REW Alignment tool - Guides or Manual?

[StevieLee]

To be more specific; if negative ms delay on the subs needed and that can't be done within the DSP then increase the ms delay of all the main speakers in the DSP (if possible).
If not possible (as mains are not controlled within the DSP) then either increase the sub distance or decrease the mains distance in the AVR by the equivalent amount.

Thank you jtalden

I have some more measurements, but so far there is a noted improvement, jeez the bass now hits harder with more control.

 

[StevieLee]

Hi guys......in order for me to post a file for you to see are these the measurements you need to see?

Front left only
Sub1
Sub2
Sub3

Measured without eq from avr, and with 80hz crossover in place, and with front left as acoustic timing reference?

Kind regards.

 

[jtalden]

Objective? What equipment are you working with; DSP, AVR other? Will you be running an auto setup routine or completely DIY?
Step 1: Setup the 3 SWs to work work together well as a mono group.
Step 2: Setup SW to mains XO.
For system setup it's better to start a new threads in the AV System Setup and Support forum.

 

[StevieLee]

Sorry, should have been clearer.

Objective To ensure all subwoofer are correctly time aligned.

Working with an anthem avm60, that does not deal with time domain/distance as these have to be input manually. So will run ARC afterwards.

I have a mini dsp 2x4hd to deal with 3 subwoofer. 2 front. 1 rear.

Each sub has own output on mini dsp.

Kind regards

 

[jtalden]

Yes, the test plan is fine.

 

[StevieLee]

Yes, the test plan is fine.

I would be extremely grateful for your advice/help getting these to work well.

Sometimes I feel I'm on the correct track, then doubt my own ability and question my results.

 

[jtalden]

I will take a look a confirm your results or offer an option or two.

 

[StevieLee]

I will take a look a confirm your results or offer an option or two.

Awesome man. Thank you.

 

[Richard Feltbower]

@jtalden I wonder if I might ask your advice.
I've been discussing phase alignment with @lexicon in relation to my own system. I'm also using Dirac but via an Arcam AV860 processor with Velodyne DD12+ sub and ATC SCM40A main speakers with 90Hz crossover. I'm not currently running a centre speaker.

After running Dirac, by default it sets the Sub delay to 0ms and FL and FR at 17.6ms and 18.5ms.

I've had a look at the alignment tool in REW and the combined Sub/FL response looks pretty good based on these default values (0 and 17.6ms), being smooth and above either separate trace. The Sub/FR combined trace however requires adding -6ms delay to obtain a reasonably smooth trace. This seems to agree after listening to some music, as the default delays push the sound and bass too far over on the LHS. With the -6ms added to the FR , so now 12.5ms vs Sub, everything becomes much more central and sounds right. This also seems to correlate with a simple comparison of the FL vs FR (no Sub), with REW showing FR optimised by adding -6ms of delay relative to the FL.

I came across your guide earlier on in the thread using the trace arithmetic approach and tried this as well. If I'm following your instructions correctly, REW suggests that the optimal delay should be -7ms added to the Sub to align phase properly. However, this doesn't seem to make sense as that would put the FL and FR delays at 24.6ms and 25.5ms. It also retains the same discrepancy in alignment between FL and FR with the sound over to the LHS.

I've attached the mdat, would you mind taking a look that I'm not doing something daft?

Many thanks,
Richard.

 

[jtalden]

@jtalden
I've attached the mdat, would you mind taking a look

I'm not really following your thought process as I get confused when it is expressed differently. Does this help:
Your FL, FR are out of phase from 60-135 Hz due to room influences. Your FL and FR rolloff is also very shallow resulting in a very wide XO range. In the case of @lexicon data, while he had these same situations, Dirac BM corrected them. In this case Dirac has not corrected them.

Dirac did selected a favorable timing. I probably would have chosen a SWs delay of +7 ms from this setting. This difference may or may not significantly influence the bass sound quality as the SPL support is similar with either setting. Both are good settings and can be compared for sound quality if you like. There are other favorable SW delay settings but those start to get too far away from the direct sound timing and it is best not to stray too far from that.

We cannot have different FL vs FR timing because the mid and high frequencies would not arrive together at the LP. That would shift the entire soundstage. We also cannot have a single mono SWgroup adjusted to different timings. That means we must compromise on a delay setting to best support the L+R average and trust that the majority of the bass is panned mono. A L or R panned bass note will deviate from from smooth SPL response that we EQ to.

