• 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!

Help me understand my REW measurements

Which method is the optimal or does it really not matter?

What matters is to test at reasonable listening levels (around 70-80dBA at mic position), because you don’t want to fry your tweeters, and to be well above the background noise level (although sweeps are quite good at dealing with noise). Where you adjust the volume does ultimately matter very little, as long as the gain structure doesn’t become too skewed and amplifies noise.
 
What matters is to test at reasonable listening levels (around 70-80dBA at mic position), because you don’t want to fry your tweeters, and to be well above the background noise level (although sweeps are quite good at dealing with noise). Where you adjust the volume does ultimately matter very little, as long as the gain structure doesn’t become too skewed and amplifies noise.
It's funny you should say that as I've mentioned this before in another thread a while back but for me, a strange thing happens whereby, once I have finalised everything, the output of my Genelecs seems boosted. Example; say my minidsp SHD volume at 75dB is -39 on the volume dial and that is the level used throughout the measurement process, once everything is done, it seems that -39 on the volume dial suddenly equates to almost 10dB higher volume than what it was before.
It may be that I have incorrectly set the GLM volume output but I'm not sure.

I'll keep trying anyway to see if I can fine-tune my process.
 
Hi all.

Ok, so can someone please have a look at these measurements and explain to me what my next steps would be to align my subs and select a suitable crossover.

To my eyes, my subs look ok when summed together but I'm not really sure?
These are full range measurements with no eq applied other than GLM calibrated on the mains.
 

Attachments

  • Full range measurements no eq or xover.zip
    2.8 MB · Views: 111
Hi all.

Ok, so can someone please have a look at these measurements and explain to me what my next steps would be to align my subs and select a suitable crossover.

To my eyes, my subs look ok when summed together but I'm not really sure?
These are full range measurements with no eq applied other than GLM calibrated on the mains.

Actually, no EQ is fine for all measurements. Could you post a full-range sweep of the subs?
 

Attachments

  • L and R Sub Full Range.zip
    2.4 MB · Views: 104
I don't know how you guys do what you do to be honest. I get all these ideas about how I'm going to fix everything but then I look at the graphs and fall apart. :(
 
I don't know how you guys do what you do to be honest. I get all these ideas about how I'm going to fix everything but then I look at the graphs and fall apart. :(

Presuming the acoustic timing reference (left speaker) you supplied is correct, the subs are actually delayed. This is unusual as mains are often the ones needing delays. But thinking about this again, GLM processing on your main speakers might be causing the additional time delay.
 
Presuming the acoustic timing reference (left speaker) you supplied is correct, the subs are actually delayed. This is unusual as mains are often the ones needing delays. But thinking about this again, GLM processing on your main speakers might be causing the additional time delay.
Yes, I believe that the DSP within GLM causes a delay but I'm led to believe it is minimal at best although I'm unsure to be honest.
I've always used the left speaker as the reference as its in an optimal position so it would make sense to use it.

I'm going to leave it for tonight now as its nearly 21:30 over here in England and I'll tackle it again tomorrow.

Thank you for your input, it is much appreciated.
 
Yes, I believe that the DSP within GLM causes a delay but I'm led to believe it is minimal at best although I'm unsure to be honest.
I've always used the left speaker as the reference as its in an optimal position so it would make sense to use it.

1650140686429.png


According to REW right speaker is delayed by 0.3 ms (correct), L sub by -10.4 ms (correct), and R sub by -8.2 ms (also correct) -- the timing reference being the left monitor.

I'm going to leave it for tonight now as its nearly 21:30 over here in England and I'll tackle it again tomorrow.

Thank you for your input, it is much appreciated.

No worries... will post back here shortly as well.
 
Caveat emptor: I do not have GLM, but I'll operate under the presumption that subs will be treated as mono (LR signals as shared input) and time aligned to both (centered) LR speakers.


Step 1 align the IR start times:

1650144727647.png

Select either both mains or subs and click on 'Align IR start'

This will automatically apply time offsets so IR start positions will be exactly (or almost) the same for both measurements.


Step 2 use the Alignment tool to sum the mains and subs -- separately before mains+sub xo -- or alternatively, use vector averaging (+6 dB offset needs to be applied afterwards):

1650145271172.png


Notice that the room is causing a phase response difference or skew primarily of the right speaker (red trace) between 90-130 Hz. This will cause a 10-20 dB cancellation dip in the summed LR response.


Step 3 come up with a good enough theoretical target which relieves the mains from heavy bass load burden. Also create a target for the sub that is high enough to assist filling-in the bass cancellation seen the mains.

1650146004174.png


Step 4 convolve the measurements with the appropriate low-pass and high-pass filters using the Trace-arithmetic A*B function. Since REW can't create these filters correctly (with phase information) on its own, I'll use rePhase to generate the needed filters.

1650148005349.png

White trace is the imported 48dB/oct 130Hz low-pass filter. Convolve (A*B) this together with the unequalized sub response and you get the predicted (red trace) response.

