MXL 990 to KM84-ish circuit with RK47

[SparkleBear]

Hey all,

Im wanting to build a little mxl 990 to km84ish circuit... but want to use a LDC that wants more polarization voltage than the 46v the circuit delivers the capsule. Should I use a DC-DC converter for this situation? Are there some resources I should check out to research this situation?

 

[Khron]

Well, it's not like you have too many other options... Unless you wanna cook up your own external power supply, and swap the XLR insert in the mic for a 4 or 5 pin one, yadda-yadda-yadda...

 

[RuudNL]

Did you realize that the output with a polarisation voltage of 60 volts is only 2.3 dB higher than with 46 volts?

20 log (60 / 46)

 

[SparkleBear]

Did you realize that the output with a polarisation voltage of 60 volts is only 2.3 dB higher than with 46 volts?

20 log (60 / 46)

I don't know much about it... Maybe its not worth it...

Well, it's not like you have too many other options... Unless you wanna cook up your own external power supply, and swap the XLR insert in the mic for a 4 or 5 pin one, yadda-yadda-yadda...

But if i was to do such a thing do you have a specific part in mind that would be good and low noise for microphone application? Take the 46v right at the end of the line before the capsule and bump it up another 15-20v?



Here is the NON DC-DC KM84 PCB I came up with. Definitely want to do a second one with DC-DC to have sound test.
.

It will be nice to see how much of an improvement it would be over the original circuit.

 

[Mworthin]

Haven’t tried this, but Chunger sells a board by Matador to do just that: U87 "AI" booster conversion PCB

 

[Mworthin]

Build thread. Official U87 "AI" Booster Conversion Board - Build and Support Thread

 

[Khron]

You're running some of those traces awfully close to the edge of the transformer cutout, and other unrelated pads. Have you run any DRC on that? (Design Rule Check)

That being said, there's no simple way about it, unfortunately, and it'll cost you quite a bit of board space and some possibly-not-entirely-jellybean components to get it done.

I did a similar thing in my sE X1, a few years back, though:

https://khronscave.blogspot.com/2017/05/se-electronics-se-x1-rebuild.html

 

[SparkleBear]

You're running some of those traces awfully close to the edge of the transformer cutout, and other unrelated pads. Have you run any DRC on that? (Design Rule Check)

That being said, there's no simple way about it, unfortunately, and it'll cost you quite a bit of board space and some possibly-not-entirely-jellybean components to get it done.

I did a similar thing in my sE X1, a few years back, though:

https://khronscave.blogspot.com/2017/05/se-electronics-se-x1-rebuild.html

I don't have a problem doing SMD for DC-DC. I think thats the only way to fit it all on there...

 

[SparkleBear]

I decided that the dcdc will take up too much room in the mxl body on the one pcb. MAYYYBE after a complete redesign I could fit it on two sided pcb… or if I could use big mlcc or smd lytics for the UF caps.

I might try a daughter board approach after I finish the first revision and I’m not thrilled with the output of the mic.

 

[SparkleBear]

Started to put the PCB together today. Got it built and cleaned up. But I could not get the circuit to pass audio... Can anyone see issues with the design/schema/build?



Im not sure about a few things. The ground scheme on the capsule might be different for the KM84 vs use of a LDC... I have the backplate of the capsule directly grounded like the km84. Will that work? I wired up an RK47 capsule. Front diaphragm connected to the c1/c2 junction. Backplate is going directly to PCB ground plane via through hole pad.






Im temporarily using my BV130 before my CM5722 come in the mail.

While making this post I realized that I connected the front diaphragm wire to the wrong point ( silk is wrong too ). It's supposed to be connected to the left turret, not the right. Now I am getting VERY low output signal with lots of noise. This is my first FET build from scratch so there is a learning curve.


Thanks so much for any advice or assistance figuring this out. CHEERS

 

[Khron]

For trimmers, it's "best practice", when used as a variable resistor (as opposed to a voltage divider), to connect the wiper to one of the ends. That's so, in case the wiper connection goes open, that node doesn't end up completely "floating".

DC voltages looking ok?

You probably can't measure the capsule bias voltage in a useful way, but might be worth suspecting the two 10meg resistors and/or the 1G R1.

I've had a similar noisy-low-signal issue with my Rode NT2-A - turned out to be a capacitor shorting the front diaphragm bias voltage to ground.

