Jump to content

Using REW to Determine Time Delays Between Drivers


Chris A

Recommended Posts

I would compare the waterfall plots to see the differences.  It is much easier to see things that way if you are only looking at the bass region.  You will see a faster decay of 20-80Hz.

 

Time aligning sub drivers with sub drivers is much easier.  Just use the impulse/etc plot to time align them.  Start with one sub, then align another to it, then align another to one of the first two subs, etc...  After that, you hook all the subs up to a common output, pick a filter, measure, then align each woofer with all the subs going at once.

 

7 minutes ago, Rudy81 said:

Last night, just coincidentally, I worked on trying to 'time' align my 4 subwoofers.  I have 4 DIY identical subs placed in the corners of the room.  I ran sweeps for all four and decided that both rear subs needed a 3.93ms. delay.  Measurement was made with the mic at ear height, room sweet spot, with mic pointing up at the ceiling.  It is a calibrated mic BTW.

 

My added delays made sense from the viewpoint that my LP is closer to the back subs than the front subs.  Based on Chris' idea, I ran the spectrogram plots and here are the results.  As you can see, the delay helped things quite a bit.  But....I think things can be improved.  In pondering how to do that, I looked at the phase plots and it is clear that the rear subs, with the delay are still a bit out of phase with the front subs.  Tomorrow I will work on aligning the phases and see if that improves the SPL plots. 

 

Clearly there is something to be said for aligning drivers, or subs in this case. I am curious to see if phase aligning all four subs will improve the SPL between 100-200hz.  I am motivated to really getting a handle on this time/phase alignment stuff. 

 

 

SPL comare 3.93ms delay.jpg

No time delay.jpg

3.93ms rear delay.jpg

 

Link to comment
Share on other sites

7 minutes ago, etc6849 said:

 

I do level, then set the crossovers, then do time alignment as I described.  Things should be done in this order.  Even setting level later will impact the crossover point (it sounds like you already are aware of this, but I'm putting it here for others).  The slope at the crossover region is seldom a brick wall, so there is plenty of overlap for the time alignment method I describe.

 

Technically, FIR filters don't impact phase (using them for woofer to midrange and midrange to tweeter), so really I could go back and adjust things out of order, but for the low order bessel filter I use for my woofers and subs I definitely can't do that obviously.

 

 

I have always been told that adding xovers, filters, EQ or doing anything to the signal will change things. So, your methodology seems correct to me. 

 

I have the luxury of setting almost any xover and playing with phase and time changes. I really need to experiment with various xovers to see which is best.  Currently, I am using L/R 24dB slopes. IIRC, I did not like the steeper slopes. It is a lot easier with just a bass bin and a two way to work on a speaker. 

 

I am hoping that focusing on my subs first will help to clear things up since there are no xovers involved and it is a simple matter of getting the best response from the 4 subs.

Link to comment
Share on other sites

@etc6849, thanks for the tip on using the IR to align the subs.  Looking at yesterdays results, I think the delay added was too much.  The IR plots for the front two subs are practically identical as are the two rear subs. The time delay I added swung the IR plots for the rear subs way past the mains.

 

Good news for me is that I don't use the subs with the mains at all.  I built the mains LF boxes to go down to 20hz....and they do.

 

I think I will work on just the subs tomorrow and then move on to the two way mains.

Link to comment
Share on other sites

14 hours ago, Rudy81 said:

In step 1 you talk about flattening the SPLs.  I have never done that and just let Audyssey Pro handle that job.  I know from taking measurements outdoors that my LF and HF drivers are actually pretty darn flat if not affected by room boundaries.  I am wondering if I should flatten the driver responses as you indicate, and leaving those settings in my crossover permanently....thus leaving less work for Audyssey to do. Opinion?

You can do that task using Audyssey, but the downside is that you're wedded to Audyssey for everything else, too.  If you separately flatten the response of each driver in the loudspeaker, then add the "ways" together, then you can control the polar lobing much more effectively using a DSP crossover.  After you flatten the responses of each "way", and set the relative channel gains so that they sum to the same overall SPL during the measurements, the next task is to set the channel delays carefully--along with the crossover filter types (Butterworth, L-R, etc.) and filter order (first, second, third order, etc.)--which are all interlinked (as you will find).  If you change a filter type or its order, the delays needed all change, too.