 

[Richard Feltbower]

I'm not really following your thought process as I get confused when it is expressed differently. Does this help:
Your FL, FR are out of phase from 60-135 Hz due to room influences. Your FL and FR rolloff is also very shallow resulting in a very wide XO range. In the case of @lexicon data, while he had these same situations, Dirac BM corrected them. In this case Dirac has not corrected them.

Dirac did selected a favorable timing. I probably would have chosen a SWs delay of +7 ms from this setting. This difference may or may not significantly influence the bass sound quality as the SPL support is similar with either setting. Both are good settings and can be compared for sound quality if you like. There are other favorable SW delay settings but those start to get too far away from the direct sound timing and it is best not to stray too far from that.

We cannot have different FL vs FR timing because the mid and high frequencies would not arrive together at the LP. That would shift the entire soundstage. We also cannot have a single mono SWgroup adjusted to different timings. That means we must compromise on a delay setting to best support the L+R average and trust that the majority of the bass is panned mono. A L or R panned bass note will deviate from from smooth SPL response that we EQ to.

Yes, very helpful thanks. I see what you mean about a Sub delay of +7ms. When selecting 'align phase at cursor', REW suggested a delay of -7ms, which is why I was questioning it as it would take the mains out to +25ms relative to the Sub. Adding +7ms to the sub (or adding -7ms to the mains) makes more sense as the overall delays on the mains relative to the Sub would then be around +10 to +11ms. I'll see how this setting sounds.

 

[jtalden]

Yes, that's good. Let us know how it works out.

Historically, I first set the reference time by shifting all measurements such that the L, R impulses are located near 0 ms. This is usually not necessary when using acoustic timing rather than loopback timing. Then the SWgroup was adjusted to find the best XO timing. That SWgroup adjustment was only converted to an 'all mains' adjustment if necessary due to settings limits within the DSP. This is a habit that eased the process when it was numerous manual operations within REW. Now with the REW alignment tool controls it is relatively easy to find the favorable delays in several ways so it is not as important, but I still get easily confused when I stray from my traditional thought process.

 

[Harbord]

I have a 2.2 audio system for music listening only.

Equipment
Mac mini - REW - miniDSP UMIK-1

Roon - RME ADI-2 DAC - miniDSP 2x4 HD - Hypex NC252MP Amp - Harbeth Stand Mount Speakers - Rythmik L12 Subwoofers

Room Layout
My main speakers are positioned 1.2m out from the front & sidewalls. My subwoofer placement is at 1/4 & 3/4 room-width on the front wall. The subwoofers are positioned against the front wall.

Volume Leveling
I have used REW to volume level, my main speakers, individually to 75dB SPL. I have volume levelled each subwoofer to 75dB SPL.

Acoustic Measurements
The miniDSP simplifies & streamlines the process of taking measurements with it's ability to mute channels etc. I am using the Right Main speaker as an acoustic timing reference for all measurements.

Assumptions
1 - Both subwoofers are against the front wall. Any phase un-alignment between each subwoofer is likely to be inconsequential. Thus to simplify and streamline the measurement process; I measure the Left + Right subwoofers as "one unit".

2 - Any low frequency 'hump' caused by volume levelling each subwoofer individually to 75db SPL will be negated by the application of REW Auto EQ; My understanding is when implementing EQ it is preferred to 'cut' rather than 'boost'.

3 - Given the room positioning of my main speakers and subwoofers; I expect to add delay to my main speakers so that they become in-phase with my subwoofers.

Objective
Identify the optimum main speaker & subwoofer integration as a function of:

• Crossover Frequency & Slopes
• Delay/Phase
• Subwoofer Gain

Process
1 - Configure miniDSP & Subwoofers with no EQ and No Delays
2 - Take the following measurements for multiple crossover points, i.e. 80Hz, 90Hz, 100Hz, 120Hz

• Left Main
• Right Main
• Left + Right Subwoofer

3 - Utilise REW Alignment Tool to model "smoothest" frequency response across the crossover point utilising the delay and gain sliders.
4 - Within miniDSP define the preferred crossover and delay; adjust subwoofer gain (if required)
5 - Measure Left Main, Right Main, Left+Right Subwoofer and apply REW Auto EQ (as required).
6 - Measure Left Main + Subwoofers, Right Main + Subwoofer to confirm frequency response matches expectations.