As mentioned before though, i'm really not thrilled with the routing of some of those PCB traces

 

[SparkleBear]

For trimmers, it's "best practice", when used as a variable resistor (as opposed to a voltage divider), to connect the wiper to one of the ends. That's so, in case the wiper connection goes open, that node doesn't end up completely "floating".

Interesting… so this is not a divider I have here is it? I have wiper to ground and one leg at the fet pin.

DC voltages looking ok?

Not sure. Don’t think so..


Probably can’t measure after the 10M resistors accurately. But the c4/r3/c2/fet1 junction is 2.2v when it should be 10v.

21.5v Point is measuring 19.33v
24v point is measuring 23v

You probably can't measure the capsule bias voltage in a useful way, but might be worth suspecting the two 10meg resistors and/or the 1G R1.

Suspecting how?
What can I check there?

I've had a similar noisy-low-signal issue with my Rode NT2-A - turned out to be a capacitor shorting the front diaphragm bias voltage to ground.

Interesting as well… so it was a bad capacitor? Or it was shorting another way?

RE: the traces you dont like @Khron , Was it just the traces near the edge cuts that sketch you out? or other things too?
I will fix that soon...this is revision 1. I will work on a few more layout things for revision 2. I dont like having the transformer wires on the corners and I think I will actually have to make a bigger profile hole for the transformer so it sets in the PCB. There is very little clearance inside the microphone. Also want to get a smaller trimmer or give it a bit more space from the FET.

I am going to also be looking at the DC-DC inside my u87. The daughter board is SO SMALL. I want to make something like this for my fet circuits... full DC polarization voltage at 60-70v!!!

What kind of tool or meter can I get to measure voltage in high z sections of microphone?

Bad fet maybe? I got a bunch just in case.

 

[Khron]

Interesting… so this is not a divider I have here is it? I have wiper to ground and one leg at the fet pin.

Well, no. Why would it be?

Your trimmer replaces R3 in the original schematic. Have you tried going through the whole range of the trimmer, while measuring the drain voltage? And while you're there, solder the wiper leg to the "free" leg of the trimmer, on the underside of the board.

 

[Khron]

Just a quick doodle in Paint with some thoughts.

- flip the trimmer 180deg and shift it to the right
- swap the placements of those two electrolytics, so you can then route those two traces under the transformer coupling cap (as opposed to "outside" it)
- that being said, you'd rather want those caps closer to the series resistors they're connected to (ie. on the other side of the board), but considering the minuscule currents involved, that may well not be an issue; just another "best practices" type thing
- why run traces so close to unrelated pads though, why route them so close to edges (like the transformer cutout), and/or why come off pads at an angle, if not necessary?
- also, since you're not exactly starving for board space, why not use thicker traces? Those look unjustifiedly thin, and it kinda freaks me out, especially when they're routed so close to things...

 

[SparkleBear]

Just a quick doodle in Paint with some thoughts.

- flip the trimmer 180deg and shift it to the right
- swap the placements of those two electrolytics, so you can then route those two traces under the transformer coupling cap (as opposed to "outside" it)
- that being said, you'd rather want those caps closer to the series resistors they're connected to (ie. on the other side of the board), but considering the minuscule currents involved, that may well not be an issue; just another "best practices" type thing
- why run traces so close to unrelated pads though, why route them so close to edges (like the transformer cutout), and/or why come off pads at an angle, if not necessary?
- also, since you're not exactly starving for board space, why not use thicker traces? Those look unjustifiedly thin, and it kinda freaks me out, especially when they're routed so close to things...

Thanks for all the thoughtful tips @Khron . I will definitely look at each of those points moving forward.

I’ve confirmed that everything in rev 1 is working. Next rev will have updates like proper mounting hole placement, better track routing and thicker tracks. I also think I want to swap out the trimmer for a right angle one so I can turn it without taking the bottom bell off. Also thinking about incorporating a pad switch with the 15pf feedback between the polarization line and the drain. ( like on the original schema ).

I ditched the dc-dc using matadors board design because the parts were just too massive for this build. But I’m thinking for the next one I want to look at the u87ai daughter board and make something like that with all small smd parts. About the size of a us quarter coin . That will give me more output… even though ruud pointed out it would only be a few db, I want to take on the challenge.

LOL sorry for freaking you out with my tiny traces.

