Banzai's KM84 DIY Body & PCB kit build thread

[CalGTR]

OK, so I'm having some issues, maybe, with biasing. I'm ending up with some strange results, seems to me. I'm basically seeing max headroom with max source resistance (20k). I am never able to see symmetrical clipping, regardless of trimpot settings. It's always the bottom that flattens out first, and backing off the source resistance only makes the bottom of the wave worse, all the way until the FET shuts off.

If I back the trimpot off to about 18.2kOhm, I can improve the distortion such that for a 200-300 mV input signal, or thereabouts, I see below .5% THD on the output, which seems reasonable enough to me (but then again I'm a bit clueless as to what to expect). Moving off of this particular bias point it either direction results in more distortion. The source-ground voltage I'm seeing at these higher source resistances is only about 4.5V, which is miles off of the 10V or so I thought I should be expecting.

Any of this seem reasonable to anybody, or should I be hunting for problems somewhere in the build? Should I seek a different FET that might be closer in spec to what one might usually expect? Any comments/advice will be appreciated.

Thanks,

Calvin

It seems to me that i had the same experience. Just set for minimal THD as you did before and you should be good.

Thomas

Thanks. I appreciate your input. Right now, adjusting for minimal THD is the plan.

I'd be interested to hear whether anybody else has ended up with this seemingly extreme trimpot setting and/or low source voltage.

 

[kato]

I'm planning to use cinemag's 8:1 (2510) instead of their BV107-alike. Is there anything majorly different except the copper tape to ground the laminate stack and the ratio? I know my output will be slightly lower, but I plan to use these for drum overheads almost exclusively so that's a feature not a bug.

https://cinemag.biz/mic_output/PDF/CM-2510.pdfhttps://cinemag.biz/mic_output/PDF/CM-5722-Data-Sheet.pdf
Is the copper tape literally just a piece of tape stuck on the laminate? Any reason I can't just sick a piece on there myself?

(I'd prefer not to have to order the "right" transformers, Cinemag is slammed right now and it took me a very long time to get the last few I bought.)

EDIT: Eh, soldered them up. Somehow managed to kill two FETs! Or maybe they were dead to begin with, they were a matched pair I had in my own stash. I replaced them with the pair that came with my kits, which actually turned out to be very very close. I don't know if that was intentional but it saved me having to measure a bunch looking for a pair. These sound really nice, too bad I finished too late to do some drum takes.

Hey Jon, I was thinking about trying the Cinemag CM-24110 (4:1 stepdown) as used in Ben Sneesby's Lulu microphone for this project. It would work, right?

But then I saw [ your thread about the center tap acting weird ] on those and I'm confused. Is that center tap necessary in this circuit?

You'd seemed resigned to shelving those CM-24110s for a while. Any reason you couldn't use one in this mic?

I'd like confirmation that they'd work in this circuit before I place an order with Dave. Any thoughts? Thanks, Matthew

 

[kato]

Hey Jon, I was thinking about trying the Cinemag CM-24110 (4:1 stepdown) as used in Ben Sneesby's Lulu microphone for this project. It would work, right?

Watch part 1 of Jon's FET Mic Build Diary. I think he answers your question sufficiently if you watch the whole thing.

 

[McIrish]

I built two of these yesterday and am biasing the FET today. Getting a bit of a puzzling reading. The trimmer is a 20K trimmer and it seems I need to have it set at 20K for symmetrical clipping. That seems wrong somehow. Both are acting identical. Is there a possibility that the PCB is marked wrong? I will go back and reread all 12 pages again but maybe someone knows the answer.

I'm using the 3U transformer wired like this:
R - Red
G - Brown
B - Orange
Y - Yellow

Also, the pinout of a 2N3819 would be on the board backwards from the the way it is drawn. Is it possible that the fet goes in backwards? Man... I'm feeling kind of stupid here but I've built so many things and I've been a tech for 30 years. I must have messed up something but for the life of me I can't imagine what it would be.