Question
As this is my first time integrating subwoofers, I wish to follow a relatively simple process and not worry about too many edge cases. Are my assumptions or the process I am following on the right path?

 

[jtalden]

Thanks for the background!
Yes, your approach is fine. The details can get confusing. If you have problems - ask.

 

[lexicon]

Thanks for the background!
Yes, your approach is fine. The details can get confusing. If you have problems - ask.

Good Morning Jtalden

I have an important question for you.

I have now moved on to testing what Dirac Bass Control can achieve with two identical sealed subwoofers.

We recently verified that the timings between mains and sub created by Dirac Bass Control using a single subwoofer seemed to be optimal.

We have established that DBC calculates its delays and levels from the first mic position only and loads these into the Processor once. Once these are loaded into their slots on the processor they never “seem” to change; in the processor at least they show always as that initial figure.

Having completed the remaining mic position readings Dirac does some “preprocessing” and you arrive at the filter screen where you can change curves. At that point Dirac has determined a number of “groups” for your speakers. In my case for example all of the four heights are placed in one group, the L&R in another, the centre in it’s own group etc. Both subs are, of course, logically placed in their own group.

At this stage Bass Optimisation is switched OFF and you can move speakers between groups. There are no crossovers to set.

Next stage is you turn Bass Optimisation ON.

The first thing that happens is that the speakers move automatically back into their original groups and having asked the question as to why, I understand that is currently by design.

You then click ”calculate” and after about 5-10 minutes going from 0-100% the filter screen re-appears with (different) recommended crossover lines for each group of speakers showing and the two subs have been moved to the top and do not show a crossover. Sort of logical.

You can then move the crossovers for any speaker group at will. Having decided what crossovers you want, you click ‘calculate” again and it runs the process again, then supposedly creating the perfect set of filter parameters to perfectly align phase across all speakers.

The question I have for you is this.

In your view, is that possible to achieve without altering levels or delays from that original calculated figure, as from what I have always understood, different delays would be needed for different crossovers? Do you think it possible that those initial delays could be maintained and the timing remain optimal whatever subsequent changes are made to curves and crossovers?

For each set of Dirac measurements you can move the crossovers and recalculate multiple times and even switch bass optimisation on and off as many times as you like.

But those initial delay figures in the processor never change.

 

[jtalden]

Yes, It's possible. It all depends on their approach to setting up the system. I have no knowledge of what Dirac is actually doing. I do know that one rePhase DIY approach would allow shifting the XO frequency without changing the timing. It's possible to EQ for a flat SPL and 0° phase of 2 drivers with a large overlap of the SPL extending beyond the range of optional XO frequencies. An LR24 IIR XO can then be implemented at any of the optional frequencies without impacting the timing. There would appear to me to be many practical problems that arise depending upon the particular situation.
It doesn't really matter how they do it. What matters is their results. With your particular setup their approach worked out extremely well. I was impressed that the L vs R phase issue in the lower bass range was effectively mitigated. I have not seen any other examples.

 

[lexicon]

Yes, It's possible. It all depends on their approach to setting up the system. I have no knowledge of what Dirac is actually doing. I do know that one rePhase DIY approach would allow shifting the XO frequency without changing the timing. It's possible to EQ for a flat SPL and 0° phase of 2 drivers with a large overlap of the SPL extending beyond the range of optional XO frequencies. An LR24 IIR XO can then be implemented at any of the optional frequencies without impacting the timing. There would appear to me to be many practical problems that arise depending upon the particular situation.
It doesn't really matter how they do it. What matters is their results. With your particular setup their approach worked out extremely well. I was impressed that the L vs R phase issue in the lower bass range was effectively mitigated. I have not seen any other examples.

Many thanks - that’s very helpful! My first attempt at a 2 sub arrangement with the second sub located at the other side of my room wasn’t as successful to my ears but I’ll need to have another attempt. Looking at the REW room simulator I think I should have inverted the 2nd sub which didn’t occur to me at the time.

Stacking the 2nd sub on the first and using one channel certainly provides a good improvement in bass headroom but probably isn’t the ultimate solution required to distribute the bass more evenly throughout the room. I’ll keep experimenting.

