Home Theater Forum and Systems banner

1 - 12 of 12 Posts

·
Registered
Joined
·
14 Posts
Discussion Starter · #1 ·
Hi all,

Could you please take a look at the screenshot below? The only difference between the measurements is the upper limit of the frequency range (50Hz, 100Hz, 200Hz, and 500Hz). Is there an explanation for why the line all of sudden becomes flat up to 60Hz when I increase the upper range of the test? Also, shouldn't the amplitude of the curve remain the same between measurements?

Thanks
 

Attachments

·
Premium Member
Joined
·
6,711 Posts
As long as you don't change anything other than the frequency range (e.g. do not change the levels) the curves should all sit on top of each other.
 

·
Premium Member
Joined
·
6,711 Posts
I don't think there is a bug, since the same test for me produces plots that overlay exactly. If you aren't changing anything I'd suspect an external factor, like a receiver applying some form of dynamic EQ or loudness adjustment.
 

·
Registered
Joined
·
14 Posts
Discussion Starter · #5 ·
I was able to reproduce the problem with a different speaker and mic (attached). My hunch is that the problem occurs around the low-end cut off of the speaker (40Hz in the second test). This time I get different curves at 20, 30, 50 & 100. Above 100 the curves start to overlap each other. I'm doing the tests in a well isolated room and at a loud-enough SPL.

I tried different sweep speeds, SPLs, and also multiple passes. But got the same result.
 

Attachments

·
Registered
Joined
·
14 Posts
Discussion Starter · #8 ·
This is the loop-back test of my audio interface done in octaves (0-20Hz, 0-40Hz, ..., 0-1280Hz). I have zoomed-in a bit on the graph, the difference between lowest and highest graphs isn't more than 3dB. So even though there is a difference between results, I don't think the audio interface is the issue. I have swapped everything else in the chain, the audio interface was my last guess.

I'll try again with my audio interface at home, and also try a different measurement software and will report back.
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
I have a hard time helping with so little info. I suspect others here may be in the same boat.
The initial charts are not mic measurements. They are way too smooth for that. The loopback may be a real measurement, but it is not normal either. It appears there is something wrong there also.

1. Can you provide info on your measuring setup; include OS, audio equipment and connections being used.
2. What are the selections in REW 'preferences/soundcard' panel? A screen shot is helpful.
3. Initially run only fullrange loopback measurements unless you are using a USB mic. There is no value in trying anything else until those results look normal.
4. Post an mdat file for us to look at.
5. If you read and followed the REW help info for setup, were there any parts that you didn't understand? If you didn't follow a setup procedure that is the best place to start. There is good info/hints in the sticky threads also.

Help us with more info and we can probably help get you started.
 

·
Registered
Joined
·
14 Posts
Discussion Starter · #10 · (Edited)
I have a hard time helping with so little info. I suspect others here may be in the same boat.
The initial charts are not mic measurements. They are way too smooth for that. The loopback may be a real measurement, but it is not normal either. It appears there is something wrong there also.

1. Can you provide info on your measuring setup; include OS, audio equipment and connections being used.
2. What are the selections in REW 'preferences/soundcard' panel? A screen shot is helpful.
3. Initially run only fullrange loopback measurements unless you are using a USB mic. There is no value in trying anything else until those results look normal.
4. Post an mdat file for us to look at.
5. If you read and followed the REW help info for setup, were there any parts that you didn't understand? If you didn't follow a setup procedure that is the best place to start. There is good info/hints in the sticky threads also.

Help us with more info and we can probably help get you started.
The initial measurements were smoothed to 1/3 of an octave, but that doesn't change the fact that they don't overlap.

1. OS: Win 10, Interface: Roland Quad-Capture, Mics: Apple Earbud's / Dayton EMM6, Speaker: M-Audio BX8
2. Attached (for the loopback I didn't use the soundcard calibration and timing reference)
4. Attached
5. No, I'm pretty good with doing measurements.

In the loopback file, notice that the last 10-100 measurements perfectly overlap (because I didn't changed anything). The issue only appears when I change the frequency range of the sweep. But regardless, these fluctuations are way smaller than the ones I got from the microphone. I'm not positive they are the same issue.
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
The measurements are fine.

SPL.png

Phase.png

IR.png

There are issues with the REW measurement settings you used:

Main Notes:
> The Impulse Response (IR) window settings was set too narrow to catch the entire IR. It is best to initially use the REW default window settings of 125/0/500. We can adjust from there if necessary to fit the needs of our analysis. Your left window was set to 0ms so much of the captured measurement was dropped.
> The REW automated 'Estimate IR Delay' function found different timings for the IRs. I manually used measurement 7 as the reference and aligned the other measurement IRs to track the same phase.

Minor note:
> The level of the 5 10-100Hz measurements were at a slightly different gain level as the SPL chart shows. I presume either the output setting, or the input gain was tweaked slightly, or maybe a different channel was used.

Comments:
> The REW 'Estimate IR Delay' function result will vary depending on several factors. Other measurement programs have this same issue. Manual shifting the IR is sometimes needed depending on the situation.
> The reconstruction filter used in the Quad-Capture REW appears to be an FIR linear phase type. This is unusual for inexpensive audio interfaces and may impact the accuracy of the REW 'Estimate IR Delay' function, but I suspect the main problem is the varying sweep lengths. In any case manual adjustment is sometimes necessary to meet the needs of the analysis as it was in this case.
 
1 - 12 of 12 Posts
Top