Calibration with REW

[jerome]

After receiving one SVS PC-Ultra unit last week, I decided to start calibrating my system using REW. I used it previously but that's the first time I use it for sub calibration. And I want to get it seriously done this time

Currently, I only own a digital RadioShack and a laptop but I plan on buying a good mic (probably ECM8000) and a good sound card (probably E-MU 0404) very soon.
I don't have any BFD (yet), that's something I will have to decide when I get further down the road ...

My room is around 4.6m (L) x 3.55m (l) x 2.05m (H). Here is a picture to illustrate my front row:
IMG_ 012.jpg
I sit in a 3-seat sofa which is more or less one meter from the back wall. I have placed the front speakers and my sitting position using some reference music only (never used REW to optimize them).

For info: I know that I have some kind of acoustic problems (naked walls for instance). I have added two identical absorbing panels on the front and four on the back (right behind my sofa). I will start a separate thread soon to discuss acoustics and what I can do to fix the room.


I have tried many position for my sub and I get the best frequency curves when the sub is placed in the middle of the room (right or left). I don't really want to do place it there so my second best option is behind one front speaker. I arbitrary chose the left one for now. Here is my current curves (with mains):
Clipboard05.jpg

For info, this is a measurement with my front speakers only (crossover at 80Hz). The first graph is a full range measurement, the second one a zoom over the 15Hz-200Hz area:
Clipboard01.jpg
Clipboard04.jpg

The peak at around 36-38Hz is due to my room, as shown below in this table:
Clipboard08.jpg

I can't find a good reason for the 7Khz peak. Anyone ?

LF Waterfall:
Clipboard07.jpg

LF Decay time:
Clipboard06.jpg

Impulse response:
Clipboard11.jpg

I have tried to increase the sub gain and reduce the 36Hz peak using the PC-Ultra PEQ but did not succeed.


I'm relatively new to this so any advice is welcomed. I am quite happy with my 2-channel setup: Instrument location, sound stage dimensions, 3d is good and I have some work to do with treatments to improve my room acoustics.
Should I start with that before integrating my sub with REW ?
What can I do to get a better curve ?
Where does this 7KHz peak come from ?
Anything shockingly wrong ? I hope not ...

EDIT: corrected REW curves according to Bruce's recommendations. Sorry guys, I should have read the guidelines a bit better ...

[brucek]

Some quick suggestions on posting graphs.

Use a vertical axis scale of 45dB-105dB every time. This is the accepted scale and allows consistency between graphs for comparisons.

Always start the horizontal axis scale at 15Hz. Extending to less than that makes little sense as you can see the signal is clearly into the noise at that point. The upper horizontal axis should not extend beyond 20KHz. When measuring a subwoofer use 15Hz-200Hz scale.

Use dBFS scale for impulse response graphs. It better shows noise and harmonics.

Use a waterfall scale that matches the subwoofer scale of 45dB-105dB and 15Hz-200Hz (same axis and log graph) for better comparisons against response graphs.

---------------------------------

Yeah, you should have been able to easily get rid of that peak at 36Hz with the EQ of the Ultra. I can't imagine why it didn't work. It's about your only resonance you have, so a single filter in your PEQ should do it.

Your next problem is a dip around 80Hz (crossover area). You should be able to address that with the subs phase and distance timing to remove the dip.

Ignore the 7KHz peak. It is no doubt caused by the cheap Radio Shack meter. That's a standard result when trying to do full range measurements with that meter. Don't attempt measures above 5KHz with an RS meter (and that's stretching it).

------

That nice cove you have in the upper left ceiling. I wonder if it would have helped to have it extend all around the theater....

brucek

[Wayne A. Pflughaupt]

What brucek said, especially about the 7 kHz spike. If you really had an 18 dB spike there, it’d be drilling a hole in your head. Still, we’re used to seeing sharp roll-outs up there with the RS meters, not hot peaks, so I dunno...

Your waterfall looks pretty impressive. I’m guessing you’re using bass traps?

Regards,
Wayne

[jerome]

brucek wrote:

Use dBFS scale for impulse response graphs. It better shows noise and harmonics.

Ok, here is what I've got:
Clipboard01.jpg
Clipboard02.jpg
Clipboard3.jpg

I must admit that I don't understand these curves. I will have to search the forum to find more info on them.

brucek wrote:

Yeah, you should have been able to easily get rid of that peak at 36Hz with the EQ of the Ultra. I can't imagine why it didn't work. It's about your only resonance you have, so a single filter in your PEQ should do it.

