Issues moving from USB mic to XLR mic + sound card

[mylesok]

I recently decided to switch from my UMIK-1 to an external soundcard (focusrite Scarlett 2i2 4th) and XLR mic (Dayton EMM-6) setup thinking it would be relatively simple. So far its been pretty f***ed


SHORT VERSION: Measurements with the new setup don't match the UMIK-1, except when I turn on 'Limit cal data boost to 20db' in preferences.

LONG VERSION: I messed up and did a whole lot of measurements with the wrong mic calibration file (same model, diff unit). I loaded the correct one, re-did some measurements expecting to see little variation but it was actually pretty significant below 3kHz (1-1.5db). Later on I opened up the file on a different computer and somehow the two measurements now line up but about 20db lower than before. They also now matched very nicely with the same measurement done with my UMIK-1 (in shape, not level). After some playing around I found that the 2nd computer had 'Limit cal data boost to 20db' ticked and unticking it returned them to the original state.

I've uploaded images of the correct Cal on the EMM-6 vs UMIK-1 with and without 'Limit cal boost' checked. I adjusted the SPL of the UMIK measurement to line up in both. I suppose it means there is an issue with my new setup but I'm not sure where to start trouble shooting. Could the noise floor be an issue? It's not exactly a dead quite environment but the UMIK has never had an issue.

Might not be relevant but calibrating the SPL levels has been a huge pain. I have to turn the mic gain up quite high in the 'check levels' section, then after calibrating the SPL with an external SPL meter, it tells me that the max SPL for my measurements will be only 95db or so, unless I turn down the mic gain. It didn't seem to stop me measuring above that SPL though so I decided it wasn't going to effect the shape of the response so I could come back later to get it right.

I have been following the instructions from VituixCAD "Measurements with Room EQ Wizard (REW)for crossover simulation with VituixCAD 2"

UMIK measurements are 48kHz, Java
UMM-6 at 88.2kHz, ASIO

Thanks!

 

[John Mulcahy]

You should attach the cal file you are using.

 

[dcibel]

Change in level in those two images is ~20dB. Calibration files should not adjust level, the values should be centred around 0dB.

If you are using "use loopback as cal and timing reference", you need to leave the "limit cal boost" option unchecked, which is detailed in those instructions you've been following. Reality is that EMM-6 calibration files are not that good, in my experience they are mostly noise, doing more harm than good. Good news is that EMM-6 has a decently flat response without calibration adjustment. Best way forward would be to create your own calibration file using UMIK as the reference, simply take measurements with each mic in about as equal location as you can manage, using calibration with UMIK, and no calibration with EMM-6. Then divide the two measurements, apply some smoothing, and shift the level if needed so its centred around 0dB SPL. Save that result and use it as your cal file with EMM-6, and both your mics should produce surprisingly similar results.

 

[mylesok]

You should attach the cal file you are using.

Sure, both cal files are attatched. 32670 is the correct one. I also uploaded the REW mdat file

Change in level in those two images is ~20dB. Calibration files should not adjust level, the values should be centred around 0dB.

If you are using "use loopback as cal and timing reference", you need to leave the "limit cal boost" option unchecked, which is detailed in those instructions you've been following. Reality is that EMM-6 calibration files are not that good, in my experience they are mostly noise, doing more harm than good. Good news is that EMM-6 has a decently flat response without calibration adjustment. Best way forward would be to create your own calibration file using UMIK as the reference, simply take measurements with each mic in about as equal location as you can manage, using calibration with UMIK, and no calibration with EMM-6. Then divide the two measurements, apply some smoothing, and shift the level if needed so its centred around 0dB SPL. Save that result and use it as your cal file with EMM-6, and both your mics should produce surprisingly similar results.

Great Idea It will be interesting to see how different it is. Worth pointing out that this assumes that the UMIK is flat, and in that case the EMM-6 is much closer to flat when 'limit cal boost' is enabled in my above measurements. I'll try calibrating the soundcard in preferences and 'use loopback for timing ref' in the measurement window.

