Banzai's KM84 DIY Body & PCB kit build thread

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Fk, same ones I got...
Maybe you will have better luck. Like I said, I only tried one out of the five in the bag. Maybe a different FET is the way to go. Something still being made with accurate specs and consistency. If I had extra cash, I might try a different transformer too. But that will have to wait a while.
 
Just note that the LS846 has a different pinout. Specs look good for a microphone too. I might give that a shot.
 
Linear Systems LS846 = 2N3819 replacement.

I'm thinking the LS846 might be the "best" of the currently available options. The reissued 2n3819's appear to be of a different spec than the originals. (Possibly hearsay.) The 2SK170's have too large input capacitance.

The RusselTech KM84 project has a breakdown of 4 or 5 different FETs and the corresponding drain resistor value. One guy, "Gord" found the LS846 to bias at a value closest to the original Neumann at 3.6k. I realize this might be an amateurish conclusion, but I decided to buy a lot of the LS846's.
gord.png
Problem is, they're sold in minimum qty of 10 at 5.06 ea. plus tax and I don't need 10.
Anyone in the US want to take 5 of them and split the cost, please let me know. $29.88 would get 5 of them shipped to you.

Edit: would take $26.89 F&F to match NacSemi price below

PS, Banzai, thanks for the tips on how the capacitor values affect the bass response.
 
Last edited:
I'm thinking the LS846 might be the "best" of the currently available options. The reissued 2n3819's appear to be of a different spec than the originals. (Possibly hearsay.) The 2SK170's have too large input capacitance.

The RusselTech KM84 project has a breakdown of 4 or 5 different FETs and the corresponding drain resistor value. One guy, "Gord" found the LS846 to bias at a value closest to the original Neumann at 3.6k. I realize this might be an amateurish conclusion, but I decided to buy a lot of the LS846's.
View attachment 84083
Problem is, they're sold in minimum qty of 10 at 5.06 ea. plus tax and I don't need 10.
Anyone in the US want to take 5 of them and split the cost, please let me know. $29.88 would get 5 of them shipped to you.


PS, Banzai, thanks for the tips on how the capacitor values affect the bass response.
Nacsemi.com sells them also and you can buy in any quantities.I used this jfet for km clones and the original fairchild still sounds better(if you can find low idss values)
 
Last edited:
What is the problem with the 2n3819's? They're from mouser so supposedly not fake? Is it just the biasing problems (just need a larger resistance trimmer) or do they sound different and how would you determine this except for swapping otherwise identical bodies with the same capsule? And different from what? Can you swap an original KM84 body in with one of these capsules?

I did notice that all of my 2n3819's needed to be biased around 21K-22K like others have noted. Reading this thread had me preemptively solder in an additional 4.7K resistor in series with the pot. No problem.

Anyway, of the 3 mics I bought, there are two capsules that are similar in volume and very close in tone, probably enough for a matched pair. I'm still listening. I like them though. There is another capsule that is a few dB louder, more bass, and maybe a little duller top end. I'll need some more accurate post level adjustment to really tell.

But the sound differences in my case follow the capsule. However, I do have all 2n3819's. 3 mics, two with cinemags and one with 3u transformer. I think I like the cinemag bodies better but it's subtle. The capsule is the main determining factor.

But transistors make a difference in sound. I noticed this a long time ago when building preamps.
 
At first I didn't think the FET type would make much difference. As a straight amplifier, it should not have much bearing on the sound, but the source resistor is going to have an effect on the overall drive current into the transformer. Will that make a difference sonically? I have no idea. I too have decided to try the LS846. I see Digikey has them but you have to buy 10 so I just bought a couple from nacsemi.com. I do think the Central 2N3819 is probably a fine FET but possibly there are better options when driving a transformer. I'm just not totally convinced there will be a difference we will be able to hear, but willing to give it a try.
 
Last edited:
I'll be curious to hear about what you all discover about the LS846 FETs and their impact on sonics.

I did a bunch of testing on C3 and C4 capacitor values and have uploaded my data here. Ignore the 60Hz bump from my mains and keep in mind, these are sine sweeps directly into the capsule input, not what the frequency response of the microphone will be with a capsule. Basically, IMO, if you have the Cinemag transformer, 4.7uF is not the correct value for C3. I'll be in my studio this weekend to test out C3=1uF, C3=2uF, and then the original C3=4.7uF. I wasn't liking the curves (or they looked redundant) when I was playing the value of C4 so I stuck with these options to test out real world recording situations. I will plan to do a little acoustic guitar and some mono drum overhead comparisons.
 

Attachments

  • km84 circuit reponses with cinemag.mdat.zip
    14.9 MB · Views: 32
Also, if you are curious about quality control on the capsules included in this kit, I've attached the results of swapping all seven of my capsules on the same circuit and body and measuring their response with a sine sweep out of one of my monitors. The final two are my Neumann KM86s in cardiod. This is not a scientific comparison at all, I had to reposition the mic stand when testing the KM86s as they are side address. I got them as close as possible to the same position as these clones as I could. The point was not to measure frequency response, but to determine how closely matched my daily use pair of SDCs are, and see if I can get these clones in the same ballpark.