 

After using the DSP crossover to get things into good shape, then you can use Audyssey to take out the remaining room acoustics issues--which is what the application was designed to do.  So we have separation of concerns, and separate actions to compensate for each concern.  Much better results.

 

14 hours ago, Rudy81 said:

Next, I wonder if the readings should be taken at the 1 meter point vs. the listening position.  I am curious about this since it is the room that seems to affect the response curve as compared to a very flat response curve when readings are taken outdoors.  Certainly not questioning your expertise, but wondering which would be most effective, and what is the downside to an LP measurement. Once the timings are set you will be listening from much farther away than the measured 1 meter.  This topic has always confused me.....

I find that nothing in this subject area creates more issues and confusion than the subject you raise, quoted just above. 

 

Generally, loudspeaker driver issues are largely "minimum phase" issues, while room acoustics issues are largely "non-minimum phase" issues.  What this means is that you can effectively compensate for minimum phase issues within the loudspeakers themselves, but room acoustics issues are usually controlled by room geometries and where everything is placed--including the listening position(s).  This is one of the reasons why Toole, et al., don't like "room correction software" (i.e., firmware)--because what it's claiming to do is fairly challenged from an effectiveness viewpoint.  You can reduce gain at the room resonances to reduce peaking SPL response in-room, but you usually can do nothing about the sharp valleys of frequency response, and if you try to correct these, all you're going to do is push a lot more acoustic energy into the room at the frequencies where you don't want to put that energy.

 

When you move the microphone back to the listening position (i.e., farther back than 1 m from the loudspeaker under measurement), all you're doing is picking up a lot more non-minimum-phase behavior (early reflections) that actually swamps what you're trying to measure from the loudspeakers in-room.  This is why you measure at 1m. Once you get the loudspeakers dialed in using a DSP crossover and 1m microphone measurements, then you can worry about the room-induced issues using something like Audyssey.  For me, I don't try to correct for any non-minimum-phase room behavior, so I go with only 1m measurements and dial-in the loudspeakers.  (The results are spectacular, BTW.)

 

16 hours ago, Rudy81 said:

I use an Ashly xover, which not an option listed in REW, so I will need to find a similar xover to the Ashly.

No worries.  REW provides a lot of other DSP crossovers in their EQ facility library that can also be used, including a "generic" DSP crossover that has the following characteristics:
 

Quote

The Generic setting allows 20 filters. The adjustment ranges are:
 

Parameter Minimum Maximum Resolution
Frequency 10 22000 0.01 Hz below 100 Hz, 0.1 Hz below 1 kHz, 1 Hz above 1 kHz
Gain -120 +30 0.1 dB
Q 0.1 50 0.01

 

16 hours ago, Rudy81 said:

That looks specific to the Behringer DCX2496 DSP crossover (a unit that I actually don't recommend due to its analog section fidelity issues).  It's a description of an example process, so it has value there, but I think you need a little more information at a more general level (such as the discussion above), as well as a series of process steps in some order.  You can actually deviate from these steps, but note that it's usually better for people to learn how to crawl before walking, and walk before running.  I find that a LOT of time was wasted using less effective techniques with DSP crossovers.

 

Chris

  • Like 1
Link to comment
Share on other sites

18 hours ago, Rudy81 said:

Are you saying that I should make a 'tunnel' type of absorbent surround from the speaker to the mic? 

No--just the width of the loudspeaker on the floor.

 

16 hours ago, etc6849 said:

I have a small room, so I just did all measurements at the listening position, with the mic pointed toward the ceiling and the tip at ear height.

You also probably have a RT30 that's in the 0.1-0.2 range, which is very damped.  If it were damped any more, you could call it your anechoic chamber. :wink: 

 

The ECM8000 and UMIK-1 are both omnidirectional microphones, so all that really matters is if you're using the right calibration file, since the microphone's response in basically the same pointed straight up or at the loudspeaker. 