 

[CalGTR]

I built two of these yesterday and am biasing the FET today. Getting a bit of a puzzling reading. The trimmer is a 20K trimmer and it seems I need to have it set at 20K for symmetrical clipping. That seems wrong somehow. Both are acting identical. Is there a possibility that the PCB is marked wrong? I will go back and reread all 12 pages again but maybe someone knows the answer.

I'm using the 3U transformer wired like this:
R - Red
G - Brown
B - Orange
Y - Yellow

Also, the pinout of a 2N3819 would be on the board backwards from the the way it is drawn. Is it possible that the fet goes in backwards? Man... I'm feeling kind of stupid here but I've built so many things and I've been a tech for 30 years. I must have messed up something but for the life of me I can't imagine what it would be.

If you see my post above, you'll see that I'm having pretty much the same experience. Max headroom at 20k (although I can't quite actually ever achieve symmetrical clipping). I've also wired up the 3U transformer the same way you did. What I ended up doing was forgetting about symmetrical clipping and concentrated on dialing in lowest THD at an input voltage of about 100-200mV p-p. My current trimpot setting is something like 18.2K or thereabouts. I'd love to hear others chime in, because these source resistance (trimpot) values seem out of whack. I think I may end up trying another FET to see if it biases up any differently.

 

[McIrish]

Just put the FET in the opposite way, which would be correct if the schematic and data sheet are correct. That would mean that the PCB legend is wrong. Either way, I got the same exact results. With the trimmer at 20k i get the least distortion and most gain. Also, with that value, the drain voltage is at ~10V, which is supposedly correct. I'd sure love to hear from others. Very frustrating.

Update: Best linearity was at 21.5K. I used a different trimmer to get there. Put a 22K in there and it tests OK. The PCB legend, schematic and a data sheet on the 2n3819 can't ALL be correct. The legend does not match up to the pinout from the datasheet. I think the board legend is backwards, though it works either way. I have one FET in reversed of the legend and both mics act the same. Hard to believe everyone else is not seeing this issue.

Second one has been recalibrated. This one required an 18.5k resistance. I used two resistors to get there. Both mics have the same output and have no visible distortions before they begin symmetrically clipping. Now to try them out with capsules.

Thinking about it, it is possible that the 2N3819 FETS sent with the kits have a different gain. Who knows. I'm fairly confident in the values I selected for the bias. It all looks good.

 

[Zim]

If you see my post above, you'll see that I'm having pretty much the same experience. Max headroom at 20k (although I can't quite actually ever achieve symmetrical clipping). I've also wired up the 3U transformer the same way you did. What I ended up doing was forgetting about symmetrical clipping and concentrated on dialing in lowest THD at an input voltage of about 100-200mV p-p. My current trimpot setting is something like 18.2K or thereabouts. I'd love to hear others chime in, because these source resistance (trimpot) values seem out of whack. I think I may end up trying another FET to see if it biases up any differently.

I‘ve had a similar issue - couldn’t get symmetrical clipping, and got best results (by ear) with the trim pot maxed out to 20k.

Because I was using a signal from my DAW via a cable I hacked together, albeit following the instructions I found somewhere here, and then using the oscilloscope in REW, I’m so far assuming it’s something I’m doing wrong. So I’ve ordered a signal generator, which will at least reassure me that the signal I’m sending is the ‘proper way’ and hoping the granularity that will provide re voltage will sort the problem. Still planning to use REW to monitor, but first things first.

Sort of reassuring to hear others are having similar results to me but, like I say, this being my first mic build and not having the proper measurement kit I wouldn’t be at all surprised if it’s just driver error on my part at the moment. Will post back if the sig gen helps (although that seems to have been delayed another 2 weeks which is irritating!!)

interested to hear how others are getting on...

 

[kato]

I‘ve had a similar issue - couldn’t get symmetrical clipping, and got best results (by ear) with the trim pot maxed out to 20k.

Maxing out over 20k seems inconsistent with all the folks from past group buy successes. A couple of examples come to mind:

FuzzFace said he "set the trimmer to 3.9K in order to get the FET at least in the ballpark for biasing. When I connected again to my DAW, everything was much more manageable & Since this uses a 20K trimmer, if you leave it at the center you have about a 10K source resistor, which in my experience is high enough to make this thing go wild…" I reckon if 10k is enough to make it go wild, >20k is extra wild.

