Author Topic: Mic Calibration  (Read 6225 times)

0 Members and 1 Guest are viewing this topic.

Robert

  • Newbie
  • *
  • Posts: 4
    • View Profile
Mic Calibration
« on: June 06, 2011, 18:00 »
hi there,

     anyone tried to use SATlive MAT module in FFT mode for mic calibration?

     I've had an access to acoustic lab at our local meteorology service and was very curious about how "flat" my dbx rta-m mic is. I couldn't use pistonphones/calibrators for calibration because (to my surprise) none of the standard adapters fit, so I decided to do a comparative calibration. I've put a speaker into anechoic chamber, fed it with pink noise from generator (SRS DS360), captured the signal with the ref mic (G.R.A.S 40AF) and analyzer (Larson Davis 2900B), and compared the results with the signal captured by dbx mic, through sound card and SATlive.

     To be honest I didn't expect to get such results (see attachment). Low freq slope looks ok for me, all cheap measurement mics have it, but the rest of the spectrum... it looks like AM modulated. Any thoughts what is the potential problem here? I'm not very convinced that rta mic could have that type of response.

     I will try to describe the setup more in detail:

     REF GEN -> speaker in ANECHOIC CHAMBER -> REF MIC -> REF ANALYZER

     DBX MIC close to ref mic -> TASCAM US-144 -> SATlive 1.40.18.1 (TASCAMs ASIO, WIN 7 Ultimate, Dell Vostro 1710)

     I've set the level of gen to produce 60db SPL (1kHz) at the ref and dbx mics, adjusted gain of the crappy Tascam preamp for a healthy level and adjusted offset in SATlive to have this 1kHz to be exactly at 0dB in SATlive MAT module in FFT mode (set to Flat with Pink Noise in FFT compensation menu) . Then run through freqs of interest adjusting the level from gen to produce 60db at the ref mic (did that to compensate for obviously not flat freq response of the speaker) and registering the level seen in SATlive.



     Any thoughts? Wrong calibration technique? Crappy TASCAM sound card (didn't have my RME FW800 at the moment)? Possible SATlive issues?

     Thanks in advance for any comments on the subject.
    
« Last Edit: June 06, 2011, 19:12 by jazzuhead »

TomyN

  • Administrator
  • Sr. Member
  • *****
  • Posts: 286
  • Master of SATlive
    • View Profile
    • SATlive Homepage
Re:Mic Calibration
« Reply #1 on: June 06, 2011, 18:37 »
Hi,

thanks for your posting. I'm a little confused about your description, because the picture you post looks very much like transfer-function, but you're talking about MAT-FFT all the time.

Tomy
Dipl.-Ing.(FH) Thomas Neumann
SIM-II Operator
DANTE Level1 certified

Robert

  • Newbie
  • *
  • Posts: 4
    • View Profile
Re: Mic Calibration
« Reply #2 on: June 06, 2011, 18:47 »
Hi Tomy,

MAT-FFT was used only for measuring. The picture in the attachment is a screen shot of SATlive Miccorrection Editor with measurement values. Let say the ref mic at 339.1Hz had its 60db SPL then the dbx showed +1.8db in the MAT-FFT.

Hope now it's more clear.

TomyN

  • Administrator
  • Sr. Member
  • *****
  • Posts: 286
  • Master of SATlive
    • View Profile
    • SATlive Homepage
Re:Mic Calibration
« Reply #3 on: June 06, 2011, 18:54 »
Okay, now I understand. I asume that your using the 'Edit values Section'.  If you enter a value in this section, the trace is modified only at the frequency of this point, and the next points remain at zero dB.

You should use the Point to Point Connection section, which connects directly between the points.

Tomy
Dipl.-Ing.(FH) Thomas Neumann
SIM-II Operator
DANTE Level1 certified

Robert

  • Newbie
  • *
  • Posts: 4
    • View Profile
Re: Mic Calibration
« Reply #4 on: June 06, 2011, 19:08 »
yes, Tomy, I understand what you mean. I was too lazy to go through this procedure. I used calibration editor only to put in the values quickly.

But my concern is why would mic have that type of "AM modulation like" freq response. I've seen quite a few mic calibration files for different mics, and never seen that kind of response.

And by the way, Im not trying to shift all the guilt on the SATlive. Just wanna to understand were is the problem.

I modified the attachment of the first post, just drawn the line by hand.
« Last Edit: June 06, 2011, 19:21 by jazzuhead »

TomyN

  • Administrator
  • Sr. Member
  • *****
  • Posts: 286
  • Master of SATlive
    • View Profile
    • SATlive Homepage
Re:Mic Calibration
« Reply #5 on: June 06, 2011, 21:00 »
Okay. The problem is the procedure you use. I'll try to explain: Any FFT works with discret, i.e. fixed frequencies. For a very coarse FFT this might be 100Hz, 200, 300, 400 and so on. If you measure a sine wave with 200 Hz all energy will be displayed at the 200 Hz Amplitude value. But if you use a value of let's say 250 Hz, the energy will be shown in the 200Hz and the 300Hz value, thus both values will be smaller.

To solve this issue you need to compare a the frequencies used for FFT in SATlive. To get those frequencies, just move the cursor in the MAT-FFT display with the right or left arrow key. This will set the cursor to the next frequency used.

To calibarate a mic, I'd suggest the following procedure:

Connect reference mic to REF, and the mic to calibrate to MIC input. Playback pink noise, and perform a MAT-Transferfunction measurement (first set the delay). If you got a stable measurement, save it using the SAVE button. SELECT 'SATlive smooth frequency response' as type of trace in the save dialog.

Now you're done and you can use the trace as miccorrection.

Tomy
« Last Edit: June 06, 2011, 21:01 by TomyN »
Dipl.-Ing.(FH) Thomas Neumann
SIM-II Operator
DANTE Level1 certified

Robert

  • Newbie
  • *
  • Posts: 4
    • View Profile
Re: Mic Calibration
« Reply #6 on: June 06, 2011, 22:20 »
This is exactly the way I did measurements.

Im not an expert on digital signal processing, but I'm aware of dsp math and how things are working in digital world :)  So knowing that, I performed measurements at discrete freqs used in mic calibration editor. In my case I chose 44.1k fs but measured the amplitude only at more common freqs (20.2Hz, 25.6Hz, 31Hz, 40.4Hz, and so on), rather than on every 44.1k step freq. Generator I used is capable of generating even more discrete steps. So no problems there, values in the first post attached picture were measured at the freqs you see in the picture.

Or maybe I didn't get you right and you trying to say that I should measure at every step freq of the chosen fs?

Thanks for the procedure suggestion! I will see if I could perform the measurements that way. The only problem is that ref mic needs 200 volt bias voltage, and if I remember well it is certainly not on TRS or XLR connector. So there's no way I could connect it to any standard gear used in pro audio.


Anyway, thanks for nice discussion, Tomy! I will to try calibrate my mic with different approach and will post details as soon as I will have the time to do it.