 

Most of the folks that are starting out doing REW and time alignment have nothing but carpet on the floor (if that) and perhaps little slivers of absorption at the "first bounce points", which do almost nothing to affect the RT30 of the listening room, and the real early reflections that mar phase and group delay measurements, as well as SPL response when the measurements allow in-room reflections to creep into them.  I don't recommend taking measurements farther back than 1m from the loudspeaker under test for some good reasons.  The tradeoffs of doing it at the LP just aren't there, in my view. 

 

16 hours ago, etc6849 said:

For the time alignment, I also did something different.  My thinking was for time alignment, the primary concern would be around the crossover region.  So I performed band limited measurements for the drivers over a reasonable overlapping region, then compared the impulse/ETC plots.

There are several plots that I use to look for channel delay adjustments, such as the spectrograms (using full sweeps) and group delay (also using full sweeps).  There is a lot of information that's useful to discern whether the issue is minimum phase or not by looking at the GD plot, and the spectrogram will always show you whether you've got real phase issues or just measurement issues of phase. I don't find the ETC plots to be useful for much other than looking for the time and magnitude of the biggest in-room reflections (which are all non-minimum phase).  Step response also shows me if I'm there or not, but it doesn't really tell you which way you need to adjust.  I've found that the most difficult measurements to do well are channel delay adjustments (after the crossover filters have been selected) and EQing for flat SPL response at lower frequencies (subwoofer frequencies), where room acoustics tend to obscure the impulse response.

 

16 hours ago, etc6849 said:

@Chris A  My step response looks very different than yours.  I think mine might look more ideal, but honestly, I have never looked at my step response so I am curious what you think.

Well, not really.  All the pre-ringing that you've got before 0ms is an issue that's pretty audible.  Step response is just that--a step in response, followed by a damped oscillatory tail--because loudspeakers don't have DC response, and there has to be as much area below zero as above to balance the step.

 

16 hours ago, etc6849 said:

Note I intentionally tilted levels when I set them up...

Why?

 

16 hours ago, etc6849 said:

One thing that is confusing is the phase shift that appears at 6kHz and 13kHz when I unwrap phase.

I'd first recommend taking your measurements at 1m with a lot of absorption on the floor, then look to see if that phase issue disappears.  If it doesn't, it says that your tweeter is time misaligned to the midrange (assuming that you're in the midrange/tweeter interference band at that point), or your driver is going into non-pistonic motion, which generally results in suddenly lagging phase as frequency increases.

 

15 hours ago, Rudy81 said:

I must be dense because I have read countless articles and books on the subject and it is still confusing on 'how to do it'.

I'm working on that.  I should have a tutorial in PDF format in the near future--in fact a two-part tutorial: one for the more basic issues, and a second one for those people that want to squeeze more performance out of their loudspeakers.

 

By the way, it turns out that setting time delays on subwoofer/bass bin channels isn't very sensitive due to the wavelengths involved and the time periods at the crossover points. Contrast that with setting the tweeter/midrange delay, and you're in the 10s of microseconds regime for setting them properly (especially if the tweeter is separate from the midrange driver (i.e., you're not using a dual-diaphragm driver), where if you stand up or sit down at your LP, you're going to have time alignment issues between the midrange and tweeter due to the very short wavelengths involved--1.35 inches at 10 kHz, which becomes an issue if the path length to the tweeter vs. the midrange changes by 1/3 inch simply because you stood up from a seated position.

 

14 hours ago, etc6849 said:

Technically, FIR filters don't impact phase (using them for woofer to midrange and midrange to tweeter)

I think you meant to say "they don't impact phase and SPL at the same time...like IIR filters do"...?  Then I understand what you've said.  BTW: I would say that you may be the only longer term member on the forum that's using FIR filters.  I would like to get to the point of being able to use them, but I've got a few things in the way from a time management perspective.

 

Chris

Link to comment
Share on other sites

12 hours ago, etc6849 said:

I would compare the waterfall plots to see the differences.

Personally, I find that the waterfall plots are nice to look at, but I can do very little with them other than to look for resonances or noise sources in-room.  You may have more luck than I do.

 

12 hours ago, Rudy81 said:

I have always been told that adding xovers, filters, EQ or doing anything to the signal will change things.

Yes.  The question is: "how much?".

 

12 hours ago, Rudy81 said:

I did not like the steeper slopes.