Thanks again for your advice

Jon

 

[Harbord]

Hi,
I have made measurements over the last couple of weeks.

To recap my Audio 2.2 equipment is; Roon - RME ADI-2 DAC - miniDSP 2x4 HD - Hypex NC252MP Amp - Harbeth Stand Mount Speakers - Rythmik L12 Subwoofers. My subwoofer placement is at 1/4 & 3/4 room width on the front wall. The subwoofers are positioned against the front wall.

Volume Leveling
I have volume levelled each subwoofer to 75dB SPL but goto clipping in REW. I have now volume level each subwoofer to 72dB SPL for a combined 75db SPL.

Acoustic Measurements
I am using the Right Main speaker as an acoustic timing reference for all measurements. All subwoofer measurements are as "One Unit" combined. I have taken measurements without crossovers as well as multiple crossover points. I then make a judgement on the "optimum" crossover point; I utilised REW Trace Arithmetic to create a measurement of Left + Right main. I then utilise REW Alignment tool against L+R Main and L+R Sub.

Objective
Identify the optimum main speaker & subwoofer integration as a function of: Crossover Frequency, Delay/Phase, Subwoofer Gain

Analysis
My inexperienced analysis of the attached REW Mdat file is:

• Optimum crossover = 90Hz with 48db/Oct slope
• Alignment tool delay/phase recommendation: L+R Main 6.45

Questions:
My subwoofers are located on the front wall back from my main speakers; REW measures their approximate delay as being 1.1ms. Thus I am perplexed that REW alignment tool is recommending a 6.45md delay to my L+R main, but I am inexperienced in subwoofer integration. Is my approach and results on the correct path?

I have not yet applied EQ but have been listening to music tracks to confirm the optimum subwoofer integration. Are there any suggestions on the type of music to listen to and what to listen for?

 

[jtalden]

You did this very well. This is a favorable setup. Improvements from this target are likely to insignificant.

The REW estimates of the measured impulse offsets are not used for when using the alignment tool for timing adjustments.

Note to clarify your results; the mains were adjusted in the alignment tool rather than the SWs and that a -6.45 ms delay change was found for the mains. That means the mains delay can be left unchanged (at 0 ms?) and the SWs delay can instead be increased by 6.45 ms. That may have been what you intended.

I also looked at the data for the L+R and the SWs for the 90 Hz XO. I deleted the other measurements. Attached are 2 alternate favorable timing settings. Both of these call for less delay of the SWs. One of them requires the SWs polarity to be inverted. Attached is the mdat of my analysis. In summary, the 2 optional timing settings are:

1. SWs delayed 1.5 ms and their polarity inverted.
2. Mains delayed 3.5 ms

Both of these will provide a very similar SPL to the original timing. One or both of them may sound differently however so you could experiment and chose the one you prefer. I would think option 2 is closest to the conventional target timing, then followed by option 1 that has a little more SW delay. Your original timing has even more SW delay than those two do. Even if the SPL looks slightly better with your original timing, I would not suggest that tradeoff unless you prefer the sound quality.

For test music, I do have recommendation. I suggest you make the evaluation with a selection of your favorite music. That should make the process much more enjoyable.

 

[Harbord]

Many thanks for your review and delay/phase settings suggestions. I did listen to both settings but I will need more time to determine if I can detect any differences when listening.

With regards to the REW Alignment tool, its automatic "settings" is to treat them as starting suggestions as to what is possible; We should experiment with manual adjustment of sliders.

 

[jtalden]

I use the sliders to confirm the tool setting. The tool is not intended to be an automatic setup in the sense of other AVR setup routines that provide just one answer. It is a tool to assist the user in finding one or more of the favorable timing settings for SPL support. Any of those timings that don't stray too far from the conventional target timing for the equivalent electrical XO timing is likely to provide good results.

