MXL 603 -> KM84 Replacement PCB

[joulupukki]

Couple of quick unpolished demo sound clips, unedited. These mics sound so similar, it's very hard to distinguish them. Do you hear a difference between the two?

Mic Parts SDC-84:
View attachment MP SDC-84 Music Test.mp3
DIY KM-84:
View attachment DIY KM-84 Music Test.mp3

I re-biased the DIY KM-84 to be closer-ish to the MP mic, but the MP mic has different clipping/distortion (probably better tuned) than the DIY version. My DIY KM-84 is now running at 10.15V.



I know very little about VO, but here are two speaking tracks with each mic.

MP SDC-84:
View attachment MP SDC-84 VO Test.mp3

DIY KM-84:
View attachment DIY KM-84 VO Test.mp3

Frequency sweep comparing the MP SDC-84 with the DIY KM-84 loaded with different capsules:
View attachment 1707837082709.png

 

[jp8]

Both the MP and the DIY versions share the same circuit topology and have capsules with almost identical FR, so I guess this explains the similarity in sound. It would have been nice if we could compare the sound of your mics with a stock CM-60 or CM-60 body with an MP or 3U capsule. But you don't have a stock CM-60 anymore, do you?

Jan

 

[joulupukki]

Both the MP and the DIY versions share the same circuit topology and have capsules with almost identical FR, so I guess this explains the similarity in sound. It would have been nice if we could compare the sound of your mics with a stock CM-60 or CM-60 body with an MP or 3U capsule. But you don't have a stock CM-60 anymore, do you?

Jan

I bought two Takstar CM-60 mics and haven't converted the second yet (but probably will soon since I can always get another CM60 for around $20). Here's a frequency sweep with the Takstar CM-60 circuit with the three different capsules:

 

[joulupukki]

Looks like pretty much the same shapes using the DIY KM-84:

 

[jp8]

Clear, thanks!

Any chance you could add recordings similar to the ones in post #221, but then with a stock CM-60 mic? And a recording with either of the MP or 3U capsules on the stock CM-60 body? No problem if you don't want to do that. I was just curious whether the differences between the capsule types and Schoeps vs KM84 electronics would be audible or not.

Jan

 

[joulupukki]

Clear, thanks!

Any chance you could add recordings similar to the ones in post #221, but then with a stock CM-60 mic? And a recording with either of the MP or 3U capsules on the stock CM-60 body? No problem if you don't want to do that. I was just curious whether the differences between the capsule types and Schoeps vs KM84 electronics would be audible or not.

Jan

I posted some comparing the CM-60 with the MP SDC-84 in post #20 here: https://groupdiy.com/threads/what-makes-a-good-microphone.86443/post-1130954

I can definitely hear a difference. I don't know if one is "better" than the other – probably depends on the application. The CM-60 has more low frequency content.

 

[joulupukki]

I converted the other Takstar so now I theoretically have a matched pair of DIY KM-84 mics. Not sure if that's actually true, but this quick test (both using 3U capsules and 3U GZT-84 transformers) says they're pretty darn close.


And here's a test comparing the Mic Parts SDC-84 stereo pair with the DIY KM-84 Stereo Pair. The MP edges out the DIY in sensitivity:

 

[jp8]

I posted some comparing the CM-60 with the MP SDC-84 in post #20 here: https://groupdiy.com/threads/what-makes-a-good-microphone.86443/post-1130954

I can definitely hear a difference. I don't know if one is "better" than the other – probably depends on the application. The CM-60 has more low frequency content.

Thanks for pointing me to that thread. I hadn't read that one before, so I missed it. Anyway, I agree with your opinion about the low-frequency content. I also had a listen to your shootouts from post #20 of this thread (). I found the best and most revealing parts of these recordings in the sections where only the SDC under test was being played back. To my ears, the clear winners were the DIY KM84 and MXL991 having the 3U capsule on, where your off-axis voice definitely sounded most natural. I found it hard to tell the KM84 from the 991, though. Just my €0.02...

Jan

 

[joulupukki]

Probably like a lot of things, once you reach a certain level of good, it's not about which is better, but which one you prefer?

I've got a temporary solution for covering the Takstar's low cut switch (and it'll likely be the permanent solution):

 

[joulupukki]

If Graeme ever updates the layout of the PCB to be different, it’d be nice if it were possible to more easily swap out the JFET for a different one without having to desolder any other components. I’d love to try out a different pair from my batch of 2N3819 JFETs or even try out a different JFET entirely, but the big blue trim pot is in the way on the back.

Maybe it could be desoldered from the top. Has anyone done that before? Just curious. I don’t have plans to do this, but just wondered … for the future.

 

[jp8]

Just curious: what differences do you expect the other 2N3819s to have on the sound or on THD or other measurements? Their specs are quite close and all of them are close to the typical datasheet values. The small differences will have some effect on open-loop gain and distortion, but the feedback around this part will iron out these differences; the closed loop gain and distortion will be almost identical with any of these JFETs. Maybe they will have slightly different noise levels, but that's all I would expect. I would be surprised if you would hear/measure a significant difference. But let's see: perhaps I've got it all wrong, overlooked something and there will be audible differences.

Jan

 

[Migs 31]

Maybe it could be desoldered from the top. Has anyone done that before? Just curious. I don’t have plans to do this, but just wondered … for the future.

Then you use these (see pics).
And you only have one pin to desolder from the top (Gate - R2 -C1).
Just make sure your soldering and solder tip are ultra clean.

But as jp8 said. I wonder if you will see much difference between two - three different 2N3819’s.
Maybe between different types, like a BF256 or others?

