First REW measurements - now what?

[spreston]

Guys:

Finally got REW up and running and had a few questions related to my 2.1 system and the attached images. Based on what I have read it seems that we should get sub sorted before moving on to mains. Would you mind providing me with some guidance relating to the following?

0. Background-
* Medium sized asymmetrical room with lots of angles; a lot of the walls run approximately 45 degrees towards ceiling
* MacBook Pro > MAudio Buddy Mic Preamp > Behringer ECM8000
* Reconditioned c.1975 2 channel pre and poweramp
* Behringer DEQ2496
* Martin Logan Descent Sub
* Customized bookshelf speakers

1. Sub location - only have two choices (see images 1-4):
- Front of room between mains or
- Alcove (far right and near rear of room)

2. Low pass filter switch on sub only provides two choices (see images 1-4):
- 70hz
- 40hz

FYI - The check levels (under "make a measurement" popup) were:
-12db in the 70hz alcove scenario
-16db in the 40hz alcove scenario
I manually adjusted the check levels for the respective front of room scenarios to facilitate graph comparison. Is that a mistake?

3. Thoughts on filters?

4. I noted some fuzziness in the REW graph of higher frequencies during the calibration of my Mac/MAudio Buddy mic Preamp; will this cause problems measuring the mains? The gain on my mic preamp is really low to avoid more fuzziness during calibration; is this a problem? (see image 5)

5. I ran a quick picture of the mains from 24" away from each speaker. Db look really high (as mic is very close) vs sub measurements and I am not quite sure how people integrate sub and mains. (see images 6 and 7). Should the higher frequencies (2k forward) be as fuzzy as they are or does this indicate a measuring issue?

6. Any other comments or suggestions would be highly appreciated

Best,

:P

1. Sub Front - 70hz low pass filter


2. Sub Alcove - 70hz low pass filter


3. Sub Front - 40hz low pass filter


4. Sub Alcove - 40hz low pass filter


5. Calibration check MacBook Pro & Audio Buddy Mic Preamp


6. Right main speaker - measured 24" from speaker


7. Left main speaker - measured 24" from speaker

[Wayne A. Pflughaupt]

Just a few quick thoughts before I shut down for the night.

All your sub locations look viable (hopefully those are unsmoothed graphs). I'd go with the 70 Hz filter, it gives the EQ more to work with, relieves strain on the mains and receiver, reduces interference from multiple low frequency generators, etc., etc.

REW can recommend filters for you.

Regards,
Wayne

[brucek]

Sub location

The alcove appears the best....

Low pass filter switch

Go with the 70Hz.

I manually adjusted the check levels for the respective front of room scenarios to facilitate graph comparison. Is that a mistake?

Your target is a bit low. Anytime you move the speaker or the microphone, simply redo the Check Levels routine and set everything back to 75dB. You are a bit low low (although the plots are fine).

Thoughts on filters?

That's what REW is for.

I noted some fuzziness in the REW graph of higher frequencies during the calibration of my Mac/MAudio Buddy mic Preamp

hehe, look at your graph scale. Each major division is 0.1dB. Set it to a sensible 1.0dB. You'll see it's fine.

I ran a quick picture of the mains from 24" away from each speaker. Db look really high (as mic is very close) vs sub measurements and I am not quite sure how people integrate sub and mains

You can measure anywhere you choose. As I said, you need to reset the Check Levels routine to 75dB. (and hopefully the first time you set Check Levels you ran Calibrate SPL, so that REW and the actual level are aligned to 75dB).

Should the higher frequencies (2k forward) be as fuzzy as they are or does this indicate a measuring issue?

It indicates standard comb filtering caused by relections. This is expected, so to reveal the underlying trend with full range measures we use the smoothing feature in REW. Set it to 1/3 octave smoothing.

Any other comments or suggestions would be highly appreciated

1. work on the area between 100-200Hz.
2. make a soundcard calibration file with the mic preamp (line-in/line-out) included in the cable loopback. It will remove some of its influence at high and low frequencies.

brucek

[spreston]

This is a lot of fun! Thanks guys for your prompt and focused responses

Wayne:

All your sub locations look viable (hopefully those are unsmoothed graphs).

* They are unsmoothed graphs

I'd go with the 70 Hz filter, it gives the EQ more to work with, relieves strain on the mains and receiver, reduces interference from multiple low frequency generators, etc., etc.

* Even though I do not have a high pass filter (yet), equalization should help relieve some strain on the mains. Both you and Bruce K recommended the 70k low pass filter...so it shall be!

Bruce K:

The alcove appears the best....

