same with the other mic and you can get them within a dB of each other.
The source voltages were different (3.2 vs 4.8 v) but they matched by a dB which is close enough for me. Interestingly when I measured distortion using REW they both were much lower this way that the other method I was originally using.

 

[pasarski]

I had a similar problem for two mic and mod km84s I bought. I was biasing to be close in voltage but the transconductance of the feta was different.
It was suggested to bias them a different way. Turn the resistor all the way down while feeding a 1k tone at -50 dB through the mic and monitor output. Bring the output level to the same the back down just a half turn. Do the same with the other mic and you can get them within a dB of each other.
The source voltages were different (3.2 vs 4.8 v) but they matched by a dB which is close enough for me. Interestingly when I measured distortion using REW they both were much lower this way that the other method I was originally using.

I've been playing a lot with km84 type of circuit in a simulation lately, and yes, you will get really close to a bias point with lowest distortion at moderate levels this way (bringing the bias resistance up to the value when the mic gets as loud as it gets). If you want maximum headroom before clipping you need colder bias (larger resistance).