UMIK calibration off

[mcmoll]

Hi there
First post, so please be gentle if this has been discussed previously. I just recently got a brand new UMIK-1 and started it with the associated cal file in REW. Now I borrowed a very fancy HBK sound calibrator 4231 from work and wanted to check UNIK is spot on cos I was a bit unsure about the UMIK. It measures 95 DB instead of the supposed 94dB so UMIK is 1dB out even with loaded cal file .Is this normal tolerance or do I have a badly calibrated mike? When I go to calibrate button on SPL meter in REW it only gives a interface to reload the cal file but no way of an offset so I can adjust. Am I missing something? Thanks

 

[John Mulcahy]

You can change the sensitivity figure in the cal file to alter the reading, but first make sure there is a good seal to the calibrator as the UMIK diameter is slightly less than the standard calibrator size. Wrapping some tape around the tip can achieve a good seal. It isn't often that accuracy better than 1 dB is needed, mind.

 

[mcmoll]

Hey John, thanks for pitching in! I'm a rerecording mixer by trade and would like to use UMIK for my home studio. So 1dB while admittedly not game changing does make a difference for me.
I got the 12mm adaptor on for the Calibrator, but i also tried to tape it off now and it makes no difference.
I tried to alter the sens factor value and the AGain , saved and reloaded the cal file but it doesn't seem to make a difference?
Can I put a general offset in somewhere into the cal file or would I have to edit every single data point? Thanks

 

[John Mulcahy]

The sens factor is the value used for the SPL calculation. Clear the cal file in REW to remove the selection then reload it with your changed sens factor value.

 

[mcmoll]

now that I have fixed the measurement, playing (dolby) pink noise changes the value of the measurement still super fast in the slow setting. Is that cos my room is not treated yet or is the dolby integration value even slower than 1s? is there any way I can tweak it so it measure even slower?
cheers

 

[John Mulcahy]

Choose Leq on the SPL meter and reset the timer when the noise signal is playing.

 

[DanDan]

I have pointed out the SPL Sensitivity discrepancy for years. I have usually measured it as 1.8dB. Rewriting the first line in the Cal Txt File fixes the issue entirely.

 

[kpjamro]

I am having a related issue. I insert my UMIK-2 into my calibrator source set for 94 dB and record some audio on a laptop (using Audacity). I am running 48ksps and 32-bit.

When I import the audio from the laptop into REW (and use the calibration file) REW tells me the level is 104 dB, This is a neat 10 dB more than the calibrator. I also have a type 1 SPL meter. It reads 94.6 dB with the calibrator.

I am using the laptop for field recordings and running REW on my (bigger) desktop PC.

Thanks

Ken

 

[John Mulcahy]

When capturing data from a UMIK (1 or 2) directly connected to the PC the conversion to SPL depends on the cal file sensitivity figure and the Windows volume control setting, since that affects the gain applied to the mic data. If you import recorded sweeps REW has no way of telling from the data what the mic gain was when that data was captured, so if you select the option to apply cal files it uses the current audio input selection as a proxy. That assumes that (1) you have your UMIK-2 connected and selected as the audio input and (2) the PC input volume setting for the UMIK-2 is the same as it was on the device you used to capture the recording in the first place. The response shapes will be the same regardless, but preserving absolute SPL with imported recordings is tricky.

 

[kpjamro]

When capturing data from a UMIK (1 or 2) directly connected to the PC the conversion to SPL depends on the cal file sensitivity figure and the Windows volume control setting, since that affects the gain applied to the mic data. If you import recorded sweeps REW has no way of telling from the data what the mic gain was when that data was captured, so if you select the option to apply cal files it uses the current audio input selection as a proxy. That assumes that (1) you have your UMIK-2 connected and selected as the audio input and (2) the PC input volume setting for the UMIK-2 is the same as it was on the device you used to capture the recording in the first place. The response shapes will be the same regardless, but preserving absolute SPL with imported recordings is tricky.

