TOPIC:

pos wrote:

JohanW wrote:

I have extracted two frequencies(802.424 and 812.413) from several calibration files.
This must be a problem in the calibration rig or the microphones of the latest batch are really bad.

(...)

Maybe this gap is due to some mismatch between two measurement procedures used to calibrate the mic: pressure measurement for the LF, and normal gated measurement for HF. 800Hz would be a logical crossover frequency between those two techniques.
If this is the case this would be easy enough to resolve by just aligning the 800Hz and 812Hz values, and add the difference to subsequent values...

It could be scripted easily enough... I could make a website script to do it if there were interest...

The question then would be what the correct SPL calibration number is...

I have plotted three calibration files before 700-0190(top diagram) and three files after 700-0190.
It looks like it isn't enough to subtract an offset. A scaling of the values after 800Hz is also needed.

It looks like versions prior to 700-0190 also had a phase column (it can been deduced from magnitude with minimum phase relation for most of the audio band though, except maybe the UHF where the capsule will have breakups?...), and also went down to 5Hz instead of 20Hz.

That said, the upper part of the post 700-0190 curves looks much more like what one would expect (from those capsules) than the prior 700-0190 ones. The lower part is much too flat though...

JohanW wrote:

I have plotted three calibration files before 700-0190(top diagram) and three files after 700-0190.
It looks like it isn't enough to subtract an offset. A scaling of the values after 800Hz is also needed.

Very interesting. Looks like the 20.1 Hz number is a glitch too. I'm not 100% sure what you mean by "scaling of the values after 800Hz" but I suspect the peak at 10k is probably correct, the UMIK I have (early version) has that at about 2.5 dB, which isn't corrected by the cal file.

Also, comparing the "Sens Factor" at the start of the file, it's accounting for the offset at 1 kHz. So my guess is that the best/simplest way to use a (post-0190) cal file would to remove everything below 812 Hz and put in one line at 20 Hz with the same value as at 812 Hz. For example, the 7000275 file would be modified to:

"Sens Factor =-18.5026dB, SERNO: 7000275"
20 2.6590
812.413 2.6590
822.526 2.6671
832.766 2.6748

... etc...

Maybe the upper part of the new files is ok then (except the offset).
The almost flat lower part with a peak on the first sample is very strange.
I don't understand how this could be fixed without sending the mic back.

Hi Johan, the UMIK without calibration is actually fairly flat down to 20 Hz, so zeroing the cal file at low frequencies will I think give a reasonable result - within tolerance for this class of mic and assuming they are all similar, that is. I'll post a comparison sweep at low frequencies later this afternoon.

What I'm not getting is how flat and uniform the freq response is from exactly 802Hz down to 21Hz... that's across all those UMIKs and then suddenly above 812Hz they all begin to differ (I mean even after transposing the curves... BTW, after transposition it appears its my UMIK that ends up having the worst response in the group) Something just isn't right about that sudden perfectly ultra-flat response across all 3 mics... I'm sorry but it just looks fake.

monomer wrote:

BTW, after transposition it appears its my UMIK that ends up having the worst response in the group)

The shape of the raw mic response (or the cal file) isn't relevant, it's only how the mic response measures when the calibration file is applied. In that regard, once the cal file is adjusted as I suggested above, yours will probably be "better" than the early ones that didn't do the 10k peak.

Wish I had one I could test the theory on

Something just isn't right about that sudden perfectly ultra-flat response across all 3 mics... I'm sorry but it just looks fake.

Other mics (Earthworks for example) don't even go below 800 Hz in their cal files. They're within tolerance at those frequencies. The UMIK looks pretty flat down to 20 Hz, I'll get the comparison up later on so you can see.

Here's the comparison of:

Earthworks M30 (red)
Uncalibrated UMIK-1 (blue)
CSL-Calibrated EMM-6 (green)



So it's pretty close, zeroing the cal file below 800 Hz doesn't seem like an unreasonable thing to do. Of course, this is one sample.