I agree, unless you're using FIR filtering, and even then, you have to watch for pre-ringing.  I find that steep-slope filtering is more a "solution looking for a problem" than a "solution that is often needed".

 

I find that I don't need FIR filters thus far as I can keep the phase swings well within 45 degrees for each driver/way pairing using a different technique.  In general, I use the natural roll-off of the drivers/horns themselves to create the steeper slopes, and use PEQs to trim off the SPL response that I don't need on one driver/way and/or the other.  That's described more extensively in the "Part 2" of the tutorial that I'm putting together.

 

Chris

Link to comment
Share on other sites

Chris, excellent comments. Thank you.  Your ideas on where to measure make very good sense.  In a while I am going to dive into the 4 subwoofers and see just what can be done to optimize the 4 subs in order to work together as one sub controlled by the LFE channel. 

 

If it's ok I will post results here unless you think we would be better served by a separate thread.

Link to comment
Share on other sites

I must say I'm happy to see others posting here. I was starting to feel like I was wearing out my welcome with @Chris A . I was also starting to think everyone under the sun must fully understand what I don't. ie feeling kind of slow on the uptake.

 

19 hours ago, Rudy81 said:

I just now stumbled onto this thread. Where are you placing your mic for these measurements?

 

I think Chris has pretty much given you what you need. Sorry for the delay in response. But as they say a picture is worth a 1,000 words. I have mine 1 meter from the front edge of the horn, centered on the driver with the mic aprox at a 45 degree angle. That is, not directly the driver and not at the ceiling. For sure not at the listening position for reasons Chris informed me when I first started down this path. My music room started out as just an HT but in more recent years I've focused more on 2-channel music. It is pretty dead with absorption on the walls and absorption on the ceiling front half and diffusion over the listening area. Also no sharp angles where the ceiling meets the walls; Auralex absorption around the entire perimeter of the ceiling/room interface of the room.  Still when I am measuring I put foam on the floor and blankets/dog beds on the walls between speaker and mic. I'm curious now if perhaps my floor absorption is not enough. So here is a couple of shots of a typical set up. I've also measured with the center channel out of the room and one speaker in centered in the room, away from front/side walls and absorption on the floor.

 

rps20191024_125118.thumb.jpg.0befb9022d7a07037562b56d10535de8.jpg

 

rps20191024_125057.thumb.jpg.6322f8ff00b19372cbdefff97ffc0288.jpg

 

 

16 hours ago, Rudy81 said:

However, I seem to recall that once you add the xovers it changes the phase and the timing

 

I'm sure most of you know this....but for those that might be reading that do not this is what I have learned. At first I thought the only goal was to get the peaks and valleys as flat as possible. Then I learned about the delays (beyond just distance between drivers) and Group Delay and Impulse, etc. I now check all readings after every change to see what different PEQs do to Phase, Group Delay, Spectrogram.  Likewise if I change delay I see how that impacts GD, Phase, etc. I'm yet to figure out why, but some times you can change one and see little effect on the others (which is nice). Other times one changes makes the others wildly different. It's like a Rubik's cube. Lately I'm trying not to obsess over the SPL being flat to +/- 2 or 3dB for the entire 20-20K range but have it "pretty" flat from 200-4,000 while keeping things in line with the GD, Spectrogram, Phase, etc. One thing I don't seem to be able to get is very flat phase like I see from @Chris A and @etc6849 If I unwrap my phase I can't seem to get it on the screen with my full SPL measurements. That is I always have at best 5X the -180 to +180 Jumps when "wrapped". Many times it is 8 or 9. Any pointers there or is that driver dependent? @Rudy81 What do your SPL and Phase plots look like with your Oris 150s.  

 

I want to try/learn as much as I can not just makes changes because I was told to do X, Y and Z. But there is so much to lean (at least for me). 3 maybe 4 times now since August I've totally reset my all my settings and started over with Chris' first couple of steps of Flatten individual drivers, match the gain, set the crossover and takes sweeps for delay. I don't spend nearly as much time tweaking the PEQs to obtain maximum flatness, I look at GD and Spectrogram much sooner and keep iteratively changing PEQs, delay, even Gain or high or low to strike the balance between flatness, GD, phase and Spectrogram. I wish I understood better what change impacts the others. I am getting pretty good at predicting what will happen based on a change. If I see a quick additional spike up/down in phase I can be pretty sure there will be a corresponding spike in Group Delay near that point....and most likely a worse Spectrogram.

