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

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Flat circuit with flat K47 in the body I use is slightly thin in the bass. What could be the easiest way to fix that? I have no C6 and shorted C9.
Using my channel strip, parametric EQ of -2dB @ 2 KHz sounds just about right (Q might be 1.89).
 
Flat circuit with flat K47 in the body I use is slightly thin in the bass. What could be the easiest way to fix that? I have no C6 and shorted C9.
Using my channel strip, parametric EQ of -2dB @ 2 KHz sounds just about right (Q might be 1.89).
Try increasing the value of C5 (in your schematic). I personally like 68n but you can even go higher, like 100n.

Or rather just remove C5 and/or R9 altogether...
 
I just wanted to chime in regarding one thing. I am late to the game, tried searching and couldn't find anything about it.

There is one crucial resistor missing (i believe in 500ohm range) from r9 to ground. @OneRoomStudio

This creates biasing (messeingang) input for correct biasing of the fet. Injecting into FET gate, or any other point doesn't take negative feedback into account. Also wrongly biased FET will throw off HF/LF response. DONT BIAS THE FET TO GET ANY SPECIFIC VOLTAGE OR CURRENT. It has to be done to get lowest thd and appropriate HF/LF reduction.

Especially important if you have random 3819 fet, or inadequate substitute.

And don't forget negative feedback tailors both high and low frequency respinse!

I am sorry if this has been adressed before, i just couldn't find anything.
 
Last edited:
Would it still be a way to just find the proper bias for lowest THD without the Messeingang as Klaus Heyne suggests?

https://repforums.prosoundweb.com/index.php?topic=21044.0

Yet another nice forum :)

I think I saw user name 'RuudNL' as a member here too?

https://repforums.prosoundweb.com/index.php?topic=37367.0

... and I can leave the head/capsule connected to the circuit board while injecting sine wave?
Klaus' method is a joke, you can disregard it. If nothing else, headphone driver will have higher THD than your mic at that much injected level. Do you want to measure headphones or the mic?

As i explained, and as Ruud explained, the correct way is to do it through messeingang that's lacking in this project. It's just one extra resistor.

If you inject any other way you are omitting the feedback network that is there to, among other things, decrease distortion. The microphone capsule is part of that network.

The following reply from Uwe is also right.


Update:
The last u87ai I measured could take 407mV RMS injected into messeingang for 1% THD @1K.

However it could take just 226mV injected through 60pF cap directly into gate for the same THD value.


Sadly i haven’t measured what input voltage would give 0.5% that's specified by Neumann for 117db. Idk either how linear THD % increase is with SPL value.

@RuudNL have you maybe checked what injection voltage into messeingang gives you 0.5% thd @1k with u87ai?

I found in a note i made quite a while ago 390mV RMS = to 117dB = 0.5% THD. However i can't remember how i came to that conclusion.
 
Last edited:
I did a quick LTSpice simulation of the way the frequency response changes with FET gain:

ORS87 Fr vs gain.png

The traces are show the overall amplifier response (from capsule voltage to output) with a basic FET gain of +5dB, +10dB, ... up to +25dB. I'm using the feedback-network component values from post #903, not the original U87 ones.

As you can see the HF and LF rolloff becomes more pronounced as the gain increases, so it's reasonable to expect the sound of the mic to change as you adjust the bias (and therefore the FET open-loop gain). Probably, therefore, you should adjust the bias first, then twiddle values in the feedback network.
 
Flat circuit with flat K47 in the body I use is slightly thin in the bass. What could be the easiest way to fix that? I have no C6 and shorted C9.
Using my channel strip, parametric EQ of -2dB @ 2 KHz sounds just about right (Q might be 1.89).
You shorted C9? That would create a DC short between the capsule polarizing voltage (~48V) and the FET drain...which would seriously mess up your biasing. Why did you short C9?
 
I just wanted to chime in regarding one thing. I am late to the game, tried searching and couldn't find anything about it.

There is one crucial resistor missing (i believe in 500ohm range) from r9 to ground. @OneRoomStudio

This creates biasing (messeingang) input for correct biasing of the fet. Injecting into FET gate, or any other point doesn't take negative feedback into account. Also wrongly biased FET will throw off HF/LF response. DONT BIAS THE FET TO GET ANY SPECIFIC VOLTAGE OR CURRENT. It has to be done to get lowest thd and appropriate HF/LF reduction.

Especially important if you have random 3819 fet, or inadequate substitute.

And don't forget negative feedback tailors both high and low frequency respinse!

I am sorry if this has been adressed before, i just couldn't find anything.
Injecting signal in series with a ~50pF capacitor in place of the capsule works fine, as show in the original simulations here, and described here.
 
How does it looks for a V1.2? :)
Nice. Which transformers would fit besides 3U Audio GZT-87? I read it's your favorite :).

Will it have a messeingang? It sounds convenient to have such after the capsule has been soldered on.

I tried to measure by creating messeingang using 6K8 + 680 resistor (didn't have the original values from U87 original schematics).
I used the holes for switches in HL-95 with fully assembled mic body and got some results. Don't know if I did it properly.
I wanted to see the noise floor being lower than the overtones.

1722067735459.png
 
Hey guys, I made this version of the circuit in stripboard. I'll share here so that anyone can use it.View attachment 131491

Hi Jedias,

I believe I see some errors here:
  • R15 contact needs to move from K7 to K6.
  • Cut on G9 needs to move higher up on the same strip 9, anywhere between N and Q.
Please have a look and let us know if these are correct.

Cheers,
D.
 
I'd like to ask your opinion about my BIAS calibration procedure (using 24V zener).
No hardware oscilloscope is used, only DAW.

1. Disconnect both capsules wires from the PCB (FD and BP)
2. Connect mic XLR to preamp INPUT and phantom +48V powered on. Connect preamp output XLR to your DAW and use some kind of scope / frequency analizer.
3. Adjust trim pot to get 11Vdc in JFET drain as a starting point (assuming 24V zener)
4. Inject a 100mVac signal from your DAW signal generator to the PCB (use a two wires unbalanced cable with (+) terminal connected to JFET gate and (-) terminal connected to GND). You should have something like this.
View attachment 131520

5. Raise signal to 200mVac and you should have a little more distorsion (harmonics)
View attachment 131521

6. Raise signal to 300mVac and you will have this (even more distorsion)
View attachment 131522

7. Now adjust trim pot to decrease this harmonic distorsion as much as possible (you won't be able to remove distorsion completely). Something like this
View attachment 131523

8. My final reading in JFET drain is 9,2Vdc.
9. Remove calibration 2 wires cable and solder capsule wires.
10. It's done. Congratulations !!!

What do you think? Do you find it correct?
I wanted to confirm using this detailed explanation.
More neophyte questions.

If I'm following correctly, I can power up the mic (Capsule disconnected) with My Focusrite Scarlet interface and feed the signal in REW on my computer. Simple enough...

The part I don't fully get is step 4: injecting a signal, what am I using to connect and the source of the signal? Am I using just 3.5mm audio output from my Mac and connecting wires directly into the circuit. I'm sure I have a couple sacrificial cables around I can appropriate for the effort.

Would one of you be so kind as to post a photo of the setup? My REW experience is primarily with a uMic and speaker testing so this is quite different.

eso
 
Last edited:
Back
Top