• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

REW Questions

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,201
Likes
16,983
Location
Riverview FL
@JohnPM

REW V5.20 Beta 13

Six sweeps, acoustic timing reference, the only setup difference is the sweep duration, using each of the available settings from 128k to 4M.

Frequency Response, as expected, overlaid nicely:

1561839802088.png


Impulse Response - what's going on here?

1561839843416.png


And Step:

1561839964145.png
 
Last edited:

Daverz

Major Contributor
Joined
Mar 17, 2019
Messages
1,294
Likes
1,451
Because the measurements are different lengths, the IR delay is different for each measurement. I suggest aligning the impulse responses. On the Impulse tab click on Controls->Estimate IR Delay->Shift IR for each measurement before viewing in Overlays.
 
OP
RayDunzl

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,201
Likes
16,983
Location
Riverview FL
Because the measurements are different lengths, the IR delay is different for each measurement. I suggest aligning the impulse responses. On the Impulse tab click on Controls->Estimate IR Delay->Shift IR for each measurement before viewing in Overlays.

I'm not looking at the time offset, although I might expect them to be the same, too.

I'm looking at the difference in the shape of the impulse and step,

"It seems like" in the past, using the different sweep times gave the same result.

I did do the "Estimate IR Delay", the waveshape remained unexpectedly ugly on the longer sweeps.
 
OP
RayDunzl

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,201
Likes
16,983
Location
Riverview FL
Example:

Old sweeps, from August 2014, one with 256k, one with 1M length, with near perfect overlay:

REW V5.01 Beta 22

1561876849844.png


1561877069380.png





And a 256 and 1M sweep right now.

REW 5.20 Beta 13

Overlays don't match up. wondering why the shape, as well as the time, are different.

1561877038930.png



1561877102325.png
 
Last edited:

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
Can you pls show overlayed phase responses in 5.20 Beta 13?
 

Daverz

Major Contributor
Joined
Mar 17, 2019
Messages
1,294
Likes
1,451
I have beta 13 on my macbook. I'll try some measurements tomorrow.
 
OP
RayDunzl

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,201
Likes
16,983
Location
Riverview FL
Can you pls show overlayed phase responses in 5.20 Beta 13?

5.20 Beta 13
Red - 256k
Green - 1M sweep

1561882497161.png


Phase measures often give surprises.

Generally, I have come to expect to see "flat" from 200Hz up, due, I think, due to a lack of disturbing reflections in my setup. The jump around 200Hz (below) is likely the Crossover from panel to woofer is 180Hz.

The slope down at the higher frequencies is something new in newer versions.

Here's from five years ago:

Blue - 256k
Red - 1M - with a little surprise

1561882549739.png
 
Last edited:

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
Have you tried measure them from a smaller distance, say 0.5m? That way you eliminate reflections and get a consistent curve.

Btw, in the one from 5 years ago both graphs are practically identical - those steps at HF are 360deg apart, not sure why they happened but they look benign. Not so good in this new version.
 
Last edited:

JohnPM

Senior Member
Technical Expert
Joined
Apr 9, 2018
Messages
340
Likes
901
Location
UK
Do you have different devices for input and output? Looks like the sample rates are a little different to each other.
 
OP
RayDunzl

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,201
Likes
16,983
Location
Riverview FL
Have you tried measure them from a smaller distance, say 0.5m?


I never listen from 0.5m, nor am I building speakers, and panels are a bit strange close up, so, no.

All my measures, unless noted otherwise, are from the listening position, at about 10 feet.
 
Last edited:

JohnPM

Senior Member
Technical Expert
Joined
Apr 9, 2018
Messages
340
Likes
901
Location
UK
Blue - 256k
Red - 1M - with a little surprise

View attachment 28628
Phase wraps like that occur where there are nulls in the response from reflections, they show deviations from minimum phase behaviour. They can be a bit variable from measurement to measurement depending on whether the null is deep enough.
 