(All three graphs were set to be 0 dB reference at 1 kHz.)

Just in case... I want to clarify that the above is a nearfield measurement of a woofer, not an acoustic calibration.

My mic is with UK customs and I'm concerned by this thread.
A wrong cal file will render the mic usless.

I've mailed minidsp to advise of my concerns and to let them know this thread is here.
Lets hope they can resolve this quickly

I was just advised my mic has left Cdn Customs and delivery is expected shortly. Very concerned about this thread.

Here you can find a full comparison of the "older" and "newer" cal-files.

www.hometheatershack.com/forums/spl-mete...e-22.html#post590250

Dear All,

Still Chinese new Year over here so thanks for your patience during this holiday break!

As I'm checking emails over the break, It came to my attention that the calibration file of the latest batch had an issue (SN# 7000190). First off, we're deeply sorry for the error and thank you for your patience in hearing back from you. There has been quite some speculation and some very good investigative work by some of you, so here is a "short" rundown on what's believed to be the cause of this unfortunate issue. Let's see how we can summarize few things:

- What's a calibration file? When microphones are manufactured at the factory, a test bed allows measurement of the specs of the systems to ensure we have a deterministic data. The purpose of the calibration file being to know the "drift" of a sensor (i.e. electret+body+USB ADC) compared to a perfect response and "compensate" for the inaccuracies of a microphone. That's the purpose of a calibration file so we're all on the same page about what we're talking about here.

- To make a microphone calibration file on the whole spectrum, you apply a different type of measurement based on the frequency range. When looking at the latest set of calibration files, it seems obvious that 2 different techniques were used pre and post 800Hz (i.e. HF measurement for more accurate results). In that process, the "scaling" of these 2 sets of data seem to have been incorrectly matched by the script used to match the 2 data. So the solution lies in matching the sensitivity of these 2 test beds and the reason why you see a major jump in frequency like a stair case.

- So the good news is that it's a simple script mistake that we unfortunately didn't catch as we were in the middle of trying to organize shipments for all of you. We're sorry of that! Being in the middle of Chinese New Year, our supplier is unfortunately on holiday till Monday but I'm sure that we'll hear back from them right away with a new set of calibration files since all measurements are backed up on servers for every serial number. The issue here being a little script and a human error.

- Why did the calibration file process changed in between the 2 batch? As we received feedback from the first batch, we realized that we could provide a better set of data especially for the high frequency. We talked to our supplier and worked on that new test bench process. Received some test samples + calibration file with the new bench and went ahead. That explains the change in calibration process. Knowing they have all the data on record for the previous batch, we'll soon update that data for all mics shipped up to date. So although it doesn't look like it from the outset with that script error on the new set, it's actually to the benefit of the miniDSP community that we did modified the process. As some of you who've been using miniDSP products for a while can attest, we're always trying to provide better value, better products by learning from our customers's feedback.

- Can I use my microphone without a calibration file? Yes, you can. Most of you will use this microphones with some sort of DSP/processing to affect the frequency response of a system based on the measurement from the microphone. Without loading the calibration file, this affordable microphone is actually already very linear (especially in the low frequency up to 1k). So even without a calibration file, it's not like you won't be able to tune your speakers and start some measurements especially if you're looking at low frequency. You can start learning REW. If you want to tune that high frequency, we'd however recommend that you wait till Monday/Tuesday so we can post a new set of files without that glitch.

Hoping this post provides a bit more insight for those of you wondering what's going on. We're happy to answer questions you have on here, so feel free to write a post or an email. My team is coming back to work tomorrow so they'll be able to start answering emails and forum posts you may have sent over the week end.

Thanks again for your patience and your support!

Tony @ miniDSP

Finally an official response on what's happened! It sure sounds like good news to me that I won't need to return my UMIK for a re-calibration. Next week is not too long to wait for a corrected calibration file. About the only question I had that wasn't directly answered however is if the new calibration file will go below 20.1Hz?