John thanks for the fast reply. I had windows gain set at 0 which I was hoping was actually zero, not +10, but who knows...

Today I connected the UMIK directly to the desktop and compared to my SPL meter.

Calibrator = 114 dB / SPL meter=114.0 dB / UMIK/REW=114.4 dB
sounds nice but
Calibrator = 94 dB / SPL meter=94.6 / UMIK/REW= 104.1

So it reads 10 dB high (at 94) even with a direct connect (same as the imported recording). I wonder if the gain in the mic is set so high that it cannot measure 114?
I note that Audacity seems to be "pegged" at full scale trying to measure 114. Since FS for the mic is supposed to be 125 dB (per the spec sheet) I was not expecting it to be saturated at 114. It should be -11 dBFS. Perhaps the 125 dB spec is without the factors in the cal file?

FWIW
Calibration file has "Sens Factor =-10.58dB, AGain =18dB, SERNO: 8104xxx"

Can I just fiddle the value in the cal file to remove the unwanted 10 dB? (It still won't be able to measure 114 of course, but 94 and below will be right). I note that, somewhere in the manual, MiniDSP recommend calibrating using 94. Perhaps that is a hint that it cannot handle 114.

Ken

 

[kpjamro]

More info. I found an additional spec that says the sensitivity of the mic is -31.9 dBFS at 94 dB.
Playing with Audacity, I found that an input volume setting of 33 percent gives -31.9 dBFS with my calibrator running at 94 dB (dBFS in Audacity using the Wave Stats add-in function).

Writing to an output file that is subsequently read by REW I get 94.6 dBA in REW. tada.
That matches the sound meter (DSM403SD).

So that fixes the problem when using a separate-record PC. Use 33 percent gain (.33 times the voltage, squared, is ~ -10 dB).

But something is off with REW direct-connect. Is there a gain somewhere inside REW other than the cal file?

Or is the factory mic calibration wrong?

Ken

 

[John Mulcahy]

The only way to get any significant discrepancy between 94 and 114 dB on the calibrator is for the signal to clip, that should be evident on the dBFS input scale on the meter.

There is something odd going on though. I have just tried my UMIK-2 on the calibrator and it is consistently reading 10 dB low at 94, 104 and 114 dB, with either ASIO or WASAPI exclusive drivers. Sens factor for my mic is -11.33, so similar to yours. I'll look into it.

 

[John Mulcahy]

As an aside, to have the full 125 dB input range on the UMIK-2 set the input volume to no more than 60 %.

 

[kpjamro]

It seemed that setting the input volume to 33% had no effect on REW. But when I closed down and then later restarted REW it started giving the expected 94 dB. I guess you cannot change it 'on the fly' (?)

I have seen other comments in forums that these mics saturate around 115 dB. Results at 114 are not going to be reliable for some of these. I do not know if changing the firmware settings fixes this upper limit. It might be an electrical/hardware issue.

BTW - I need the REW SPL info to match a real SPL meter. My calibrator has only 94 and 114 (not 104) so I have only 1 useful data point - 94 dB.

I opened a ticket with mini dsp asking about the 33% setting in windows for UMIK-2.

Ken

 

[John Mulcahy]

It seemed that setting the input volume to 33% had no effect on REW

REW compensates for the volume setting (provided the UMIK is selected as the input device and input). That compensation happens live as the volume changes. The SPL figure should not change but the input level will change. Take care to ensure the UMIK is always selected as the input, if it is disconnected while REW is running the input selection will revert to default.

To have sufficient input headroom for 125 dB the input volume needs to be at 60% or lower.

 

[sm52]

There is something odd going on though. I have just tried my UMIK-2 on the calibrator and it is consistently reading 10 dB low at 94, 104 and 114 dB, with either ASIO or WASAPI exclusive drivers. Sens factor for my mic is -11.33, so similar to yours. I'll look into it.

What happens if you select a line not with EXCL, but the same line without EXCL? Will the readings be correct?

 

[John Mulcahy]

Yes.