I got my cm5722 transformer in the mail and put the whole revision together. It took me a while to realize that I had to turn the fet source resistor to get the fet to start doing its thing and pass audio ... *facepalm*
After this discovery it would pass audio reliably. I tried to follow the fet bias procedure. I injected signal into gate from daw and tried to get the clipping even and such. I had no idea really what was the correct way. I attempted to balance the distortion on top and bottom with test signal. However, after this procedure it would not pass audio from the capsule so I obviously did not do it correctly… this is my first fet mic project. Still getting my tool belt in order.

Also, please excuse the slight singe on the film caps. Woopsie.

Can anyone help point me In the right direction beyond what I learned in this post?
( Mic Biasing )

When trying to bias the fet, I was using the daw signal generator and an oscilloscope plugin. Is there anything I need to do to condition the signal to interact with the fet? Such as passing through a high value resistor or anything? Just measure the output voltage of the daw signal? Do I feed the signal straight into the gate? Or do I feed it where the capsule signal is fed? I was feeding it directly to the gate.

Any tips would be so helpful for dialing in the fet bias …

Cheers!

PS. No pictures of the installed board yet: forgot to take those before I went on this LA work trip.

 

[SparkleBear]

Ok! Made some edits to the design as per previous comments a la @Khron ...

Made the mount holes wider and more flexible for mounting possibilities... Thickened all of the traces and improved routing... Cleaned up silk and graphics. Fixed capsule arrow error... Connected the two legs on the trim pot.



Ill be back from LA next week and will post some pics of the rev1 build. I still need to figure out how to properly bias the transistor.

I just sent in rev 2 in to JLCPCB. Getting it in black! Stoked on that... also sent in a few more goodies that will post soon

 

[Khron]

You really have a thing for running traces as close as physically (or even theoretically) possible to each other, or unrelated pads, don't you? But if the DRC (if you ran any), or JLCPCB haven't complained... "Too close for comfort" anyway, at least for me

Also, any particular reason for that utterly humongous 250v output cap, in a 24v-powered circuit?

And since this is not an SDC, and thus one side of the capsule is not mandatorily connected to ground (through the case), you could go with the "ChinaMicMod" wiring (from micbuilders / Zapnspark), and have the capsule connected between the bias voltage and the JFET gate. That then rids you of one of the 1g resistors, as well as the C1 DC-blocking cap. But that then requires swapping the two output wires going to the XLR as well, to maintain the same polarity.

 

[SparkleBear]

You really have a thing for running traces as close as physically (or even theoretically) possible to each other, or unrelated pads, don't you? But if the DRC (if you ran any), or JLCPCB haven't complained... "Too close for comfort" anyway, at least for me

Lol, I guess only that one audio line is close… I tried to pull the traces further from edge cuts. I run drc but I guess my rules allow what I’ve been doing….. jlc has never complained. I have their rules loaded into my kicad.

Also, any particular reason for that utterly humongous 250v output cap, in a 24v-powered circuit?

Well the output cap is actually way bigger in the render than in real life. But I totally forgot that the output cap can be more like 75v rather than what I’m used to in a tube circuit. Thanks for the reminder…

Rev 2 will use a smaller cap perhaps… but I already sent it in and it’s in production… so I will just use a smaller cap in the same footprint

And since this is not an SDC, and thus one side of the capsule is not mandatorily connected to ground (through the case), you could go with the "ChinaMicMod" wiring (from micbuilders / Zapnspark), and have the capsule connected between the bias voltage and the JFET gate. That then rids you of one of the 1g resistors, as well as the C1 DC-blocking cap. But that then requires swapping the two output wires going to the XLR as well, to maintain the same polarity.

I’ll have to take a look at that… i basically followed the km84 circuit as I have not fully designed my own mic circuits from scratch yet.

Thanks for the support and such!

 

[SparkleBear]

Also, any particular reason for that utterly humongous 250v output cap, in a 24v-powered circuit?

Would you just use a 1uf 50v wima? One of those smaller yellow radial film caps 50-60v?

 

[Khron]

Would you just use a 1uf 50v wima? One of those smaller yellow radial film caps 50-60v?

Lowest usual standard voltage rating for those is 63v (or even 50v), so... Unless your confirmation bias is irrationally attached to that "bigger is better" type thing... Yes, a suitable-value small box film cap should be plenty indeed.

If you don't mind, i'll just highlight the spots on the board that "rub me the wrong way", and i might also then ask why