I'm just realising that I haven't paid much attention to the gain setting on the loopback input of my soundcard. Maybe that has something to do with the 20db change in level?

 

[dcibel]

There's nothing wrong with the cal files other than -40db offset. But, REW seems fine with them, open your mdat, go to the SPL tab and check the box for "mic cal" at the bottom legend to look at them yourself.

You should check input levels always and adjust your gains accordingly. It affects the SPL, but it really doesn't matter unless you really need absolute SPL. Unchecked box for "limit cal boost" is the correct way, leaving the box checked and different in input level between channels > 20dB will effectively remove the calibration.
Aligned your measurements at 1kHz, "correct cal" gets you closest to the UMIK. But, no calibration loaded at all gets you very close to the UMIK, so if you want your mics to produce the most similar measurements, just toss those EMM-6 calibration files in the bin.

 

[dcibel]

Here, try and load the attached cal file. It should make your mics match this well:

 

[John Mulcahy]

other than -40db offset

That is the problem. Cal files should be 0 dB at a reference frequency, usually 1 kHz. As far as REW is concerned the file is asking for ~40 dB correction everywhere by saying the response is -40 dB. With the cal limit enabled the cal file is being ignored as all the entries require more than 20 dB of correction. Add 40 dB to all the cal file entries (can do that in a spreadsheet) and use that instead. The start of the file is also a little odd (lines highlighted bold below), best delete those lines.

CALSOD
Frequenzgang EMM-6 32670
2
0.0
0.0
1.0
0
20.00 -39.50 0.00
20.55 -39.80 0.00
21.11 -40.00 0.00

 

[dcibel]

Dayton EMM-6 calibration is awful, I guess you get what you pay for. For an equally cheap mic with proper noise-free calibration that isn't off by 40dB, go for Sonarworks SoundID. The example calibration I downloaded was only off by 3dB .

At least, the old EMM-6 that I used to have with 4 digit SN was centred around 0 and didn't have any extra junk at the start of the file. It still didn't do much to correct the mic response, the added noise made it less flat.

 

[mylesok]

There's nothing wrong with the cal files other than -40db offset. But, REW seems fine with them, open your mdat, go to the SPL tab and check the box for "mic cal" at the bottom legend to look at them yourself.

You should check input levels always and adjust your gains accordingly. It affects the SPL, but it really doesn't matter unless you really need absolute SPL. Unchecked box for "limit cal boost" is the correct way, leaving the box checked and different in input level between channels > 20dB will effectively remove the calibration.
Aligned your measurements at 1kHz, "correct cal" gets you closest to the UMIK. But, no calibration loaded at all gets you very close to the UMIK, so if you want your mics to produce the most similar measurements, just toss those EMM-6 calibration files in the bin.

I can see your logic and believe you, but how is it possible for Dayton to create a calibration file so bad it makes the microphone worse? Especially if I can make a fairly good one with my budget USB mic . The UMIK cal file looks highly smoothed compared to EMM-6. Any ideas on why that is? Without knowing better, I would have guessed it meant the Dayton had better resolution in the calibration rather than noise. I'm not even sure how they would introduce that much noise...

That is the problem. Cal files should be 0 dB at a reference frequency, usually 1 kHz. As far as REW is concerned the file is asking for ~40 dB correction everywhere by saying the response is -40 dB. With the cal limit enabled the cal file is being ignored as all the entries require more than 20 dB of correction. Add 40 dB to all the cal file entries (can do that in a spreadsheet) and use that instead. The start of the file is also a little odd (lines highlighted bold below), best delete those lines.

CALSOD
Frequenzgang EMM-6 32670
2
0.0
0.0
1.0
0
20.00 -39.50 0.00
20.55 -39.80 0.00
21.11 -40.00 0.00

Thanks for that. Is there much point if the mic is actually very close to the UMIK when the cal if effectively disabled?