OPEN SOURCE DIY Mic Project - ORS 87 - Stripped Down u87

[oran.outan]

Did you use LSK189?

Yes !

 

[micolas]

Yes !

Is R12 33k or 47k?
Did you do the BIAS of the jFET on the oscilloscope and signal generator, 1kHz?

 

[oran.outan]

Is R12 33k or 47k?
Did you do the BIAS of the jFET on the oscilloscope and signal generator, 1kHz?

I use 33k

I set the bias with multimeter and then look the waveform with oscilloscope and 1khz… but I have a cheap oscillator and not really confortable with that part with poor knowledge…the waveform looks ok (regular wave and not saturated).
I mostly follow instructions write by Oneroomstudio and wordsushi…

 

[Khron]

but I have a cheap oscillator and not really confortable with that part with poor knowledge

Do you have an audio interface and a computer you can install RoomEQ Wizard on? If the answers is "yes" to both, you don't need any (more) fancy hardware than that

 

[micolas]

I use 33k

I set the bias with multimeter and then look the waveform with oscilloscope and 1khz… but I have a cheap oscillator and not really confortable with that part with poor knowledge…the waveform looks ok (regular wave and not saturated).
I mostly follow instructions write by Oneroomstudio and wordsushi…

I use the method described by @poctop and @Matador with good results

https://groupdiy.com/threads/aurycle-fet-mikes.46109/post-587802

 

[oran.outan]

Do you have an audio interface and a computer you can install RoomEQ Wizard on? If the answers is "yes" to both, you don't need any (more) fancy hardware than that

Yes, already use REW for frequency response…
Right, i could use it for that

 

[Khron]

Yes, already use REW for frequency response…

It's got a signal generator, it's got a scope, it's got a spectrum analyzer / RTA (with live THD / harmonics measurement)

 

[Wordsushi]

Hi all,

This is my build on V1 board, using khron's BOM, Neutrik transformer (on the back) and Vami K67 capsule.
pretty Happy with the result !
Using 36 zener, I get 33v and set to 16,5v at the drain.
Thanks to @anvl for the body tips !
Thanks to @Papa Tango Charly for the PCB !

Enjoy°
0-0
View attachment 131004View attachment 131005

NICE!!!! That's a good looking build. Congrats, and welcome to the club!

 

[JsLR]

Hi! I'm trying to bias with the 1khz and scope method (REW), like I did before succesfully with an oktava 219 mod, but can't seem to get good readings.

2n3819 with 33v zener build, around 27.4v before R12 (47k) and 11.4v after R12 if I target for minimum THD (looks a bit low to me). In the REW scope, one side of the wave is clipping earlier, I assume because voltage post-R12 should be closer to half the voltage pre-R12, but...

If I bias only looking for that voltage, I can get 28.8v and 14.4v (pre and post R12), and the clipping is more symmetrical, but THD is too far from being minimum. Furthermore, level on one side of the wave is bigger.

I know biasing has a subjective component related to "taste", but the fact that I get the minimum distortion when the wave clips assymetrical is right?Thanks in advance and a BIG thank you for all the incredible info and support in this project.

 

[micolas]

Hi! I'm trying to bias with the 1khz and scope method (REW), like I did before succesfully with an oktava 219 mod, but can't seem to get good readings.

2n3819 with 33v zener build, around 27.4v before R12 (47k) and 11.4v after R12 if I target for minimum THD (looks a bit low to me). In the REW scope, one side of the wave is clipping earlier, I assume because voltage post-R12 should be closer to half the voltage pre-R12, but...
View attachment 131050
If I bias only looking for that voltage, I can get 28.8v and 14.4v (pre and post R12), and the clipping is more symmetrical, but THD is too far from being minimum. Furthermore, level on one side of the wave is bigger.
View attachment 131051
I know biasing has a subjective component related to "taste", but the fact that I get the minimum distortion when the wave clips assymetrical is right?Thanks in advance and a BIG thank you for all the incredible info and support in this project.

If you measure the voltage across the Zenner diode, you will find that it stabilizes less than 33V (it has a tolerance)
So, connect a 330K resistor in parallel across R14 (orig 10K) so that R14=9.7K.
Now adjust Udrain=11.5V and measure the voltage at the connection R12 to R14, if it is closer to 29V (+/-0.1V)
When I had problems with Zenner's tolerance, I used a resistor between 220...390K in parallel, so that at a correct BIAS I measured 29...29.1V.
*
If it's ok, adjust the BIAS again, from R*, with a 1kHZ input signal high enough for both half-sinusoids to enter clipping at the same time, to be symmetrical. The 1kHz signal level is then decreased by until there is no limitation. If even then the shape is symmetrical, it's ok. If not, the procedure is repeated again, until a compromise is reached for the highest undistorted level with good symmetry.

Note:
Of the two BIAS adjustments you presented, the first one is closer to the correct one.
*
The "correct" BIAS can differ depending on everyone's preferences, the character of the desired sound (cleaner, more saturated, etc.)

 

[untamedfrontier]

I'm here to continue bias talk, because I don't know if I understand yet how to correctly inject the 1k tone. Is a 50p or similar capacitor only needed if there is not a capsule connected?

I connected a 50p to the R7 when I sent a 1khz sine from REW, but my tone ended up being more like 1.7k, but couldn't shake the feeling that I was doing it totally wrong. Oh the life of a n00b

 

[JsLR]

If you measure the voltage across the Zenner diode, you will find that it stabilizes less than 33V (it has a tolerance)
So, connect a 330K resistor in parallel across R14 (orig 10K) so that R14=9.7K.
Now adjust Udrain=11.5V and measure the voltage at the connection R12 to R14, if it is closer to 29V (+/-0.1V)
When I had problems with Zenner's tolerance, I used a resistor between 220...390K in parallel, so that at a correct BIAS I measured 29...29.1V.
*
If it's ok, adjust the BIAS again, from R*, with a 1kHZ input signal high enough for both half-sinusoids to enter clipping at the same time, to be symmetrical. The 1kHz signal level is then decreased by until there is no limitation. If even then the shape is symmetrical, it's ok. If not, the procedure is repeated again, until a compromise is reached for the highest undistorted level with good symmetry.

