Home Theater Forum and Systems banner

1 - 20 of 22 Posts

·
Registered
Joined
·
99 Posts
Discussion Starter #1
FR file sizes smaller.JPG


(smoothed 0.2ms for easy comparison)

These graphs were taken one after the other. No changes at all.

Red = 128k
Orange = 256k
Green = 512k
Blue = 1M

Why do the reflection levels change based on measurement file size?

The data for the other measurements changes as well.
 

·
Premium Member
Joined
·
6,711 Posts
The details of each measurement (including sweep length) are also in the info panel.

The main difference in the ETC plots is from normalisation, here are the levels without normalising each to the peak:

etc.png

Why the IR peaks are so variable is more difficult to answer, I'd suspect some issue with either the measurement setup or background noise. For comparison, here are a 128k and a 1M sweep from the system on my desk, measured at a lowish level (rest of the house have gone to bed):

irs.png

The two impulse responses are very consistent, as I'd expect them to be. Here are the 128k and 1M from your file at the same scale:

others.png

Very different, there is something quite wrong there. One possibility is a sample rate conversion problem somewhere, worth checking the sample rate the various parts of the chain are set to in REW and in the Windows audio properties.
 

·
Registered
Joined
·
99 Posts
Discussion Starter #5 (Edited)
1) I did a delete properties and shut down just to reset everything.
2) checked mic properties in windows to make sure was set to 16bit 1 channel 48k
3) checked REW to make sure its sampling was set to 48k
4) re-ran measurements. same results :(

5) I tried setting REW and the mic to 44.1k. No difference.

When you speak of the REW sample rate chain, I only know of the one setting in the initial preference screen.

I am using a Dayton USB mic (Umm6)

I also turned off the air to make room as quiet as I can get it.

What now?
 

·
Registered
Joined
·
99 Posts
Discussion Starter #6
I wonder if it could be my USB path?

Computer -> 15ft usb cable -> Usb Hub

From the Usb Hub one output goes to my DAC on my stereo system via another 15ft Usb cable, a second output on the Usb Hub goes to my Usb mic via another 10ft Usb cable.

It is done this way because my measuring computer is a desktop in the next room about 20feet away.

Does that sound problematic?
 

·
Registered
Joined
·
99 Posts
Discussion Starter #7
I tried making two separate paths:

Computer Usb1 -> DAC
Computer Usb2 -> Hub -> Mic

Same results :(
 

·
Registered
Joined
·
99 Posts
Discussion Starter #9
Here is more info to help in trouble shooting:

REW Analysis.JPG

REW preferences.JPG

Java version: Version 7 Update 67
REW Version: 5.01 Beta 22 build 3423
Vista version: Home Premium SP1 (with all MS updates)
 

·
Registered
Joined
·
99 Posts
Discussion Starter #10
View attachment REW 256k twice.mdat

rew 256 twice 2.jpg

rew 256k twice.jpg

I ran the same (256k) twice, one right after the other (without moving or changing anything or course), and even in this case, my IR's dont match exactly.

I dont use or can utilize the loopback function. Perhaps this is the problem? As can be seen, the timing reference changes slightly, even in this case.
 

·
Premium Member
Joined
·
6,711 Posts
Do you have anything else you could use as an input for a measurement, an SPL meter maybe? Might even be worth trying a built-in mic just to see if the results are more consistent at different sweep lengths. Could also try a low sweep level in case there is some clipping happening somewhere, and try selecting the "High Pass" input option in REW in case there is some DC offset in the input path.
 

·
Registered
Joined
·
99 Posts
Discussion Starter #12 (Edited)
Do you have anything else you could use as an input for a measurement, an SPL meter maybe? Might even be worth trying a built-in mic just to see if the results are more consistent at different sweep lengths. Could also try a low sweep level in case there is some clipping happening somewhere, and try selecting the "High Pass" input option in REW in case there is some DC offset in the input path.
I just tried the high pass option and then a lower level measurement (-16db instead of -6db) with no change in results.

Over at AVS, in the "Simplified REW" thread, several of us are comparing results and setups on this topic. So far, two users are NOT having the problem, and one other (besides me) user IS. We havent been able to determine what is different in setup from those that are having the problem from those that are not, so far.

I dont have another mic to use for measuring.

Edit: IN playing around with different sweep frequency ranges, I notice that the disparity between the long and short sweeps narrows at t=0 when the higher frequencies are excluded. For instance, when the sweep frequencies were limited to 15-1000hz, the difference in IR were in sync at t=0 when 128k and 1M sweep lengths were run, but are off by 5db or so at 15-20000 . Dont know what this means, but thought it might mean something to you John.

But the IR graphs still dont match up nicely like yours.

15-1000.jpg

15-1000hz

15-20000.jpg

15-20000
 

·
Registered
Joined
·
261 Posts
We havent been able to determine what is different in setup from those that are having the problem from those that are not, so far
I thought markus said he doesn't have a USB mic? if so that is the main difference so far; 2 USB mic users have a problem, 2 analogue mic users do not.
 

·
Registered
Joined
·
99 Posts
Discussion Starter #14
I thought markus said he doesn't have a USB mic? if so that is the main difference so far; 2 USB mic users have a problem, 2 analogue mic users do not.
IN your post at AVS, you said "Dayton EMM-6 mic via a usb preamp/audio device (RME Fireface 800), no loopback, IRs do not vary with length".

I took that to mean you were using a USB mic since you mentioned a USB preamp. If that is not the case, then your correlation might have some bearing on this, yes.
 

·
Premium Member
Joined
·
6,711 Posts
My measurements were with a UMIK-1.

Jim, see if reducing the buffer size settings on the REW soundcard prefs to 32k input and output makes any difference. Also take a look at the Scope plot for the Captured trace after each measurement to see if there are any obvious differences in those - the plot only has data for the last measurement made, so you need to capture the graph after each measurement.
 

·
Registered
Joined
·
261 Posts
IN your post at AVS, you said "Dayton EMM-6 mic via a usb preamp/audio device (RME Fireface 800), no loopback, IRs do not vary with length".

I took that to mean you were using a USB mic since you mentioned a USB preamp. If that is not the case, then your correlation might have some bearing on this, yes.
I meant I go USB to the preamp but the mic is analog. Anyway John's reply knocks that theory out anyway.
 

·
Registered
Joined
·
99 Posts
Discussion Starter #17 (Edited)
My measurements were with a UMIK-1.

Jim, see if reducing the buffer size settings on the REW soundcard prefs to 32k input and output makes any difference. Also take a look at the Scope plot for the Captured trace after each measurement to see if there are any obvious differences in those - the plot only has data for the last measurement made, so you need to capture the graph after each measurement.
Changing the buffer size makes no difference.

scope 128.jpg

128K

scope 1m.jpg

1M

Here are the scope results for 128k and 1M at 1s
 
1 - 20 of 22 Posts
Top