OP
RayDunzl

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,201
Likes
16,983
Location
Riverview FL
Do you have different devices for input and output? Looks like the sample rates are a little different to each other.

In the newer measurement:

Using Java REW-> PC USB -> Topping D10 -> Optical -> switch -> DEQ2496 -> miniDSP OpenDRC-DI -> Benchmark DAC2 HGC for output,

UMIK-1 for input.



Also indicative of clock mismatch.

Well, there are plenty of chances for sample rate drift/change that in the string above.



Earlier measurements would likely have been:

Optical from the PC board -> switch -> DEQ2496 -> DAC2

Two ASRC there.

Thank you.

I'll experiment again.
 

JohnPM

Senior Member
Technical Expert
Joined
Apr 9, 2018
Messages
340
Likes
901
Location
UK
It isn't sample rate conversion that is the problem, but the absolute accuracy of the clocks. See this thread for a similar issue.
 

Daverz

Major Contributor
Joined
Mar 17, 2019
Messages
1,294
Likes
1,451
So you can't get an accurate impulse response with the UMIK-1 and a long sweep because of clock drift? Seems obvious in retrospect, but I'm out $90 for a measurement tool that's inadequate for the job.

I've been doing the longest measurements possible because of the DRC-FIR recommendation to use at least a 45 second sweep to reduce noise. The author does caution that "Using two different soundcards or a CD Player for reproduction and a soundcard for recording usually provides worse results unless they are accurately time synchronized", but I didn't want to deal with the extra cables and wall warts!
 
Last edited:

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
I never listen from 0.5m, nor am I building speakers, and panels are a bit strange close up, so, no.

All my measures, unless noted otherwise, are from the listening position, at about 10 feet.

Near field loudspeaker measurement is a very common thing when measuring speakers in a room. You can't measure phase correctly in a room from that distance because of reflections. Give it a try and you will see the difference.
 

JohnPM

Senior Member
Technical Expert
Joined
Apr 9, 2018
Messages
340
Likes
901
Location
UK
So you can't get an accurate impulse response with the UMIK-1 and a long sweep because of clock drift?
No, that's not correct. I've seen a couple of examples where the clock appears to be slightly offset from 48 kHz, but all the UMIK samples I have tested (5 to date) have not shown any offset and have identical impulse responses at all sweep lengths.
 

Daverz

Major Contributor
Joined
Mar 17, 2019
Messages
1,294
Likes
1,451
No, that's not correct. I've seen a couple of examples where the clock appears to be slightly offset from 48 kHz, but all the UMIK samples I have tested (5 to date) have not shown any offset and have identical impulse responses at all sweep lengths.

I'm using my music server to play back sweep files, so I was more worried about any errors introduced by drift between the clock in my DAC and the clock in the UMIK ADC.
 

edechamps

Addicted to Fun and Learning
Forum Donor
Joined
Nov 21, 2018
Messages
910
Likes
3,620
Location
London, United Kingdom
No, that's not correct. I've seen a couple of examples where the clock appears to be slightly offset from 48 kHz, but all the UMIK samples I have tested (5 to date) have not shown any offset and have identical impulse responses at all sweep lengths.

@RayDunzl's impulse response measurements for various sweep lengths look quite different. Obviously, they can't all be correct at the same time. Most likely, all of them are wrong. It is worrying that REW would display incorrect impulse responses without warning the user that something might be off. Should REW attempt to detect clock drift and warn the user if it happens? Or better yet, should REW attempt to compensate for clock drift automatically (e.g. by resampling)?
 

dkfan9

Member
Joined
May 15, 2019
Messages
54
Likes
26
Location
Columbus, Ohio
Near field loudspeaker measurement is a very common thing when measuring speakers in a room. You can't measure phase correctly in a room from that distance because of reflections. Give it a try and you will see the difference.
Even for panels? I know a .5m panel measurement will give innaccurate FR--is phase immune to that issue? And .5m seems like it would create issues in many box speakers too, at least at crossover points.
 
Top Bottom