* Most controversial response I have received yet. Upon posting I was really pleased with the flat response of the Front position with 70hz low pass filter (image 1); REW did not even suggest any filters. But after reading your comment I looked a bit more carefully at the image and indeed there is a big dip between 50 and 90 hz. That dip is significantly less extreme in image 2 (alcove position), and may be covered by the mains (to be tested)

You can measure anywhere you choose. As I said, you need to reset the Check Levels routine to 75dB. (and hopefully the first time you set Check Levels you ran Calibrate SPL, so that REW and the actual level are aligned to 75dB).

* I am using the Behringer DEQ2496 with the ECM8000 mic to calibrate SPL and just bumped into your recent posting noting that the calibration needs to be manually tweeked when measuring SPL with a mic...I was puzzled by the difference between REW's db response of mains vs. sub and suspect this was the issue (to be tested)

Should the higher frequencies (2k forward) be as fuzzy as they are or does this indicate a measuring issue?
It indicates standard comb filtering caused by relections. This is expected, so to reveal the underlying trend with full range measures we use the smoothing feature in REW. Set it to 1/3 octave smoothing.

* Wow! That is more in line with what I expected to see

make a soundcard calibration file with the mic preamp (line-in/line-out) included in the cable loopback. It will remove some of its influence at high and low frequencies.

* Did soundcard/mic preamp loopback calibration before measurements, which was a challenge. (If I had to do it all over again, I would not buy the MAudio "Buddy, I may get the Behringer 802 regardless)

==> So, now need to lug the sub back to the alcove (which actually is a less obtrusive location), will remeasure sub levels and apply filters. Then look forward to integrating mains. Any other comments or recommendations are welcomed.

Thanks!

:P

[brucek]

Then look forward to integrating mains. Any other comments or recommendations are welcomed.

To integrate the mains, simply turn on the mains and redo the Check Levels routine (still using sub test signal). Then try and get the levels matched by using the subs amplifier volume control, and get the transition smooth as possible by using the subs phase control.

brucek

[spreston]

simply turn on the mains and redo the Check Levels routine (still using sub test signal)

Do you mean in the parenthetical (still using sub test signal) that in the "make a measurement" tab I should keep the 0-200 start and end frequencies I used for the sub also to check levels on the mains for purposes of mains integration?

Then do I run the "start measuring" process on the mains with the same 0-200 fq or should I change them to something like 0-20000?

If I test both mains, the bass response is higher in level than if I just run one main; for purposes of integration, does it make sense to run both mains?

Then do you just compare low fq response to that of the sub and try to integrate them at similar levels?

Thanks!

:P

[brucek]

When you add the mains to the sub, to integrate the two together, change nothing in REW from when you checked the sub alone (except you should redo Check Levels routine).

You still measure from 0-200Hz. Your chart still extends from 15Hz to 200Hz.

On the Settings page pull down for signal type for levels, you still choose Subwoofer to Check/Set levels.

Yes, it's best to put a Y-splitter at the input to the receiver, so both mains will play with the sub.

In a perfect world, you're attempting to reproduce the flat resultant signal shown in the pic below.

You can see that the area of interest is still from 15-200Hz after adding the mains.

Attempt to get the levels the same and use the phase (or in a pinch, sub distance trim in the receiver) to get the smoothest transition.




brucek

[spreston]

Thanks guys for the guidance and clarifications. The REW software and Shack are incredibly helpful, especially once you begin moving up the learning curve.

Last night I got the sub smoothly integrated with the mains using REW; sub sounded superb and was not localizable. Interesting my intuition regarding sub positioning, volume and the 25hz adjustment on the sub was generally ok, but the extra info provided by REW made a the difference between a good and superb install.

Today I decided to refine the system and set up a house curve/ hard knee; spent the day and studied the intelligence posted by Wayne & Co; highly entertaining and useful information.

I believe that a 6db house curve is in order for my room based on what appeared to be similar volumes for 28db and 70db (crossover) sine waves from my main seating position.

For everyone's entertainment, I have attached my sub set up with hard knee house curve below:

Sub low filter - 70hz
EQ on sub - 25hz, BW?, -6db
PEQ1 - 62.5hz, BW1/5, -7db
PEQ2 - 100hz, BW1/3, -12db

As recommended in Wayne's house curve memo, I will probably kill the PEQ at 100hz, given that it is above the crossover threshold of 70hz

House curve data with hard knee uploaded into REW:
Hz , db
25.0, 8
31.5, 6
40.0, 4
50.0, 2
63.0, 0




How does this picture look? I had a big node at 50hz in my prior integration and hope that has been somewhat tamed here; it reared it's ugly face when I ran sub with the mains earlier this week. Next step for Friday is to integrate mains with sub using brucek's comments above.