M

 

[joulupukki]

@Migs 31 that’s clever! Wish I would have thought of doing something like that before.

Just curious: what differences do you expect the other 2N3819s to have on the sound or on THD or other measurements?

Mostly I’m just curious how a different pair of 2N3819 FETS I have would change. The ones I have in there now I measured before installing to have an Idss of 10.1mA and Vp of 3.4V. I’m curious to know if the other pair I have with an Idss of 11.8mA and Vp of 3.8V would result in more output volume from the mics … or whatever other change. Just curious.

The MP 84 mics have 2-4dB more output at the same preamp level than these DIY mics and I’d like to understand why. Of course, the FET is a J305, the polarized cap values are much higher, and drain and source voltages in the MP are quite a bit different … along with a couple of extra caps that the DIY doesn’t have. I’m sure that all contributes to increased output and different THD behavior.

 

[jp8]

I think the gain difference can be explained by:

A) different transformer ratio (+0.5dB for the MP transformer)

B) Different values for the feedback capacitor C2. In Graeme's circuit it is 4 pF. Original KM84 has 3.3 pF. According to simulations, this yields an extra 1.65 dB of gain. If MP took an even smaller capacitor value, the gain would be even higher.

To a certain extent, the circuit is designed to be insensitive to JFET and other circuit differences. Even if you would swap the 2N3818 by a J305 and re-adjusting the DC bias, the closed loop circuit gain would only differ by less than 0.1dB.

Jan

 

[joulupukki]

I think the gain difference can be explained by:

A) different transformer ratio (+0.5dB for the MP transformer)

B) Different values for the feedback capacitor C2. In Graeme's circuit it is 4 pF. Original KM84 has 3.3 pF. According to simulations, this yields an extra 1.65 dB of gain. If MP took an even smaller capacitor value, the gain would be even higher.

To a certain extent, the circuit is designed to be insensitive to JFET and other circuit differences. Even if you would swap the 2N3818 by a J305 and re-adjusting the DC bias, the closed loop circuit gain would only differ by less than 0.1dB.

Jan

Really good to know. Thanks for explaining! My MP mics I left the C2 capacitor out (an option in their instructions) to have the mic be more sensitive for things like acoustic guitar. Yeah, my GZT-84 tx is 7:1 vs the 6.5:1 in the MP. The MP has at least one more cap that the km84 circuit doesn’t have. Looks like a 0.022uF ceramic from the FET’s source to ground (as far as I can tell). Not sure what effect that has.

 

[jp8]

Without C2, you'll get the maximum gain the circuit allows. And maximum THD. Without C2, I would have expected it to have more than 10 dB extra gain, not the 2-4 dB you mentioned. I'm puzzled...

The 0.022uF Source to ground capacitor is in parallel to the source resistor R3 and polarized cap C3. The sense or nonsense of parallelling a small cap with a polarized cap has been discussed recently in another thread. Anyway, in this circuit it will not affect the gain.

Jan

 

[joulupukki]

Without C2, you'll get the maximum gain the circuit allows. And maximum THD. Without C2, I would have expected it to have more than 10 dB extra gain, not the 2-4 dB you mentioned. I'm puzzled...

Interesting. I wasn’t taking exact measurements … and was just measuring via the output of my preamp, and not at the drain. Not sure if that’d make a difference. If I can figure out a way to cleanly do it, I may try a 1pF or 2pF cap at some point. Or just leave it as-is since it seems pretty decent.

About the only main difference I’m hearing sonically is that this DIY version seems a little warmer than the MP. The MP has a little more high end definition than the DIY. Not sure why – most likely the transformer?

The 0.022uF Source to ground capacitor is in parallel to the source resistor R3 and polarized cap C3. The sense or nonsense of parallelling a small cap with a polarized cap has been discussed recently in another thread. Anyway, in this circuit it will not affect the gain.

Ok, good to know. I tried searching around to find that thread so I could understand more, but couldn’t find it.

 

[jp8]

As to the difference you hear between the MP and the DIY, I have to make some guesses. Could be the transformer, like you mentioned. But to be honest, I have very little experience with audio transformers and I am still in the learning phase. That is why I am now in the process of designing several transformer based circuits I'd like to experiment with. I'm almost finished now with a KM84 based circuit with polarization voltage generator, which gives you more freedom to increase the JFET current and have a polarization voltage higher than 46V. Both with the idea to reduce noise. And THD should be reduced as well, I expect. The PCBA will be for the CM-63.

But getting back to the perceived differences between the MP and DIY: what I do know of transformers is that they have rising THD at lower frequencies. That could be responsible for the warm sound. The MP on the other hand, will have more even order distortion throughout the whole audio band due to the removal of the feedback capacitor. This distortion is best heard in the upper mid frequencies, where our ear is the most sensitive. I can imagine this may sound nice on a guitar. As with any effect, if applied in moderation it can have a benign effect on the sound.

The thread I was referring to:
https://groupdiy.com/threads/parall...er-capacitors-for-optimising-the-sound.86394/

Jan

 

[joulupukki]

Thanks for posting that thread. I didn't really "get" most of it, but that's fine. Besides the difference in dB output between this DIY and the MP 84 mics, the DIY has less high end & definition using the 3U capsule and 3U GZT-84 transformer. It's not much, but it's definitely there. I'm thinking at this point, the main difference would be in the transformers used. I may try (temporarily) moving one of the MP transformers onto the DIY board just to see how that'd affect things (and confirm my suspicion). MP doesn't sell their transformers separately, but maybe the CM-5722 might give more defined output?

 

[doqmemory]

MP doesn't sell their transformers separately

Matt will likely sell the transformer separately, if you email him.