Please comment on my REW measurements

[kifeep]

Hello everyone. I'd really appreciate your thoughts and comments on my first serious go with REW. The data file includes what I hope is a reasonably comprehensive set of measurements and here's a capture of a few key ones:

Data file (dropbox)

I'm not sure why the MMM levels are so much lower than the sweeps, but I assume I missed a calibration step. All measurements were taken with the same REW output levels and AVR volume setting.

My understanding of the data is bit patchy:

1. FR seems not bad in the MMM measurements given how little I've done to address it, but of course there's still room for improvement.
2. Distortion graphs make sense to me, but I don't really know what are good or bad values. I'm especially concerned with the distortion in the left channel at around 70hz.
3. Don't know what to make of the impulse data.
4. I don't understand filtered IR at all.
5. I understand GD, but again, don't know what good/bad values would be. That said, the left channel seems much worse than the other two.
6. Only sort of understand RT60, but was able to find recommended values. These aren't bad, right?
7. Don't know what to make of clarity values.
8. I understand decay graphs, but don't know whether they are good, bad, or indifferent.
9. Same comment for waterfall graphs.

Any general suggestions where to focus efforts on improvement?

Thanks.

 

[staticV3]

I'm not sure why the MMM levels are so much lower than the sweeps, but I assume I missed a calibration step.

 

[OCA]

Hello everyone. I'd really appreciate your thoughts and comments on my first serious go with REW. The data file includes what I hope is a reasonably comprehensive set of measurements and here's a capture of a few key ones:

View attachment 334756

Data file (dropbox)

I'm not sure why the MMM levels are so much lower than the sweeps, but I assume I missed a calibration step. All measurements were taken with the same REW output levels and AVR volume setting.

My understanding of the data is bit patchy:

1. FR seems not bad in the MMM measurements given how little I've done to address it, but of course there's still room for improvement.
2. Distortion graphs make sense to me, but I don't really know what are good or bad values. I'm especially concerned with the distortion in the left channel at around 70hz.
3. Don't know what to make of the impulse data.
4. I don't understand filtered IR at all.
5. I understand GD, but again, don't know what good/bad values would be. That said, the left channel seems much worse than the other two.
6. Only sort of understand RT60, but was able to find recommended values. These aren't bad, right?
7. Don't know what to make of clarity values.
8. I understand decay graphs, but don't know whether they are good, bad, or indifferent.
9. Same comment for waterfall graphs.

Any general suggestions where to focus efforts on improvement?

Thanks.

You haven't shared your mdat file so cannot comment on GD, RT60 or impulse graphs. 70Hz dip in the frequency response of left speaker is not distortion, it's a dip in the volume of sound generated at 70Hz due to wall reflection cancelling out the actual signal at the LP. It will be caused by roughly 2.5m of extra distance travelled by the sound from the left speaker to the LP relative to the direct sound. Although the calculation is more complex, look for a surface 1.25m away from your left speaker for the cause of it. Interestingly, right speaker doesn't have that so I think your speakers are not symmetrically placed in the room.

Both speakers also suffer from multiple reflections between 100Hz-200Hz. MMM doesn't seem to show it but you might hear lack of bass in that region. These are related to 1m-1.4m extra distances travelled by the reflected sound, again look for 0.5m-0.7m surfaces from the speaker woofers. If you can move the speakers further away from these surfaces, you might dig them below the subwoofer XO and eliminate them. You will not be able to remove those dips with DSP.


Edit: Sorry, I only saw your dropbox mdat link after Keith's comments below!

 

[Keith_W]

Welcome to ASR! Thank you for sharing your .mdat file. There are a lot of things to see in those graphs, so maybe i'll just focus on a few. Otherwise this post could run very long and it would take me forever to type it It is easier to understand if you take it step by step. Others have a different approach, but MY approach is to make sure the FR is correct first before I look at anything else. In reality everything is inter-related, but its' just easier this way! Your RT60, GD, impulse, etc are important, but you should worry about them later. Trying to do everything all at once can quickly overwhelm you.