I kind of got rid of it but I could not get a nice curve. It looked like an inverted house-curve (lower LF, higher HF). That's what I would like to achieve first. Removing the last 36Hz peak will probably be easy if I remove it at the end.

brucek wrote:

Your next problem is a dip around 80Hz (crossover area). You should be able to address that with the subs phase and distance timing to remove the dip.

Well, I tried to modify the sub phase but real-time measurements showed a very little modification (by 1-2dBs, not more). I will try to change the distance in my receiver as well, maybe that will help more.

brucek wrote:

Ignore the 7KHz peak. It is no doubt caused by the cheap Radio Shack meter. That's a standard result when trying to do full range measurements with that meter. Don't attempt measures above 5KHz with an RS meter (and that's stretching it).

Oh, I didn't know that. I thought that the calibration file was there to correct the Radio Shack meter. Another good reason to buy a good mic and sound card, quickly !

brucek wrote:

That nice cove you have in the upper left ceiling. I wonder if it would have helped to have it extend all around the theater....

I had to build it to hide some pipes (it's filled with isolation materials). Never thought that it could go on both sides...
Do you think I should run separate measurements on both front speakers without the sub first?

Wayne A. Pflughaupt wrote:

Your waterfall looks pretty impressive. I’m guessing you’re using bass traps?

Thanks! No, I don't use any bass traps. I only have a standard 3-seat sofa in the middle of the room.
But the walls behind are in concrete and the right wall is next to the ground outside the house. Could that be the explanation?

[Sonnie]

I'm a little confused...

For info, this is a measurement with my front speakers only (crossover at 80Hz):

Then you post two graphs:





So are these two graphs your mains only... or is the second one just your sub?

Do you have a graph of just your sub only from 15Hz to 200Hz?

[Wayne A. Pflughaupt]

jerome wrote:

Thanks! No, I don't use any bass traps. I only have a standard 3-seat sofa in the middle of the room.
But the walls behind are in concrete and the right wall is next to the ground outside the house. Could that be the explanation?

That’s probably a question best posed for the guys at our Acoustics Forum.

Regards,
Wayne

[jerome]

Sonnie wrote:

I'm a little confused...

So are these two graphs your mains only... or is the second one just your sub?

These two graphs are for my main only. I thought that it could be useful to have a 15Hz-200Hz of it to compare it with future sub+mains measurements. I have edited my post now to make it clearer.

Sonnie wrote:

Do you have a graph of just your sub only from 15Hz to 200Hz?

No, I haven't done that yet. Will do!

[jerome]

Just got my new Yamaha RX-V1800 receiver yesterday. I just had time to play around with it and try the YPAO calibration. I also have to read the user guide.

I guess that I have to start the calibration using REW all over again. That will be quite interesting to see how the curves will be with RX-V1800 vs Denon AVR3806.

Otherwise, anybody who could interpret my impulse response graphs and explain me shortly if they show that something is wrong ? Haven't had the time to read much about impulse responses yet ...

[JohnM]

There's nothing wrong with the impulse response, but it would be worth tracking down the source of the reflection at approx 6ms and trying some acoustic treatment to reduce it. The linear %FS y axis is best when doing that.

[jerome]

I've been using some time the last days to try out my new receiver and to try to find why I can't generate the digital output (SPDIF) and the SPL meter at the same time in REW. Sounds like a Windows and/or REW bug to me but I'll come back to that later on.

First, some subwoofer measurements. My sub is tuned to 15Hz, no PEQ used, crossover set to 80Hz on both receiver and sub. I thought that the most important was to get mains+sub right (even if measurement on mains or sub is incorrect). I see that many people measure sub first so I have to be wrong. Anybody has an explanation for it?


Subwoofer behind the left front speaker - Center seat:
sub_15hz_leftfront.jpg

Subwoofer behind the right front speaker - Center seat:
sub_15hz_rightfront_centerseat.jpg

Subwoofer behind the right front speaker - Left seat:
sub_15hz_rightfront_leftseat.jpg

Subwoofer behind the right front speaker - Right seat:
sub_15hz_rightfront_rightseat.jpg

Subwoofer behind the right front speaker - Left - Center - Right seats:
sub_15hz_rightfront.jpg

I took the measurements on the left and right seats to see how much the frequency curve varies across my sofa. My favorite is Subwoofer behind the right front speaker - Right seat. I tried to get the same kind of curve on the center seat but I can't get rid of this 60Hz valley. Strangely, modifying the phase does not help so much. I just saw in the RX-V1800 manual that one of the manual EQ frequencies is 63Hz. Might help to improve the curve.
Generally, what do you think of the curves?



