Neumann KH 80 DSP room measurements

[Matias]

I am using REW + UMIK-1 for room EQ of my office setup, which is based on RME ADI-2 DAC fs + Neumann KH 80 DSP.

The 1st 2 bottom lines were done using MMM: pink noise on generator, RTA with 32 samples.

The 3rd and higher line is a single measurement holding the mic roughly where my head would be, on axis with the speakers.

Why are they so different, specially on the mids and highs? I am probably doing something (or many things) wrong, so I appreciate some feedback.

 

[staticV3]

By default, REW will not apply your UMIK Cal file to MMM measurements created using the RTA window.

There's a check box in REW's preferences which enables that.

Edit: On 2nd thought, that check box is only required for stepped measurements using the RTA window. Regular RTA measurements work fine either way.
So that's not it.

Your screenshot shows the RTA window in Spectrum mode, which is the wrong mode for PN MMM measurements.

Are you certain that all measurements were done in 1/x Octave mode?

It seems to me like your first two MMM measurements may have been created in Spectrum mode and only the last single-point measurement was taken in the correct 1/x Octave mode.

 

[Matias]

Changed to RTA 1/48 octave now, you are right, much better.
Any other tips for measuring?

 

[staticV3]

Changed to RTA 1/48 octave now, you are right, much better.
Any other tips for measuring?

View attachment 365597

As per the MMM whitepaper, use the 90deg cal file, point the mic anywhere but directly at the speakers, move slowly (<30cm/s).

https://www.ohl.to/audio/downloads/MMM-moving-mic-measurement.pdf

Also, maybe set the Pink noise Low cut to 40Hz so as to not bombard the speakers with ULF energy.

 

[Matias]

Pink noise now with high pass 40 Hz, 90° calibration file, UMIK-1 pointing at the ceiling, moving around my head position. Removed the first 2 MMMs as they were garbage anyway.

The new 90° MMM changed mostly the highs a little. Its trend is closer to the top single sweep, with less wobbling above 800 Hz.

And a tentative 5 band EQ on this last one.

 

[staticV3]

And a tentative 5 band EQ on this last one.

.8dB/Oct is rather far field target response that may sound too dull in near field.

Though if it suits your preferences, nothing wrong with that

 

[Matias]

Updated to latest REW version 5.40 beta 27, did another 90° MMM, set the RTA - Appearance option "Adjust RTA levels" to align with the single sweep.

Another tentative EQ now using 5 bands (ADI-2 limit).

 

[staticV3]

Updated to latest REW version 5.40 beta 27, did another 90° MMM, set the RTA - Appearance option "Adjust RTA levels" to align with the single sweep.

Just in case:
Always use REW's Java EXCL input type. Not regular Java.

EXCL bypasses Windows' audio pipeline and talks directly to the hardware, giving you more reliable and repeatable results.

ASIO does the same, but with less stability and you can't easily use a different DAC and ADC, so it's kind of pointless now.

 

[Matias]

Yes, it was set to EXCL on input.

I edit the post before to change the EQ to 5 bands.

Now I stored the EQ and remeasured 90° MMM same as before. Looks nicer.

Red: single sweep.
Yellow: 90° MMM.
Green: 90° MMM real measure with EQ applied.