Note:
Of the two BIAS adjustments you presented, the first one is closer to the correct one.
*
The "correct" BIAS can differ depending on everyone's preferences, the character of the desired sound (cleaner, more saturated, etc.)

Thanks for your answer Micolas! Unfortunately, changing R14 has little effect...
I can get a little more voltage at the connection of R12-R14, but when I try to bias again changing R*, I always end up with voltages around 27.7v before R12 and 12.0v at drain.

Furthermore, when I adjust R* for symmetrical shape THD is high (especially 2nd harmonic, and you can clearly hear it). If I try to get low THD, it only works for that 1khz signal level, as soon as I raise or lower the 1khz level signal THD is off again.
I mean, in the oktava 219 when I get the minimum amount of THD after two or three passes it stays low no matter the gain of the signal being applied, until it clips. Here, if I tune it to low THD at the point of almost clipping, THD raises a lot again if I lower the 1khz gain, and you can clearly hear the harmonics raising as the signal is lowering...

Probably I'm wrong, but I supposed that the voltage measured at Drain marks a center point for the wave to "swing", up until max voltage (around 29v), down until 0v, so the bias "correct" point is usually in the ballpark of 15v...
That's why my readings (12v) look strange to me, because one half-sinusoid is going to clip earlier (0-12v vs 12-27.7v), but it shows the lowest THD??

Anyway, the mic is sounding great, c5=33nf and c6=150pf, clear and full, and probably I'm not going to hear so much differences in the harmonic content, so maybe I'm just overthinking...

 

[micolas]

Thanks for your answer Micolas! Unfortunately, changing R14 has little effect...
I can get a little more voltage at the connection of R12-R14, but when I try to bias again changing R*, I always end up with voltages around 27.7v before R12 and 12.0v at drain.

Furthermore, when I adjust R* for symmetrical shape THD is high (especially 2nd harmonic, and you can clearly hear it). If I try to get low THD, it only works for that 1khz signal level, as soon as I raise or lower the 1khz level signal THD is off again.
I mean, in the oktava 219 when I get the minimum amount of THD after two or three passes it stays low no matter the gain of the signal being applied, until it clips. Here, if I tune it to low THD at the point of almost clipping, THD raises a lot again if I lower the 1khz gain, and you can clearly hear the harmonics raising as the signal is lowering...

Probably I'm wrong, but I supposed that the voltage measured at Drain marks a center point for the wave to "swing", up until max voltage (around 29v), down until 0v, so the bias "correct" point is usually in the ballpark of 15v...
That's why my readings (12v) look strange to me, because one half-sinusoid is going to clip earlier (0-12v vs 12-27.7v), but it shows the lowest THD??

Anyway, the mic is sounding great, c5=33nf and c6=150pf, clear and full, and probably I'm not going to hear so much differences in the harmonic content, so maybe I'm just overthinking...

How much did you measure the IDss and Vgs (off) of the 2N3819?
It has a large dispersion of parameters, more must be tested.

 

[JsLR]

How much did you measure the IDss and Vgs (off) of the 2N3819?
It has a large dispersion of parameters, more must be tested.

I only measured Idss in some of the jfets, from 5 to 8.5mA, within what's adviced for this build...
As I've read, the variability of 2N3819 characteristics may be problematic...I went for 8.5mA, but next step its for sure try several 2N3819 and see what happens!!

 

[micolas]

I only measured Idss in some of the jfets, from 5 to 8.5mA, within what's adviced for this build...
As I've read, the variability of 2N3819 characteristics may be problematic...I went for 8.5mA, but next step its for sure try several 2N3819 and see what happens!!

When I was building guitar pedals, I biased the jFETs at half the supply voltage, but in the U87 topology, Neumann biases Udrain=11V for a voltage of 29V. It is true that they had the opportunity to select the jFET for optimal circuit parameters.

 

[Khron]

Probably I'm wrong, but I supposed that the voltage measured at Drain marks a center point for the wave to "swing", up until max voltage (around 29v), down until 0v, so the bias "correct" point is usually in the ballpark of 15v...

Well, that presumes the JFET clips 100000% symmetrically, but there's no reason or rule that says it should. Ah, the joys of real-world imperfect components...

And yes, the voltage before the 33k / 47k drain resistor WILL change, depending on the bias / quiescent current through the JFET because you have at least the 10k resistor in series there, after the zener. Ohm's law always applies

 

[micolas]

I remember once @RuudNL found the 'sweet spot' for a drain voltage of about 9V or something like that.

 

[RuudNL]

Even though a voltage of 11 V. at the drain is mentioned on Neumann schematics, I measured voltages between 8.5 and 13 V. at the drain of several U87ai microphones. (Original Neumann.)
Personally I never calibrate for symmetrical clipping, but for lowest distortion at fairly high levels. (But before serious distortion sets in.)

 

[micolas]

Personally I never calibrate for symmetrical clipping, but for lowest distortion at fairly high levels. (But before serious distortion sets in.)

Did you notice this on the oscilloscope also in the case of the original Neumann U87i and U87ai?

 

[RuudNL]

I never check the output waveform on an oscilloscope, I only measure the distortion on a (hardware) audio analyzer. I have measured the drain voltage on several Neumann U87ai microphones.
(An oscilloscope will only show you distortion if it is around 3% or more, so the scope picture is not important to me.)