Now my problem with REW and the SPDIF output: because my PC generates a lot of high-frequency noise on the analog output when I move the mouse or when REW updates the screen, I am trying to use the digitial/SPDIF output.
I don't experience any problems when playing music (in Winamp) or generating test sounds in the REW generator window.
But sound output is stopped as soon as I click on the Record button in the SPL Meter window. Then I hear that some kind of high frequency sound comes from my main speakers, first very low then increase slowly until becoming loud. It's not like getting an external loop between a speaker and a microphone because it stops when the sound becomes loud.
Stopping the recording makes the digital output work again, high frequency noise disappears. As expected, I experience the same problem when running a full measurement.
I experience this problem even if I disconnect the RadioShack. I don't have any built-in microphone. My laptop is an old Dell Inspirion 8100. I tried to reboot but that did not help either.
Anybody know what could be wrong here?

[JohnM]

Is REW set to control output and input volume on the SC settings? If so, does it make any difference if you uncheck the boxes to control the volume? Does sound like there is some kind of feedback loop though, may be worth posting pics of your sc mixer settings.

[jerome]

JohnM wrote:

Is REW set to control output and input volume on the SC settings? If so, does it make any difference if you uncheck the boxes to control the volume? Does sound like there is some kind of feedback loop though, may be worth posting pics of your sc mixer settings.

REW is never set to control the output volume and the master volume in Windows is set to 0 (not used for SPDIF).
I have attached to pictures to show you the different windows.

I tried to play a bit with the input volume in the settings and this is what I found:
1) If the RadioShack is turned off or the analog cable connected to LINE_IN unplugged then I don't hear any feedback into the speakers. If I modify the RadioShack scale from 80dB to 90dB then the feedback sound gets very weak but I still don't get any sound out. Scale over 90dB don't help much more
2) Having the input volume box checked or not does not make much difference. The feedback noise first gets lower then increase again progressively.

I only tried with my sub now but it sounds like the feedback sound is both louder and at a higher frequency than my 50Hz sinus test tone.

The two pictures below show how REW was configured when I tried with and without input volume control. As I wrote before, I reproduce the problem by pushing the record button in the SPL Meter window.

[brucek]

Strangely, modifying the phase does not help so much.

Phase will show no effects when viewing a single subwoofer. Phase only has an effect when two or more speakers are played together.

Now my problem with REW and the SPDIF output

Your calibration file is not valid when using digital out. The file is used to compensate for the loopback of analog line-out and line-in. I would work on the analog problem and see if you can sort that out.

Either way, I don't see how you're getting an output when your Master volume is at zero (as shown in your pics).

Here's a pic of the mixer with the associated controls and what they affect. Have you checked the soundcards own application for monitor modes being turned on?

Windows MIXER plus REW Mixer control interaction.jpg

brucek

[jerome]

brucek wrote:

Phase will show no effects when viewing a single subwoofer. Phase only has an effect when two or more speakers are played together.

Oh, I didn't know that I really thought that modifying the sub phase would modify the SPL at my listening position. Isn't the phase used to work against standing wave effects (peaks and valleys)? I do see some differences (<10dB) when going from 0 to 90 or 180.
Could you explain it to me or provide a link to a full explanation?

brucek wrote:

Your calibration file is not valid when using digital out. The file is used to compensate for the loopback of analog line-out and line-in. I would work on the analog problem and see if you can sort that out.

I tried to fix my analog problem but it's not easy. The problem is that the jack connector in my PC is a bit loose and the PC itself generates high-frequency noise. It works ok sometimes and badly most of the time, but it's ok for low-frequency measurements. I think that I will have to buy an external soundcard sooner than expected

brucek wrote:

Either way, I don't see how you're getting an output when your Master volume is at zero (as shown in your pics).

Well, I don't understand it either but the reality is that the master volume doesn't have any effect on my digital output. Looks like it works like a pre-amp output. I only have to enable the SPDIF output in the Options->Advanced Controls in the XP Master Volume window.
The master volume has an effect on the analog output of course. And there I get correct measurements and no loopback problems.

brucek wrote:

Have you checked the soundcard own application for monitor modes being turned on?

No, actually. The sound card is a ESS Maestro3i and I can't remember having any specific application for it. But I will double-check it !