I managed to find 1 very closely matched pair, two pretty good matched pairs and then the one massive derivation. I'm glad I have an odd number of capsules. I suppose if you only have one pair of these and they aren't matched level wise, you could probably get away with sticking small value caps in parallel with C2 until you get them close, but I'm happy I have 3 good pairs and one odd duck...
 

Attachments

  • km84-capsules-testing.mdat.zip
    19.2 MB · Views: 10
Nacsemi.com sells them also and you can buy in any quantities.I used this jfet for km clones and the original fairchild still sounds better(if you can find low idss values)
Thanks for this resource! They also have the hard-to-get LSK389. Bookmarked!

Still, if anyone is looking for exactly 5, hit me up. The price above is an exact split with me covering 1/2 of the $3.35 first-class postage (boxed) within the US to your house. (Could also subtract the $1.13 paypal fees for those who know and trust me.) I seriously doubt I will use all 10 in my lifetime.
 
I'll be curious to hear about what you all discover about the LS846 FETs and their impact on sonics.

I did a bunch of testing on C3 and C4 capacitor values and have uploaded my data here. Ignore the 60Hz bump from my mains and keep in mind, these are sine sweeps directly into the capsule input, not what the frequency response of the microphone will be with a capsule. Basically, IMO, if you have the Cinemag transformer, 4.7uF is not the correct value for C3. I'll be in my studio this weekend to test out C3=1uF, C3=2uF, and then the original C3=4.7uF. I wasn't liking the curves (or they looked redundant) when I was playing the value of C4 so I stuck with these options to test out real world recording situations. I will plan to do a little acoustic guitar and some mono drum overhead comparisons.
I can't get these to open in REW. Could you post a screenshot of the graphs?
Thanks,
Craig
 
I can't get these to open in REW. Could you post a screenshot of the graphs?
Thanks,
Craig
Hi Craig,

Hm, I could but there are 13 files and it'd probably be most useful to be able to thumb through them in REW. Do you have the newest version of the software? Not sure why they wouldn't open, I just checked on my other computer and they opened fine.
 
There are many alternates that aren't over $1 that work well in these Neumann circuits. A few I've tried successfully are the J113 and J305.

The key to finding a good JFET is to buy 10 of them, and measure IDSS, and find the one that matches the circuits operating point (typically 0.7mA to 1mA in KM84 and U87). Any JFET with an IDSS in the range of 0.5mA to 2mA, with a VGSoff of less than -3V should work well in this circuit.
 
nacsemi says $4.54 for to-72 1-99 pcs is the shipping from them more?
I had just assumed it was, and that they would also charge sales tax.

But your Q prompted me to go through their checkout process. They offer a $6 shipping option and they *don't* charge sales tax. So qty. 5 from NacSemi comes to $26.89. ($3 cheaper than what I was offering.) Needless to say, if you want 5 and want to pay via F&F, I'd send them your way at the NacSemi price. I still haven't bought the Cinemags for this project so I could use some extra dollars, but I'm not sweating it either way. Digikey was the only option I knew about at the time.

Incidentally, I tried to order a sample direct from Linear, but they correctly sussed that I was not a manufacturer. After 2 weeks of it not arriving, I ordered from Digikey.
 
Last edited:
There are many alternates that aren't over $1 that work well in these Neumann circuits. A few I've tried successfully are the J113 and J305.

The key to finding a good JFET is to buy 10 of them, and measure IDSS, and find the one that matches the circuits operating point (typically 0.7mA to 1mA in KM84 and U87). Any JFET with an IDSS in the range of 0.5mA to 2mA, with a VGSoff of less than -3V should work well in this circuit.
does anyone have any tips for measuring IDSS and VGS as Matador mentions - or is specialist equipment needed? I’ve got a good DMM and an LCR meter but that’s it at the moment...
 
The key to finding a good JFET is to buy 10 of them, and measure IDSS, and find the one that matches the circuits operating point (typically 0.7mA to 1mA in KM84 and U87). Any JFET with an IDSS in the range of 0.5mA to 2mA, with a VGSoff of less than -3V should work well in this circuit.
Thanks for that tip Matador.

Here's a nice tutorial on Measuring JFET Idss and Vgs(off) for anyone else curious.
 
Can it be measured with the Peak Atlas DCA75?

also this Chinese component meter is pretty cheap and able to test a lot of different components and parameters. I highly advise it:


0DF931EB-7FBA-4F6D-BCC0-D752F8A7731A.jpeg
 
Last edited:
The MK328 tester I posted is sold in many different shapes and cases, it's a well known tester based on the ATmega328 chip.
this is what it measures as far as J-FETS is related:

Screen Shot 2021-08-28 at 19.42.59.png
 
Back
Top