REW test variations

[Truett]

We are planning to put surrounds, heights, and ceiling speakers behind wall fabric. We want to test how the different wall fabrics affect the sound. I have set up a single Focal 1000 ICW6 In-wall speaker to test various fabric coverings for frequency response with REW Room EQ Wizard. So far we have tested with no grill, metal grill that comes with the speaker, and a thick cotton shirt, because we do not yet have the fabric samples large enough. They are coming.

Why is every sweep that we do different? Every one of them. We've done dozens in each setup. We eliminated the outliers, and we can certainly see a theme. But is this variation on a test normal? It can be 5 to 10 DB different at some frequencies on one test to another.

 

[John Mulcahy]

But is this variation on a test normal?

Difficult to know without seeing where the variations are occurring (best attaching an mdat file or a link to one) but repeated tests should be extremely consistent anywhere levels are above the noise floor.

 

[ddude003]

Any chance you can tell us or show us more about your test rig? What holds the speaker, test materiel and mic still during the testing and how far away from the speaker is the mic? And where are you and/or others in the room during the test?

 

[Truett]

Here is graph showing 5 sweeps. One after the other. Could it be something to do with input level. If I go louder it clips.

Speaker is mounted in-wall. UMIK-1 mic is 12 feet away at MLP.

 

[John Mulcahy]

Such huge differences suggest a measurement issue (e.g. dropouts) , assuming the mic is in a stand. What driver type are you using?

 

[Truett]

Using Java driver. Input level was at .53. I changed it to 1.0. But the IN and REFERENCE IN meters do not go high. Not even close to the OUT meter.

 

[John Mulcahy]

It's not a mic level problem. 0.53 is a good setting for the input level (it corresponds to 0 dB digital mic gain). USB mic input levels are typically around -40 to -50 dBFS. How is the mic supported? What is the replay chain?

 

[Truett]

Mike is supported on the little tripod that came with it. At 90° angle pointed straight up. 12 ft away from speaker at MLP. We have also put the microphone very close to the speaker, but we still get these big variations.

Computer to amplifier on hdmi. The one speaker is connected to Center channel.

Why are the IN and REFERENCE IN meters not coming up to even get close to OUT meter. OUT is -12db.

 

[John Mulcahy]

Matching levels is for line level electrical connections, not for USB mics. How long is the HDMI cable? Does the sweep sound smooth? Does it sound like a similar level to a processor level cal signal?

 

[Truett]

The HDMI cable is 20'. Sweep seems smooth to me? Sound from speaker is very loud, but IN and REFERENCE IN meters stay very low. When I change from .53 Output to .85 or 1.0 those meters seem to go up, but not much.

 

[John Mulcahy]

Can you try attaching the mdat file for some of the measurements?

 

[Truett]

Where do I find mdat files?

 

[John Mulcahy]

They are what you get when you save measurements from REW.

 

[Truett]

Here is one.

 

[Truett]

And a few more.

 

[John Mulcahy]

Huge distortion in those measurements, up to 100 Hz basically all distortion. Suggests the sweep is too loud, try turning down the amp. You should be able to comfortably listen to the sweep.

 

[Truett]

I will do that now to post the file. But something still not right. Computer then says not high enough db.

 

[John Mulcahy]

Select just one output channel rather than L+R. Can you attach screenshots of the Measure dialog and the REW soundcard preferences?

 

[Truett]

Here is lower volume sweep. Amp says -20db

 

[Truett]

Screenshots

 

[Truett]

Meters active screenshot. Look at low IN levels

 

[John Mulcahy]

Those levels are certainly not normal, background levels would normally be -70 to -60 dBFS. The USB connector in the back of those older UMIKs is very stiff, are you sure it is fully home?

 

[Truett]

Cable is tight. Brand new mic and cable. But I agree the input level must be the problem. ?? What can I test or check?

 

[John Mulcahy]

The problem might be noise cancellation processing, which Windows tends to apply by default to any mic input that provides stereo data (it assumes that one channel is signal and the other is background noise). Make sure enhancements are not enabled on the Windows audio input properties Advanced tab (System - Sound - Related Settings on right hand side - Sound Control Panel - Recording - select UMIK, click Properties button) or Enhancements tab. Some PCs may have a separate application with similar noise cancellation features for mic inputs, e,g, Waves MaXXAudio, HP Audio Center, myASUS.

 

[Truett]

We are looking into that. When we "check levels" my voice makes the IN meters go up. But the white noise does not much. My voice on top of white noise does make IN meter go up though.