Injecting an audio signal into a microphone circuit

About injecting an audio test signal into a microphone circuit:
I have bought a number of 5840-tubes for my M49 project, burnt them in and checked them for noise and microphonic. I’ve built a mockup of a M49 mic in a cookie jar (I have replaced the capsule with a 70 pF capacitor).
I have recorded noise and the “sensitivity to tap them with a pencil” and found that some of the tubes are better than the others.
Now I would like to check things like gain, frequency response and distortion.  I know how to do that in a recorded signal, but first I need to inject the audio test signal into the mockup mic circuit.
So, can anyone tell me exactly where to inject the signal from my signal generator? Should I inject the signal directly at the gate? Should the ground/minus be connected to the ground of the mic or maybe at the other side of the stand in capacitor for the capsule?
Might sound like a stupid question, but I simply don’t understand electronics well enough. Please help me.
- Stefan

I don't know if this helps you, but you can take a look at the u67 mic and psu schematics and look how they have done it. They have built in "test rig".

50-100pF capacitor between positive node of generator and 1st grid.
Negative node to microphone ground.
That's it.
If your generator have a really small DC offset, then you can omit capacitor and connect positive node to the 1st grid

+1. 

It depends exactly on what you would like to test, but cap couple it to the grid of your tube via a 100pF to best simulate what a capsule would do in situ.

If you want to get fancy, you can tie two 47pF caps in series, and inject your test signal to where they meet.  Then, the outside two leftover legs can be the two capsule connections, and the circuit behaves like a two-terminal capsule in omni mode.

I've tested my M49 Setup yesterday with exact these setup. Works very well.
For me is interesting the Impedance of the tube bevore the signal goes into the output transformer.
I used this site to count the 'Innenwiderstand' of the Tube.

Sorry in german :-\

http://www.sengpielaudio.com/Rechner-EingangsAusgangsWiderstand.htm

Today i will test with Room EQ Wizard and a Sweep the Frequenzy  Response of my M49 Circuit with Transformer installed.
I dont have any expensive measure material at home, only three DMM's, Oszilloskop and my Signal Generator.

Is his the right way?
I've measured 14,3kohm @ 1khz Output Impedace from the EF762/732 Tube.
5:1 - 7:1 Output  Transformers like the Sowter 9995, or Lundahl 1538, or the Cinemag for High Impedance Circuits seems like the right one?
The Haufe BV11 with 10:1 Ratio is a little bit High i think.

Any Impedance Experts out there? :

Best
Andreas

A 5:1 transformer will make 14K3 look like about 570 ohms. A 10:1 transformer will make it look like 143 ohms but the level will be 6dB lower. Take your pick.

Cheers

Ian

ruffrecords said:

A 5:1 transformer will make 14K3 look like about 570 ohms. A 10:1 transformer will make it look like 143 ohms but the level will be 6dB lower. Take your pick.

Cheers

Ian

Click to expand...

OK, i count the same this morning.

5:1 Transformer is 5x5:1 = 25:1 Impedance Change = 14300:25=572Ohm
7:1 Transformer is 7x7:1 = 49:1 Impedance Change = 14300:49 = 292Ohm
10:1 Transformer is 10x10:1 = 100:1 Impedance Change = 14300:100= 143Ohm

7:1 is the best Choice?!.....

I've make some Measurements with two Microphone Circuits, after deinstalled the capsule and solder in capacitors.
Then i send the Sweep direct to Tube Grid, correct the Levels in Room EQ Wizard and measure.
see picture.

when i'm wrong, please correct me!

There's a (good) chance the coupling capacitor starts to have an influence there too. For a fixed capacitor value, if you drop the impedance it "goes into", the cutoff frequency goes up.

That, plus since the transformer is a bit of a reactive load, that could potentially explain that low peaking with the 10:1. But i may well be wrong

Andreas Grosser told me via Telephone, is the Tube Impedance to high the Microphone is very very sensitive to low Frequenzy's. Is sounding "boomy" or in german "poltrig". Is the Impedance to low, then you have less bass.

Will make measurements with 4 different Tubes EF762/32/5840, EC71, 5703 and 6s6b-V.

For Haufe and Lundahl.