 

So how do I minimize the number of phase changes (what is the correct term for that) from 180 to -180 and in turn get an unwrapped phase plot that fits on one screen?

 

Here is one with 5 trips up and down both wrapped and unwrapped. 5XPhase.jpg.f611352b61fcf29c020f0f2afc6137d6.jpg

 

 

1881678178_5Xunwrapped.jpg.3202ca5d5cfd6d4118527a8e24bc0d72.jpg

 

 

Here is one with, what looks to me, like a much flatter SPL plot between 100-2K but it has 8 trips of phase zig-zagging...so is the flatter SPL "better"?

 

8XPhaseFlatter.jpg.e13e0172ebce4473217dedf9bed1f2d1.jpg

 

I won't post them but take my word for it blue and yellow above both have horrible GD around 200.

 

On the other hand this one has what I think looks to be "worse" SPL flattening, but some of the best GD I've seen on my measurements. It in fact has zero PEQs just the XO and 7.5ms of delay with the phase to the HF inverted. Also a linkwitz riley XO, as I've recently started to see what differences that has over Bessel. Again more iterations as we add another dimension to the array.

 

5XPhaseNo2.jpg.d513cb59fd6c564d3733a5e801525b3e.jpg

 

Here is the GD if anyone would like to tell me if there is room for improvement. Again I'd like to know so of @Rudy81 's settings and see some plots since you are the only one I know with same HF Horns. Maybe its time to try different drivers. I like the detail and presence of these Audio Nirvana drives over the more laid back Tang Band but I wish I had kept both just to check measurements. How to the Foxtex compare on the plots?

 

5xPhaseGroupDelay.jpg.3f3aa0c77fe17d0d2d27b07d1b2fb95a.jpg

 

 

 

 

Link to comment
Share on other sites

Rich, I will work on the Oris and bass bins next.  I spent some time working on the 4 subwoofers.  Learned a lot in terms what can be done with active xovers and a little patience. In some instances, I am not sure I have done things correctly, but here are the results of today's efforts.

 

First, the setup.  4 Large "Full Marty" 18" ported subs in the corners of the room.  Room is approximately 25'8" L x 19'9"W x 9'1" H. LP is approximately 9'9" from rear wall.  Room has quite a bit of absorption and diffusion treatment. 

 

I looked at the individual sub plots with no changes of any kind and the two front speakers have nearly identical phase plots.  Same for the rear speakers.  I figured I needed to delay the rear subs by 4.08ms.  I then verified IR and phase plots.  The IR are right on top of each other.  The unwrapped phase plots stay very close until about 128hz, where there is some anomaly I cannot figure out.  I then ran a sweep with all 4 subs and then went on to EQ the DCX2496 that runs the subs.  I applied the EQ recommendations and individually tweaked a little to get the flattest response.  Here are the plots.  Overall looks good to me, I just can't figure out that 128Hz issue.  I tried all sorts of things.  The only thing that worked was playing with the rear phase of the right rear sub, but at the cost of messing up the response everywhere else.  I guess that frequency must be near room modes in all 3 dimensions.

 

If anyone has any suggestions, please chime in.

 

 

Full Sweep compare Subs.jpg

 

Above are the SPL measurements.  The green line is prior to making any time or EQ adjustments to the rear subs. The Orange line is shows the result with the rear subs delayed by 4 ms. The blue line is the final result after applying EQ to the 4 subs acting as one.

EQ Complete SPL.jpg

 

The lone blue SPL is the final result after time delay and EQ at the LP.  Overall pretty good looking graph.  BTW, I never use any smoothing, so this is the raw result.

 

REW EQ target.jpg

 

Above is the forecast that REQ showed if I were to apply the recommended filters to DCX2496.

 

Sub GD.jpg

 

Group delay plot of the 4 subs working as one.

EQ Complete 4 subs.jpg

Unwrapped Phase 4 subs with time align.jpg

 

These are the 4 subs' phase, unwrapped.  The two rear subs have the time delay applied.  Without the time delay applied, their phase was not close to the two fronts.  Not sure why the divergence nearing 120Hz, but maybe someone can help explain this result.