Also, BramK got his pair biased "at 6.73K on mic 1 and 7.96K on the other."

I wonder if your FET is out of whack, or possibly one of the rebadged counterfeits?

 

[McIrish]

Whatever the resistor value, to get full gain without distortion, the FET needs to be biased so the voltage at the drain is around 10v. In my case, with the bias resistor lower, the distortion was extreme. Once mine was in the 18k-22k range, it worked as expected and the voltage was correct at 10v.

 

[Banzai]

The PCB legend, schematic and a data sheet on the 2n3819 can't ALL be correct.

They can

2N3819 drain and source are interchangeable. You can install it any way round you prefer.

FET's included with this batch are the Central 2N3819's from Mouser. Previous batches were shipped with Fairchild's.

I'm using the 3U transformer wired like this:
R - Red
G - Brown
B - Orange
Y - Yellow

Transformer pad legend is for the 8-wire Cinemag 5722 (5722W wasn't yet available)

See post #3 here for 3U wiring.

 

[McIrish]

For the 3U wiring, it just seems that the left to right leads were reversed but the primary and secondary are still the same. Red and Brown are primary and Yellow and Orange are secondary. I don't see how it would have any effect on the circuit since both are swapped. I think I will leave mine as the leads would be too short to change them.

Thanks for the clarification on the FETs. I wouldn't think they would be that different but apparently they are.

 

[CalGTR]

For the 3U wiring, it just seems that the left to right leads were reversed but the primary and secondary are still the same. Red and Brown are primary and Yellow and Orange are secondary. I don't see how it would have any effect on the circuit since both are swapped. I think I will leave mine as the leads would be too short to change them.

Thanks for the clarification on the FETs. I wouldn't think they would be that different but apparently they are.

That is indeed a nice clarification on the FETs. I've ordered a few 2N3819 from Small Bear. No idea which manufacturer, but I'm hoping they'll be something different than Central so I can see what happens if I swap brands. If/when I do swap FETs, I'll report back.

As for transfo wiring, I can't see any reason why simply reversing the two primary leads and the two secondary leads would make any difference. I wired mine up the same way you did. If anyone can think of a reason why wiring up the 3U transformer this way is problematic, by all means chime in.

 

[kato]

FET's included with this batch are the Central 2N3819's from Mouser. Previous batches were shipped with Fairchild's.

Arg, sorry, I forgot you supplied the FETs for the kit. I did not mean to imply they were substandard.

Interestingly, I found a post from Jim Williams mentioning that when Fairchild put the 2N3819 back into production, the specs were all over the map. So I think we should all be willing to try out different FETs or possibly try using 25k trimpots.

 

[midwayfair]

Hey Jon, I was thinking about trying the Cinemag CM-24110 (4:1 stepdown) as used in Ben Sneesby's Lulu microphone for this project. It would work, right?

But then I saw [ your thread about the center tap acting weird ] on those and I'm confused. Is that center tap necessary in this circuit?

You'd seemed resigned to shelving those CM-24110s for a while. Any reason you couldn't use one in this mic?

I'd like confirmation that they'd work in this circuit before I place an order with Dave. Any thoughts? Thanks, Matthew

Hey, sorry I'm not around much these days and didn't see this.

Anyway -- Cinemag makes the STOCK version of that transformer as a replacement for a microphone that uses the center tap on the "1" side, whereas to use it as a step-down with the center tap as the phantom supply, we need it on the "4" side. If you order the transformer from Dave you can just specify you need the center tap on the high side and they'll wind it that way. (The ones I just got from them actually have the center tap on both sides, though, so you still need to check which is which.)

The last post I made in that thread you linked mentions this, I don't know if you read that far.

By the way if it helps anyone, I also maxed out my trim pot when experimenting by ear, even though it put the FET a little above half supply (and much higher than the 10 or 11V that's typical). If I had known I was likely to do this ahead of time, I might have increased the value of the trimpot to get a little more range. Eventually I might bias it "properly" but they sound super crisp with the drum kit so I'm inclined not to mess with a good thing.