Due to the strong influence of room on the phase at the LP it is difficult to easily identify what that equivalent timing is. Using FDW settings in the 2-5 octave range will greatly help in this regard. The SW timing and polarity that provides the closest phase tracking throughout the XO range with the FDW applied is the conventional target timing for that acoustic XO. Once that timing is determined the FDW can be removed to see if the SPL support is still still strong given the room influence at the LP. Often the timing needs to be adjusted a little to maximize the SPL. I recommend that it is better to adjust timing the the minimum additional amount needed to get strong SPL support at the LP. We can expect to find strong SPL support at the LP without the FDW using additional timing adjustments within the range of t = ±0.25 / f (where f = the XO frequency). Greater SW timing settings can still provide good SPL results, but the bass sound arrival at the LP differs from the mains more than necessary. It may be hard to detect any sound quality difference until the delay is much greater than this recommendation so it is not necessarily an big sound quality concern to stay within this range. Eventually excess delay timing can be heard.

 

[lexicon]

@jtalden Can you please kindly confirm that a positive 16ms delay is needed to Sub 2 to phase align it to Sub 1 on the attached at 40hz. In other words Sub1 will be set at 0ms and Sub2 at 16ms for say a 40hz Xover.

For some reason Dirac is allowing under 2ms between the two

I presume you align the 2 subs first?

Adding another sub has complicated this somewhat! I'm not sure of the process for that.

Many thanks

Jon

 

[jtalden]

I would align the 2 Subs first.
I found Sub 2 to need about +2 ms delay to account for the difference in distance between them. That setting results in the best overall SPL support across the entire range from 10-180 Hz. At 40 Hz they are almost exactly out of phase due to their room positions. Based on the SPL, this setting should be fine for EQ. Many other settings would also likely work okay also for such a low XO frequency when doing a DIY setup. When using Dirac, it is probably best to use the setting Dirac recommends as the EQ it calculates should be based on its calculated delay setting. That is, the SPL may not be as smooth in this range if the timing is shifted from the one Dirac recommends.

I didn't look at hte impact a 16 ms delay would have, but that is about 1/2 wavelength away at 40 Hz so the 2 subs would then be in phase near 40 Hz. Most of the rest of the range would then be out of phase. A DIY EQ may work okay for that situation as well. I might anticipate that using the Dirac calculated EQ would rolloff <30 Hz and sag some >45 Hz. Only measurements with Dirac EQ would tell.

FDW = 2

FDW = 6

No FDW

No FDW

No FDW

 

[lexicon]

Many thanks again - I’m just not getting my head around this delay calculation I don’t think so it’s probably easier just to ask you!

One point - would the timing not sound better if they were in phase at crossover and if the crossover is set low do we need to worry about the EQ being flat up to 180 or just up to a point just above the Xover?

I have limited choices really in my room:-

1. Stack the two subs in the position of sub 1 which gives me a 6db increase in headroom. Dirac loves the position of Sub1 and seems to give a fairly flat response on its own down very low. I can bump up the curve at low frequencies and it allows the adjustments to the curve. Loads of output! You only have to time one sub position.

2. Use Sub1 and Sub2 in their different locations see diagram below. Sub 1 is located exact middle of my right hand wall pointing in to the room. Sub2 is located to the left rear of my sofa. See attached Room Sim. This spreads the bass more in the room but it seems at the expense of some cancellation and SPL. Dirac struggles with the Sub2 trace with big dips at about 25hz and it can't be eliminated by tweaking that curve.

Front speaker locations are fixed to L&R of an in ceiling projector screen, there is a gas fire in the front right corner and the unit that holds all the AV electronics is located between L&R. There is a large glass patio door which takes up most of left wall. Door to room itself is in back right corner and generally left open so room is not sealed. So really my positional options for Sub2 are limited.

1. Based on the technical results only which option is best?
2. The simulation indicates gains could be made by inverting Sub2 - is that worth exploring? I presume a new Dirac run would be needed after Sub2 is set as inverted. I measured the output at 40hz sine but it did not seem to increase when it was inverted.
3. Would you still personally go for a 80hz Xover with the two subs and the JBLs? From what you say 40hz seems to be Ok?
4. Would stacking two subs in Sub 1 position + adding a third in Sub2 position be worthwhile?

Finally......

Could you kindly please explain the procedure for EQing the two subs and calculating the filters in REW if I want to try this manually. There is a Storm format output option so the .txt EQ filter files can be exported from REW and imported into the Storm by individual sub/speaker but I am not sure whether I would need to create one filter per sub or combine them in REW somehow and put the same filter on both.............. and if so how?!

Thank you again

Jon

 

[lexicon]

Here are the Left and Right JBLs to same timing ref (no sub set at large)