Link to comment
Share on other sites

31 minutes ago, Rudy81 said:

BTW, I never use any smoothing, so this is the raw result

 

Why is that?

 

 

 

32 minutes ago, Rudy81 said:

Sub GD.jpg

 

Group delay plot of the 4 subs working as one.

 

 

I am by no means an expert and I realize these are subs so we are dealing with stuff well below 200Hz but I think you need to adjust your scale. Look up at my red Group Delay plot a couple of post above. Single digit ms readings most everything below 4ms with the stuff that matters below 1ms. Yours look mostly look to be 20-40ms. Mine below 100Hz rises quickly so many I'm off the mark here, hopefully @Chris A will set me straight if I am off but 30, 40, 50ms seem very high to me. Perhaps apples to oranges once we get below 200Hz. I'll be curious to see what sort of Phase, GD, Spectrogram and Impulse plots you have for the Oris 150 + your beautiful dual driver bass bins. I need to commission a pair from you.:emotion-21:

Link to comment
Share on other sites

@rplace, good point on the GD scale.  In all honesty, I barely understand what group delay is telling me but added the graph for Chris since he discussed the GD issue earlier. I will expand the scale and post again.

 

I played around with various delays and phase changes to get the max SPL when using all 4 subs.  I could not improve much on what I posted.  Let me know if there anything else I can change to improve things.

 

I just moved the mic over to the Oris, so here we go with the mains.

Oris setupSM.jpg

Link to comment
Share on other sites

The vertical minor scale of group delay needs to be in the 1 ms or 0.5 ms range (not 50 ms) in order to see all the issues that are likely audible.

 

Chris

Link to comment
Share on other sites

5 minutes ago, Chris A said:

The vertical minor scale of group delay needs to be in the 1 ms or 0.5 ms range (not 50 ms) in order to see all the issues that are likely audible.

 

Chris

 

Well, this isn't pretty! 😨

 

Chris, what does the graph tell me?

 

 

GD expanded scale.jpg

Link to comment
Share on other sites

Looks like a fair amount of measurement noise from room boundaries (notably the floor) and other nearfield objects within 3 feet of the microphone and loudspeaker under measurement, assuming the microphone is 1m from the loudspeaker. 

 

The GD trace is typically well below 1 ms from 1-20 kHz when everything is dialed-in, with local disturbances only around the points where there is non-pistonic motion of the driver diaphragm, internal driver resonances, or time misalignments between "ways" in a tweeter/midrange crossover interference band.

 

Chris

Link to comment
Share on other sites

5 minutes ago, Chris A said:

Looks like a fair amount of measurement noise from room boundaries (notably the floor) and other nearfield objects within 3 feet of the microphone and loudspeaker under measurement, assuming the microphone is 1m from the loudspeaker. 

 

The GD trace is typically well below 1 ms from 1-20 kHz when everything is dialed-in, with local disturbances only around the points where there is non-pistonic motion of the driver diaphragm, internal driver resonances, or time misalignments between "ways" in a tweeter/midrange crossover interference band.

 

Chris

 

Chris,

In the case of trying to get the 4 subs to 'play' as one, I thought the best way to go about it was to place the mic at the LP.  One sub at each corner of the room.  Needless to say, there are a lot of objects between the subs and the mic. 

Link to comment
Share on other sites

On to the Main speaker experiment.  Two way mains.  DIY Full Marty bass bin.  Oris 150 with Fostex driver. Mic on axis with speaker, 3 feet away per Chris' instruction and psycho acoustic smoothing to make the graphs usable.

 

I have been crossing the Oris at 300Hz and think I will keep that.  L/R 24db crossover. Now on to EQ.

Left main Drivers.jpg

Link to comment
Share on other sites

Sorry, I misread the horizontal scale.  I usually take full-range sweeps, so I didn't notice.  In general, below 100 Hz, GD values will go as high as 120 ms.  Here is a more representative GD plot for bass frequencies:

 

1920310933_SPUDandJubBassBInGroupDelay.thumb.jpg.abd61adf88d18e0f7a1a48b3e4c30918.jpg

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...