In the meantime, take a look at the REW help page for a primer on what all these measurements mean.

Let's start by comparing your left speaker with no sub, before and after RC:

We can see that your RC has done a decent job of flattening out the curve, but it could do better. As @OCA* has pointed out, the dips in your uncorrected curve at 70Hz and 150Hz is due to your room cancelling these frequencies at your listening position. These should NOT be corrected by boosting these frequencies (which is what your RC has done), as we will see shortly.

(* - you should check out his Youtube channel)

Comparison between sub on and sub off. I can see you are running your mains full range and your sub extends your bottom end only a little bit ...

... which might explain why your bass distortion is so high. See that peak in distortion at 70Hz and 150Hz? Compare that to Graph 1, which shows the effect of your room correction. Your RC software has boosted the output of the driver at 70Hz and 150Hz, which has resulted in these distortion peaks. Speakers naturally have higher distortion in the bass anyway, but it is considerably worsened when RC software tries to boost the bass. It is effectively forcing the speaker to do something it doesn't want to do.

So the first suggestion from looking at these curves alone would be:

1. Don't run your mains full range. I would suggest crossing them over at 80Hz.
2. Don't set your RC software to boost dips. Strategies to deal with this might involve: repositioning your speakers, getting more subwoofers, sacrificing headroom (i.e. cutting the peaks to get them closer to the dips), or ignoring them depending on the Q (i.e. width) of the dip, since these are less audible than peaks. Some would say you should use room treatment, but if I were to say that in this thread it could easily derail it. Instead I suggest you read other threads on ASR like this debate on room treatment.

 

[OCA]

Some quick additions to @Keith_W comments above from your mdat...

I assume you've measured with Audyssey reference and Dynamic EQ on (unless you're using Dirac). That will affect the shape of frequency response graphs massively at the low and high end depending on the AVR volume during measurement. Ideally, you need to measure at the level you listen to your system.

Since you correctly used left speaker as acoustic timing reference in all your sweeps, we can see that your center speaker distance setting is off by 11cm (4.38 inches) and its distance needs to be INCREASED by that amount in the AVR set up menu. Right speaker is also arriving at the LP a little too late. It's quite typical of Audyssey to set speaker distances a little off:

It can be seen from your GD graphs that your subwoofer also needs to be applied a different distance setting but hard to tell the exact amount (my guesstimate is it's arriving 0.7ms too early and needs around 25cm shorter distance setting) because you haven't measured everything in full frequency range (0-24,000Hz) needed for 48kHz sampling rate you have used. It's good REW practice to do so to be able to use trace arithmetic operations effectively. Also, you will not be able to use cross correlation alignment tool for the sub because REW will decimate IR and drop sampling rate due to short frequency range used during measurement. Subwoofer distance setting will change a lot in your bass responses and is very important to get right:

 

[kifeep]

Thank you all for your helpful responses. Sorry I didn't reply earlier, but sometimes like happens at inconvenient times. I'll followup individual points below.

 

[kifeep]

View attachment 334760

Ah, great, thanks. Not sure I ever would have found that myself.

 

[kifeep]

You haven't shared your mdat file so cannot comment on GD, RT60 or impulse graphs. 70Hz dip in the frequency response of left speaker is not distortion, it's a dip in the volume of sound generated at 70Hz due to wall reflection cancelling out the actual signal at the LP. It will be caused by roughly 2.5m of extra distance travelled by the sound from the left speaker to the LP relative to the direct sound. Although the calculation is more complex, look for a surface 1.25m away from your left speaker for the cause of it. Interestingly, right speaker doesn't have that so I think your speakers are not symmetrically placed in the room.

Both speakers also suffer from multiple reflections between 100Hz-200Hz. MMM doesn't seem to show it but you might hear lack of bass in that region. These are related to 1m-1.4m extra distances travelled by the reflected sound, again look for 0.5m-0.7m surfaces from the speaker woofers. If you can move the speakers further away from these surfaces, you might dig them below the subwoofer XO and eliminate them. You will not be able to remove those dips with DSP.