Quick question relating to the SPL meter function of the Behringer DEQ2496 with 8000 mic which I would like to use to manually calibrate REW (I just guess calibration now using the DEQ). I can set the Behringer meter using db(a), db(c) or no weighting; the SPL readings are quite different based on weighing but I am not sure which reading is correct. What level weighting should I be using to calibrate REW or am I stuck buying a Radio Shack SPL meter?

Thanks again and look forward to hearing how hard knee sounds with mains!

Best,

:P

[JohnM]

If you are calibrating the REW levels using a speaker cal signal you can use C weighting or no weighting, if your are using a subwoofer cal signal best to use no weighting.

[spreston]

Hi guys. I have finally incorporated the low frequency hard knee curve and mains into my measurements and was looking for some guidance from the board members. System sounds lifeless :(

The sub sounds fine and well integrated into the system (I ended up running a sine wave at crossover level of 70hz to match sub's and mains' SPLs).

Please find image #1, the response of my alcove subwoofer alone. I am pretty happy with the results, with the exception of the void around 55hz, which I seem to be stuck with in this room



Image #2 is the sub with the mains incorporated. The 55hz void moved up to 60hz and unfortunately got bigger and deeper. There is also a big void between 125hz and 175hz. Are these voids related? The room is not close to a box and has some 45 degree ceilings



Image #3 is the sub with the mains incorporated, graphed out to 20khz. This image is smoothed (at 1/3 octave). I am not too happy with this graph. More importantly, system sounds a bit lifeless.



I am running the DEQ2496 to the sub and mains. The measurements incorporate the following filters and house curve:

Sub low filter: 70hz
EQ on sub: 25hz, BW?, -6db
PEQ1: 62.5hz, BW 1/5, -7db
PEQ2 91HZ, BW 1/10, -4 (PEQ2 was not included in image #1, as this is to adjust mains)

House curve data with hard knee uploaded into REW:
Hz , db
25.0, 8
31.5, 6
40.0, 4
50.0, 2
63.0, 0

Maybe I should start by killing PEQ1, that will make the image #1 (sub alone) worse. But image #2 (sub + mains) should improve, which (I think) is what I am really interested in. I am a bit disappointed now so any suggestions would be greatly welcomed!

[brucek]

But image #2 (sub + mains) should improve, which (I think) is what I am really interested in. I am a bit disappointed now so any suggestions would be greatly welcomed!

Yeah, the end game is certainly to have the sub+mains corrected, so if that means you need to change some filters, then fine.

So, without any filtering, you feel the system is not disappointing - you liked it better? You wouldn't be the first person who doesn't like the sound of their system after properly setting it up. I wonder if it might take a while to get used to it......

brucek

[thewire]

Try your usual location with a new calibration for brief periods of time but attempt to go out of your way. Remember how this sounds. Enjoy what you are watching. Hearing is not are only sence. Try covering one side of your ears for a brief moment if you are distracted. Try and enjoy the material, have fun, do some escapism. Then the next day sit where you usually sit, and listen to the same material again. Try another location to sit if you have another and see if it sounds like right after you made your changes. Your focus of attentioin in your hearing should be more difficult to identify one sound being more better than the next, and you are not bias because your memory will have learned what has changed. I never judge how my system sounds in less than a day or two. Notice how my threads are kind of long... The less significant the change, the more challenging. If you have not trained yourself to hear, the proccess may be more difficult.

[spreston]

Thanks guys for the advice and words of encouragement...it really makes a difference. I think I got the low frequencies and integration of the sub and mains nearly as good as possible in my challenging room.

I was very frustrated with the DEQ2496's autoEQ function, which I used above 200hz. With other mains I have owned, the DEQ functioned like a dream. With my new bookshelves, everything sounds a bit veiled. I followed TheWire's advice to move around and work in shorter sessions and that really helped. Key success factors (so far):

- Moved my listening position closer to the speakers
- Lowered the maximum autoEQ adjustments to just +/- 4db per bandwidth
- Set up a full range house curve (using +8db at low end and -8db at high end - might try a bit less steep in the future, say +/- 6db)
- Removed lousy grill covers from mains - what a difference. Seems like vodoo science

System sounds pretty good...look forward to enjoying it and some more tweaking in the (not too distant) future. Thanks again for great software and an excellent support system.

:p

[spreston]

Just wanted to give you an update; system sounds fantastic!