 

[Zim]

Right then, so back to biasing... my signal generator arrived, so had another go hoping the increased granularity would help. It certainly makes a difference in terms of knowing exactly what signal I’m sending the mic, but hasn’t actually sorted the biasing Here’s what I got...

Previously I had the resistor wide open for the best results by ear. Since then I finished my second mic so decided to use that one in case there was something specific about the first mic that was causing the results. I had to use the transformer out of the first mic though (AMI T8) as I don’t have my second transformer yet.

Once again, only with the resistor wide open (Ie. turned fully counter clockwise) do I get the best results, but now I can put some actual figures and pictures to it.

with the resistor fully CCW the max voltage I can get at R4 is 9.64V, so just short of the 10V in the circuit diagram. Clipping starts on the top waveform at 350mv peak to peak


it starts leaning off to the left slightly, and turning the resistor clocwise (the only option available) just exagerates the clipping on the top of the waveform and has no effect on the bottom of the waveform. With the resistor fully CCW again I continued to increase the signal - here it is at 500mv p/p: (the signal generator was set to 75ohms btw)


Any thoughts / recommendations from anyone?

thanks

Zim

 

[JW]

Can't folks having trouble biasing just add a 5K series resistor with the pot to dial in the correct value? That would be my first instinct if 20K doesn't seem to do it.

 

[kato]

Anyway -- Cinemag makes the STOCK version of that transformer as a replacement for a microphone that uses the center tap on the "1" side, whereas to use it as a step-down with the center tap as the phantom supply, we need it on the "4" side. If you order the transformer from Dave you can just specify you need the center tap on the high side and they'll wind it that way. (The ones I just got from them actually have the center tap on both sides, though, so you still need to check which is which.)

The last post I made in that thread you linked mentions this, I don't know if you read that far.

I did read that far! However, I'm not sure why a person would even want a phantom center tap on a mic output transformer. I'd probably just skip it unless it was required to make the transformer behave properly.

It's even more mind-boggling to me that STOCK, it would be for a 1:4 step-up configuration, because I've never seen a mic schematic that called for a 1:4 step-up on the output. I'm a little newbish in this dept. I'm stuck between using the safe bet 7:1 from the original KM84, which I know will work, or experimenting and possibly learning something new.

Anyway, thanks for chiming in. I recently discovered your open-source fet mic build diary video, and I realized I recognize you from the stompbox world! Your FET mic design looks like the most promising DIY mic project out there! (I'm definitely building one.)

 

[kato]

Can't folks having trouble biasing just add a 5K series resistor with the pot to dial in the correct value? That would be my first instinct if 20K doesn't seem to do it.

The board is pretty tight for space, but if you can find room to fit another resistor, that would work.

For simplicity, I'm planning to order a 25k trimpot before I start building. But yes, adding a 5k resistor would work.

 

[Zim]

Can't folks having trouble biasing just add a 5K series resistor with the pot to dial in the correct value? That would be my first instinct if 20K doesn't seem to do it.

I’m going to need to give that a try - it’s definitely tight, but getting that trimmer out to replace it looks like it. Would be a nightmare!

Can I just double check from someone with more experience of electronics & circuit diagrams than me (essentially, most everyone here)... to put 5k in series I’d just need to desolder one end of R2 and solder that directly to. 5k resistor and then solder the other 5k leg to pin 2 of 20k trimmer, is that right? If i leave R2 connected to it’s pad I’d be putting the 5k resistor in parallel wouldn’t I?

Also need to order some resistors before I can give this a try - only have some 4.7k 1/8W or a 7k ¼W and I“m assuming I need a 5k 1/2W In there

 

[McIrish]

Leave R2 alone. You have to either cut a trace or remove the trimmer to add a resistor to one leg or the trimmer or just change it to a 25k trimmer. R2 has nothing to do with that. I removed the trimmer and added a 4.7K to one leg and tack soldered it to the board. Once I found the correct bias point, I measured the resistor&pot out of circuit and replaced it with a single resistor. getting the trimmer out is a bit tricky as one side connects to ground and it's a bit hard to heat up enough to pull it. But... I did it twice (two mics) so it can be done.