Edit: Sorry, I only saw your dropbox mdat link after Keith's comments below!

This is curious. You are correct that the speakers are not symmetrically placed in the room, but there is nothing that distance from the left speaker. The left wall is 74in (1.88m) from the left speaker. (The right wall is 62in (1.57m) from the right speaker.). The distance to the floor is considerably less than 1.25m and the distance to the ceiling is considerably more. I'm quite stumped.

Both speakers are 15in (0.38m) from the front wall. This 25% less than the bottom of your estimated range of 0.5m-0.7m but is maybe still the source of the 100Hz-200Hz reflections?

 

[kifeep]

Your RC software has boosted the output of the driver at 70Hz and 150Hz, which has resulted in these distortion peaks.

Aha! That makes perfect sense, but I hadn't connected the dots. RC is just Audessey, so it's interesting to learn that it can be overly aggressive in that area.

1. Don't run your mains full range. I would suggest crossing them over at 80Hz.
2. Don't set your RC software to boost dips. Strategies to deal with this might involve: repositioning your speakers, getting more subwoofers, sacrificing headroom (i.e. cutting the peaks to get them closer to the dips), or ignoring them depending on the Q (i.e. width) of the dip, since these are less audible than peaks. Some would say you should use room treatment, but if I were to say that in this thread it could easily derail it. Instead I suggest you read other threads on ASR like this debate on room treatment.

1. Don't run your mains full range -- The speakers have a F3 of 35Hz and I've been running them crossed at 40Hz. I can see your point that the sub doesn't extend the range very far, but when I tried crossing them higher I subjectively didn't like it as much. I should probably spend more time listening at a higher crossover setting.

2a. Don't set your RC software to boost dips -- As I mentioned, the RC was just Audesssey, so no real say in that. However I have started to play around with @OCA's 'Audessey A.R.T.' method which provides more control.

2b. repositioning your speakers -- Unfortunately I don't have a lot of leeway on position. And I'd feel better about moving them if I could figure where the reflection was...

2c. getting more subwoofers -- Worth considering

2d. sacrificing headroom -- I would be fine with this, and if I understand correctly I should be able to do this to some extent with REW filters

2e. use room treatment -- I've made a couple acoustically-relevant decor changes (thicker carpet pad, heavier drapes), but serious room treatment is definitely in the later-if-ever category.

Thanks for your suggestions.

 

[kifeep]

I assume you've measured with Audyssey reference and Dynamic EQ on (unless you're using Dirac). That will affect the shape of frequency response graphs massively at the low and high end depending on the AVR volume during measurement. Ideally, you need to measure at the level you listen to your system.

Sweep measurements were with Audyssey on, but Dynamic EQ off. I did add a small (2.5db) bass boost manually in the MultiEQ Editor app.

Since you correctly used left speaker as acoustic timing reference in all your sweeps, we can see that your center speaker distance setting is off by 11cm (4.38 inches) and its distance needs to be INCREASED by that amount in the AVR set up menu. Right speaker is also arriving at the LP a little too late. It's quite typical of Audyssey to set speaker distances a little off:

Your points are well-taken, but since I have started to mess around with your method of inserting REW filters into Audyssey, I'm going to table IR concerns until the next round of measurements. Hopefully within the next week or so.

BTW, your MultiEQ Editor hack is really cool and very clever. Thanks for sharing it.

 

[OCA]

Sweep measurements were with Audyssey on, but Dynamic EQ off. I did add a small (2.5db) bass boost manually in the MultiEQ Editor app.


Your points are well-taken, but since I have started to mess around with your method of inserting REW filters into Audyssey, I'm going to table IR concerns until the next round of measurements. Hopefully within the next week or so.

BTW, your MultiEQ Editor hack is really cool and very clever. Thanks for sharing it.

Make sure you use the latest scripts linked in the description of the Audyssey A.R.T. video, they are better optimized than the ones linked in earlier videos.