After lots of trial and error in my tough room, key set up factors that worked for me:

* Sub and mains placement/integration via REW provided most improvement (well worth the time and effort!)
* System sounds best using only a few EQ adjustments
* Finally threw in the towel regarding EQ at higher frequencies with my mains; used PEQ only at low frequencies (music sounded veiled and unconvincing otherwise)
* Ran DEQ2496 via the tape in (to EQ both mains and sub)
* Set sub volume dial at 2 for typical listening volumes, phase dial at 90 degrees and low pass filter at 70 hz

Final PEQ settings for sub + mains, which effectively provide the hard knee house curve graphed in an earlier post (July 19, image 2):

25hz, BW?????, -6db (this is done via dial on subwoofer, not via DEQ)
56hz, BW 1/10, +6db
90hz, BW 1/10, -4db
146hz, BW 1/6, +6db

Future tweaks:

1. Figure out how to incorporate Dynamic EQ from the DEQ2496 to boost low frequencies when listening at low gain levels (like the loudness button on your receiver, but fully customized for system and room! Please see my post on the BFD forum for more exciting details)

2. Still willing to experiment with house curve for higher frequencies

Punchline: less is more

Thanks,

:p

[Wayne A. Pflughaupt]

Sorry to leave you hanging, spreston, but for some reason my e-mail notification has been spotty lately.

Are you sure those two 1/10-octave notch filters are doing anything audible? If not I'd loose them.

About your 55 or 60 Hz problem, one of your other sub locations might eliminate it. When I said back on Post #2 that all your positions were "viable," what I meant was that all of them would be easy to equalize. So if you're still having that 55-60 Hz problem, you might try one of the other locations. (Yes I know - more measurements and tweaking... )

One of the reasons your main speakers sound "veiled" might well be because response above 2 kHz is very poor, based on the graph in Post #10. A filter set on the DEQ about 1 octave wide, centered somewhere between 4-5 kHz, boosted ~5-6 dB would probably make a world of difference to your top end. If flattening that depression ends up pushing up the area above 10 kHz even higher, you can use another filter to correct that, or perhaps a shelving filter, if the DEQ has those.

Just keep in mind that this will cost you some headroom in that frequency range, so hopefully you have headroom to spare. And you might get some increased noise up there, but hopefully it won't be audible unless you get right up to the speaker. And hopefully, you'll get a substantial improvement in SQ despite these drawbacks.

Regards,
Wayne

[spreston]

Hi Wayne:

No apologies required!!! Thank you very much for taking the time to review my posting from the very beginning and responding in such detail; great comments.

- As you keenly noted, I confirmed that the two 1/10 bandwidth filters provide no audible improvement, but did make the graph look better. I will kill them

- I was quite successful dealing with the 50-60hz void when dealing with the sub alone, but once I incorporated the mains, the void reappeared. As recommended, I will try moving the sub between the speakers this week

--Someone also recommended that I try plugging the port on my mains to reduce their impact in the low frequencies; how do you do that, with a sock? Should the plug be airtight? Anecdotally, my old speakers came with some fluffy grey foam plugs for the ports, but at higher volumes they just fired out of the speakers like projectiles

- I like your idea of boosting the 2-10 Khz response; interestingly, this void does not appear in nearfield measurements of the individual mains so I suppose it is a caused by the deep shag rug, excessive cloth furniture and odd angles in the room. With respect to headroom, that is not a problem at this juncture, fortunately

Thanks again and will let you all know how this goes next week!

[bjs]

Wayne A. Pflughaupt wrote:

[FONT="Comic Sans MS"]
...One of the reasons your main speakers sound "veiled" might well be because response above 2 kHz is very poor, based on the graph in Post #10. A filter set on the DEQ about 1 octave wide, centered somewhere between 4-5 kHz, boosted ~5-6 dB would probably make a world of difference to your top end. If flattening that depression ends up pushing up the area above 10 kHz even higher, you can use another filter to correct that, or perhaps a shelving filter, if the DEQ has those.

Doesn't hurt to try it. But EQ'ing it may well make it sound worse. It would really depend if the system is minimum phase in this area (and I suspect it is not).

A great addition to REW would be showing the associated phase response including indicating the theoretical minimum phase response so we can tell which frequency areas are good candidates for EQ and which ones likely best left alone. It would also help us converge on the optimum EQ as well.

[Wayne A. Pflughaupt]

bjs wrote:

Doesn't hurt to try it. But EQ'ing it may well make it sound worse.

That's my Rule #1 for equalizing: "Don't apply any that doesn't result in an improvement in sound." See references to 1/10-octave filters in Posts #15 and #16.

Regards,
Wayne