1650151233586.png 1650147270121.png 1650147275949.png 1650147394490.png 1650147401314.png

One can also simulate predictions within REW the individual post filter xo magnitude and phase responses -- however, time offset needs to be applied manually so one would need to be familiar reading phase traces to be able to approximate the alignment for the latter correctly.


**edit: Actually, there is no need to use rePhase for the manual filter(s) creation and convolution step

1650198373548.png

I completely forgot about this in-built REW functionality -- used it once a long time ago -- and only remembered it now! :facepalm:


Step 5 finally align the filtered main and sub responses:


1650148590320.png

I've already applied the appropriate delay offset to the sub response manually beforehand so there is nothing to adjust above.


-----------

Overlay graph views to give you a better idea of what is happening:

1650148916629.png 1650149066340.png 1650149158591.png

I tried other combinations, BTW:

1650149276862.png



To create some very simple post alignment EQ within REW, I applied frequency dependent windowing to the response. If you don't do this beforehand, REW will create too many high Q specific filters.

1650150179240.png

*Note: if FDW 15 cycles were used, only 3 filters are required, but the correction would be more course/rough.

Remember, I only generated the ff. filters automatically -- and, loaded them to the individual responses as a bonus. If I wanted to be less conservative and more meticulous, I'd apply a few boosting EQ's as well as manually create / re-adjust some of the filters.


1650150571708.gif



GLM could do most of the stuff I showed here, I guess. But this at least gives you an idea how one might determine what's a good xo candidate, time delay offset & EQ simulation using a much more hands-on / manual method.
 

Attachments

  • 1650147993997.png
    1650147993997.png
    85.2 KB · Views: 100
Last edited:
Hi @ernestcarl

The efforts you have gone to are beyond anything I could've hoped for to be honest and I appreciate it very much!
I just have to learn how to interpret what you have done now! :D

You are correct in that I am summing the subs to mono and using the L main as a timing reference.
I have looked at aligning the subs as one via the alignment tool but they play quite well with zero delay applied.

So I kind of fell at the first hurdle really as I followed what you suggested in step 1 but clicking on "Align IR Start" for Sub L & Sub R didn't seem to do anything? Am I looking for something that isn't obvious to me?

The rest I can kind of follow so it's as simple as aligning the summed L&R plus summed Sub1 & Sub R to find the most appropriate blend between the two?

Is that right?

Do I need to input delays into the minidsp SHD? What delays did you determine to be the most appropriate?
 
I'll have to come back to this as it's Easter Sunday over here so it's an important day for my family.

Be back when I can. Thank you for the assistance up to now though! :)

If there is anything you need me to do at all please let me know.

Mike.
 
So I kind of fell at the first hurdle really as I followed what you suggested in step 1 but clicking on "Align IR Start" for Sub L & Sub R didn't seem to do anything? Am I looking for something that isn't obvious to me?

Measurements highlighted in the ALL SPL tab will have their time offsets automatically modified to approximately align at t=0

1650199670002.png


Open the 'Impulse' tab in the Overlays Window:

1650199679923.gif


Instead of manually applying time offsets to each measurement or entering delays into your EQ & crossovers software interface (e.g. miniDSP) and re-measuring again, this method is much quicker -- we are only simulating everything inside REW first before going to the final route of verification.

Getting all IRs to begin close to t=0 lets us 'align sum' or vector average optimally. Moreover, it makes it easier for us to "eyeball" and apply less time offsets to the overlayed mains & sub phase responses over in the 'alignment tool'.

An important thing to note after the initial time offset delay corrections have been applied is that whatever additional processing we apply to the responses e.g. mains and/or subs, we may have to re-adjust the time offsets again. Why? Well, processing will alter the original transfer function, but more importantly, phase profile and possibly polarity -- and in the case of certain FIR filters, cause additional overall time delay as seen in your post GLM corrected left (acoustic timing reference) and right measurements in relation to your unprocessed or unequalized subs.

The rest I can kind of follow so it's as simple as aligning the summed L&R plus summed Sub1 & Sub R to find the most appropriate blend between the two?

Is that right?

Do I need to input delays into the minidsp SHD? What delays did you determine to be the most appropriate?

Yep, just enter the EQ, ~ms delays, and HP-LP crossovers into whatever software/hardware interface it is that you use -- such as a miniDSP. You will still need verify the measurements, of course. Look at how closely the actual measured magnitude, phase, and step response of the new Subs+L+R is compared to the simulated 'aligned' one generated within REW. They should be very close or nearly identical.

Just as a starting point, use the estimated 'Delay x ms' time listed for each measurement which I also noted in post #29. Then verify and re-adjust with actual individual and summed (Subs+L+R) measurements -- look at the IR, step, magnitude and phase and compare with our simulated "optimal" EQ'd responses.


*edited Step 4 which skips the need for rePhase altogether.
 
Last edited:
This short video from Nathan Lively about picking crossover filters and pre-crossover equalization may be of interest to some:


I normally apply some EQ to the mains and sub prior to summing and don't always jump for LR filters -- sometimes it might not work as well as other options. Just remember that there isn't really one default solution that works universally well for every situation.
 