 

[kpjamro]

OK more experimenting.
Selecting line input w/o EXCL raises the indicated reading by about 2.5 dB for both 94 and 114 input levels.. But the reading at 94 is about 2.5 too high so I guess that is not a good idea.

I can watch the REW SPL meter as I adjust the volume slider in Windows and see that the SPL number remains constant, while the headroom changes. That is neat.

But here is the problem: regardless of what I run the volume level at (provided I have some headroom) the difference between the 94 and 114 sound levels (on the calibrator) result in only 15 dB change in reported signal level on REW not the expected 20 dB; same for the recorded signal level on Audacity - the dBFS numbers in Wave Stats differ by 15 dB, not 20.

Remember the difference on the hardware SPL meter differ by 20 as expected.

I checked the background noise level in this room as an extra data point (around 50 dBA), As long as I put the two microphones right next to each other, the readings agree quite well - less than 1 dB.

So I think the readings from UMIK-2 will be OK as long as we are below 110 dB or better yet 107 dB.

Is it a coincidence that this number is 125 minus AGain ?


Ken

 

[kpjamro]

It seemed that setting the input volume to 33% had no effect on REW. But when I closed down and then later restarted REW it started giving the expected 94 dB. I guess you cannot change it 'on the fly' (?)

I have seen other comments in forums that these mics saturate around 115 dB. Results at 114 are not going to be reliable for some of these. I do not know if changing the firmware settings fixes this upper limit. It might be an electrical/hardware issue.

BTW - I need the REW SPL info to match a real SPL meter. My calibrator has only 94 and 114 (not 104) so I have only 1 useful data point - 94 dB.

I opened a ticket with mini dsp asking about the 33% setting in windows for UMIK-2.

Ken

I have to say the cause of this was disconnecting and reconnecting the microphone as mentioned in one of the replies. You can, in fact, change the windows gain on the fly.

 

[John Mulcahy]

The internal analog gain of the mic is adjustable, perhaps your mic's gain has somehow been set higher than it should be causing premature compression/clipping.

 

[kpjamro]

MiniDSP technical help says that clipping/compression normally starts at 110 dB. This can be increased, but results in less sensitivity at the low end. Having the 110 number on the spec sheet would be useful. This is with a single tone, which I suppose is kind of a worst-case scenario, It's also how you calibrate a microphone.

 

[armagedon]

thank you for sharing your experience.
can not understand the values in REW:

why REW do not show -12,37 value ( ?
can't figure out the 3,01dB difference

the full scale should not be 106,37 (94 + 12,37) ?
this is what the doc seems to tell : calfiles

is the "AGain =18dB" as something to do in this ?
REW do not take Again into account.
minidsp was providing a software to change this gain: link

the value of -31,9dB in the documentation is universal or must be adjusted for each mic regarding his sensitivity ?

if I lower the volume to get a value close to -31,9dB it seems to be close to the reality:

do I need to edit the calibration file to change the sens Factor to include the Again into (-30,37dB instead of -12,37dB) so I can increase the volume to 100% ?

 

[John Mulcahy]

If you change the internal analog gain you have to make a corresponding change to the sens factor, the default 18 dB AGain is already taken into account in the sens factor figure. REW compensates for the windows volume setting.

REW takes account of the "Full scale sine rms is 0 dBFS" View preference in the dBFS figures it displays, that may cause a 3.01 dB shift.

do I need to edit the calibration file to change the sens Factor to include the Again into (-30,37dB instead of -12,37dB) so I can increase the volume to 100% ?

No.

 

[armagedon]

thank you for your response.
If I uncheck the option "Full scale sine rms is 0 dBFS" the 94dB shift but in the wrong way and the full scale stay unchanged :

still feel lost

 

[John Mulcahy]

If you want to use the UMIK-2 as delivered all you need to do is load the cal file. If you doubt the SPL figures check with a mic calibrator. There's no point to setting the Windows volume level to anything other than 0dB/100% as it acts on the digital data after conversion.