But that Tube soldering sucks : :-[

Andreas

For the testing of tubes with wires, I use an IC socket.
That makes it very easy to push the wires of the tube in, and test the next tube...

is the Tube Impedance to high the Microphone is very very sensitive to low Frequenzy's. Is sounding "boomy" or in german "poltrig". Is the Impedance to low, then you have less bass.

Click to expand...

Given the plot you linked, the system Q seems 'underdamped'. It really looks like modifying Q(tc) factor with a sealed loudspeaker alignement.

https://www.diyaudioandvideo.com/Calculator/SpeakerBoxEnclosure/Help/boxQ.gif

What Andreas Grosser told you confirm that from a frequency response point of view. It can be added that transient response/behavior and efficiency is linked too: for best transient response (for a closed box loudspeaker, and overly simplified) target is Q(tc) of 0.577 (also known as critically damped). It gives the deeper frequency response too. Efficiency is less a concern imho.

A good balance/compromise between freq response, transient behavior and overall efficiency is a target of Q: 0.707  (this is usually what is used for closed box design, but it is arguable depending of listening space but it's an other subject on his own...).

As Khron said playing whith coupling cap value will modify Q of overall system.   

played a little bit with dc blocking caps, changed value from 1uF to 2,2uF.
Now is better, but the High Freq Roll Off from 10khz to 20khz is 7db. Maybe a Problem? Measured with 48K. The Resolution at very High Freq is horrible in 48Khz Sample Rate, but i can't set the Resolution higher in Room Eq Wizard.
Measured Impedance in this Circuit after Sengpiel Method, Result is with 5703 Tube 14700ohm @ 1khz.
I think the Cinemag CM-6511 would work perfect. The Lundahl 1940 is maybe an Option.
A original AC701K is a High Z Tube and will work very well with this Haufe Transformer, but with 5703 Tube is the Cinemag the better Choice.

Ericbazaar said:

the High Freq Roll Off from 10khz to 20khz is 7db.

Click to expand...

I suppose you did calibrate the soundcard you are using to measure with?
(Because I never experienced this with a tube stage.)

RuudNL said:

I suppose you did calibrate the soundcard you are using to measure with?
(Because I never experienced this with a tube stage.)

Click to expand...

Yes measure it, sweep straight out and direct in. 100% linear (Universal Audio Apollo).
The Resolution from 13khz - 20khz is horrible at 48k, it looks like Square Waves. I think a Sinewave with 192k resolution brings a better Result. Or i'm totaly wrong?

Thank you all for your input and advice!
Tested one of my Mullard 5840-tubes tonight. It works like a charm.

Hi stelin,

courious, what kind of transformer you have used ?
Thanks for your input as well,

Manfred.

Hi Manfred,

It’s an AMI T49.

Below 20 Hz, the frequency response curve I posted does not say that much about the transformer.
The response curve depends on how hard you hit the transformer and the diagram from REW also depends on the speed of the sweep.

First, I fed the mic with -12 dbFS (the default in REW). That produced a resonance top of about 5 db at 10 Hz.
I then reduced the output from REW to -50 dbFS. That gave me a god signal from the mic with 30 db gain in the pre amp, about the same as you would expect when recording a singer. That’s the curve you see in my post.

But, that was with a fast sweep (256k in REW terms, or 5 seconds). When I increased the sweep to 1M (about 20 sec) the curve was flat down to 10 Hz and then dropped off without any anomalies.

But then again, I’m not that interested in what the curve looks like below 20 Hz... That's a part of the spectrum I always try to get rid of anyway.

- Stefan

Hi Stefan,

thanks, your answer is giving me big relief, because I acquired already the Haufe "BV11", the T8974 with a 10:1 ratio like your AMI T49.
Building a M49 turns out to big the biggest deal I had so far.
Everything seemed well planned in the beginning, but the difficulties came, before I even started.

OK, let´s see what the Haufe will turn out with the 6S6B-v.
I have to finish my pair of C12 first, before I start with the M49s.

I´m not familiar with REW so far, just got the hint of Ericbazaar to make measurement with that prog, what I will do.
For me, as well, frq below 20Hz are of no interest; but these freq absent without having to block them  is the better solution.
Have a nice week,

Manfred.