@ernestcarl

Hi. I've been playing around with REW and I think have this kind of right now.
Can you do me a big favour please? Can you confirm my findings and report if correct or not?

Using the original measurements we have been working with, I have found the following to be beneficial:

Mains 80hz LR24 xover
Subs 140hz LR48 xover
Subs Delayed by 12.42 ms
Subs Inverted
Subs +3dB (from 0dB within the alignment tool)

How does that look to you?

Obviously, I have to carry out a measured response as I don't know how the Genelecs will respond to any EQ applied via the minidsp SHD before the GLM software kicks in.
I will have to apply the necessary xovers and PEQ first.
I might try to apply the xover first, carry out a GLM sweep to see if GLM actually recognises it via the minidsp SHD.

Its all fun and games! :D

I have a question about house curves too but I'll leave that until after I am at a point where I have everything working well together......
 
Can you do me a big favour please? Can you confirm my findings and report if correct or not?

I suggest using REW's RTA window and running full-range periodic PN -- no need for a MMM, just keep the mic steady or in the MLP. This will allow you to change the delays, xo & EQ settings in real-time (don't forget to press 'reset' when you change DSP settings) and see the magnitude difference for yourself.

The amount of SPL increase in the sub is going to depend on what actually sounds good to you in your room. If you normally listen at lower volumes, you might want more of a boost -- and if you listen at louder volumes, you might want to shelve the HF a bit.
 
Could someone take a look at my measurements? I'm guessing my untreated room is screwing things up since nothing seems to make sense... but I'm probably just doing something incorrectly. Here's what I've done:

1. Removed crossovers, delays and EQ everywhere
2. Measured right speaker with acoustic time reference to itself (ref_r)
3. Measured left speaker and the sub placed next to it individually with acoustic time reference to right speaker (l and sub)
4. In addition to previous full-range measurements, I measured a shorter sweep (20-160Hz) like some people seem to prefer (prefix b_ in the measurements)
5. Applied FDW with 6 cycles to all measurements - not done in the attached file since it doesn't seem reversible?
6. Used the alignment tool to align l and sub to ref_r at around 80Hz (intended XO frequency) or whatever frequency/delay seems to work the best

I can have the phases matching OK around 80Hz, but it only seems to make the impulse and step responses worse. I tried unwrapping the phases but it didn't seem to make a difference. In practice, I'm not having huge issues with my summed responses, but they could be better and things just don't sound very tight to my untrained ear. I'm running a 2x4HD to an active DCS205-4 sub and an SMSL AD18 into Monitor Audio Bronze 2s, FWIW. Thanks in advance!
 

Attachments

  • phase_mlp.zip
    4 MB · Views: 70
Just my 2 cents:
- you have extremely high RT60 values over the full range. Above 100Hz you have values between 400 and 750ms - ideally in a living room the RT60 shall be around 400ms max (above the Schroeder frequency at least). You can battle this with physical room treatment
1656067674327.png

- the same can be seen on the Waterfall diagram too (you see that the waterfall does not come to a rest within 300ms but they 'crash' on the 'front wall') - even in the midrange, not only in the bass/sub bass region
1656067688351.png


- also I guess that you have some reflective surfaces, this can be seen from the IR curve (those spikes above 0ms)
1656067838033.png


- I guess you have a 2-way system since there is a delay between the drivers as seen on the Step Response curve - you can correct this with a DRC like Dirac Live
There is also some pre-ringing (the oscillation before 0ms) - not sure what causes that though - probably the DRC will fix that
1656067899824.png


- distortion is relatively high compared to the volume level of the measurement - again some DRC usually helps with that

1656068093267.png



I will let some more experienced members here to comment on the Excess Phase and the recommended crossover frequency (personally I would play with it around 70-80Hz and with various steepness (24-96dB/octave) and if you can vary the phase type (min-linear-mixed) that would also be interesting to see/hear
 
I thought reviving this thread was a little better than making a whole new thread. My questions is, is there a reason not to trust REW and its tools the the setup is spotless?


I made a full system FIR-EQ up until to 500 Hz by using REW and eAPO for Windows. Thing is, I somehow never got the bass right.
To specifically analyze this I measured my Front speaker and subs only between 50-70 Hz the phase for about 60 Hz on both and THEN use the alingment tool.

I aimed for phase alignment marked at 60 Hz and to my suprise REW suggested a phase flip on the subs, which I tried and did a new measurement. It also suggested a 55ms delay but I thought this a little "intense" so I opted for only the flipping for now.

Now trying the alignment tool again on this new measurement and my initial reference measurement, now the results are: instead of 55ms delay "only" 48ms delay was suggested on NO phase flip anymore.

It also sounds more coherent - mostly the light dip I had before at 60Hz is not as pronounced anymore and it all feels a little more in sync.

So, is the suggestion REW made mostly correct or did I fall into a trap here?

The mean part is, it is hard to measure since ASIO would bypass all my eAPO filters and java only supports L/R.

1724260967004.png
 
Back
Top Bottom