Thanks. What is the difference between the black solid line and the one that is dashed? One I assume is the estimated SPL given the better phase alignment but not sure what the other one represents?
-
Welcome to ASR. 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!
Time align with Wiim amp using Rew
- Thread starter Arianoxx
- Start date
-
- Tags
- rew time alignment wiim
- Thread Starter
- #23
So, I’ve just finished redoing everything for the fourth (or fifth, I've lost count) time, and I’m promising myself this is going to be the last one.
I followed every step carefully, the chirp is definitely there, and while the delta-T is still far from perfect, it's beyond my current understanding. Maybe it's something with my MacBook (M1 Pro), my mic (IMM-6C), or just something I’m not aware of.
Anyway, this time I measured each mains and the sub with HPF and LPF applied for a few different crossover points (and yes, it was a pain). But now, when I use the alignment tool, the result matches exactly what I see later in the measurements.
I also tried the "align phase at cursor" feature and even aligned things manually, and what I'm sharing is the best result I've managed to get so far.
If anyone wants to take a look:
I followed every step carefully, the chirp is definitely there, and while the delta-T is still far from perfect, it's beyond my current understanding. Maybe it's something with my MacBook (M1 Pro), my mic (IMM-6C), or just something I’m not aware of.
Anyway, this time I measured each mains and the sub with HPF and LPF applied for a few different crossover points (and yes, it was a pain). But now, when I use the alignment tool, the result matches exactly what I see later in the measurements.
I also tried the "align phase at cursor" feature and even aligned things manually, and what I'm sharing is the best result I've managed to get so far.
If anyone wants to take a look:
- The "reference" file includes the timing measurements (with different crossover settings)
- The "20-04-2025" file includes some of the results
alankila
Member
- Joined
- Mar 28, 2025
- Messages
- 81
- Likes
- 64
OK, but here is what I have. It is basically the same thing as before. Your mains are rolling off under 50 Hz, so phase alignment will be lost below 50 Hz. Your L main is showing unevenness in response after 80 Hz, and sub also rolls off after 100 Hz, so we will lose the phase alignment for these two reasons above there. This leaves us something like 50-80 to work with, and I chopped a bit more around both because of the severe phase divergences at both extremes.
Phase is usually not correct if magnitude is not showing the same kind of changes due to similar acoustics environment, as that is just a part of how minimum phase systems work, so that is why it is critical to identify part of the response that seems to be comparable and focus on that. Thus, I manually optimized the range between approximately 55 and 75 Hz for both gain and phase. In the top magnitude trace, I used +2.2 dB as the gain for mains to make the blue and red curve overlap on the region. I previously recall saying sub should be 1.7 dB louder -- well, that was then. This trace suggests +2.2 dB more on mains, whatever the reason.
I then adjusted the delay to maximize the overlap of the phase curves in the same region. I didn't use the filtered versions but just the full bandwidth versions because I think that you should select between 60/70 Hz and I think one delay roughly optimizes for both regardless of the crossover setting as they should experience similar phase adjustment due to the crossover used. I previously mentioned this as well, that the crossover is 2nd order and it should flip phase 180 degrees, but I think Wiim is going to flip it for you, and you don't have to worry about that. The phase traces show good overlap between about 55-75 Hz, if I use 4.5 ms delay on the mains. I continue to maintain that this is roughly the correct delay for your system. The resulting system's phase angle is fairly flat between 30 and 80 Hz, so it can't be far off like this.
In these measurements, I could not use the automatic tools. They optimize across a wider section of the response and because it diverges so hard near 80 Hz, I guess it throws it off. It would be better if alignment was done on extremely narrow region like 10 Hz at most, but alas, that's not how the tool works. So manual adjustment was needed, though I already knew the figures and they work this time just as well as they did last time.
View attachment 445590
Phase is usually not correct if magnitude is not showing the same kind of changes due to similar acoustics environment, as that is just a part of how minimum phase systems work, so that is why it is critical to identify part of the response that seems to be comparable and focus on that. Thus, I manually optimized the range between approximately 55 and 75 Hz for both gain and phase. In the top magnitude trace, I used +2.2 dB as the gain for mains to make the blue and red curve overlap on the region. I previously recall saying sub should be 1.7 dB louder -- well, that was then. This trace suggests +2.2 dB more on mains, whatever the reason.
I then adjusted the delay to maximize the overlap of the phase curves in the same region. I didn't use the filtered versions but just the full bandwidth versions because I think that you should select between 60/70 Hz and I think one delay roughly optimizes for both regardless of the crossover setting as they should experience similar phase adjustment due to the crossover used. I previously mentioned this as well, that the crossover is 2nd order and it should flip phase 180 degrees, but I think Wiim is going to flip it for you, and you don't have to worry about that. The phase traces show good overlap between about 55-75 Hz, if I use 4.5 ms delay on the mains. I continue to maintain that this is roughly the correct delay for your system. The resulting system's phase angle is fairly flat between 30 and 80 Hz, so it can't be far off like this.
In these measurements, I could not use the automatic tools. They optimize across a wider section of the response and because it diverges so hard near 80 Hz, I guess it throws it off. It would be better if alignment was done on extremely narrow region like 10 Hz at most, but alas, that's not how the tool works. So manual adjustment was needed, though I already knew the figures and they work this time just as well as they did last time.
View attachment 445590
alankila
Member
- Joined
- Mar 28, 2025
- Messages
- 81
- Likes
- 64
Hold the phone. I noticed that if you invert the phase, the alignment in phase can be made much better. The time delay is now on the sub side, due to negative delay on mains, so rather than delaying the mains, sub should be delayed instead. I really should have checked this option before, and I'm rather sorry that I didn't. Clearly this is a better phase alignment than any attempt before, staying basically almost exactly correct all the way from 30 to 90 Hz. I used the other channel to compare because it seems to have less of that 80 Hz acoustic problem and it continued to run really nicely up to that cancellation and then whatever is going on > 100 Hz. This is way better.
Well, I don't know what to say, except that I recommend this setup from now on that I happened to use the phase switch and noticed that it was much better this way. Sorry about the noise from before, I really should have checked this. I am actually a bit surprised that sub needs to be further delayed, as it is my experience that subs are usually late in their sound output to begin with. Maybe the mains are DSP speakers? Or is the Wiim playing tricks and has some assumed delay built-in that serves as 0 reference value? I don't know.
Well, I don't know what to say, except that I recommend this setup from now on that I happened to use the phase switch and noticed that it was much better this way. Sorry about the noise from before, I really should have checked this. I am actually a bit surprised that sub needs to be further delayed, as it is my experience that subs are usually late in their sound output to begin with. Maybe the mains are DSP speakers? Or is the Wiim playing tricks and has some assumed delay built-in that serves as 0 reference value? I don't know.
Keith_W
Major Contributor
@alankila you can not compare a sub measurement with a mains measurement if they are not both anchored properly in time. Yes you can play with the curves and come up with something like "if you shift your sub by 3.5ms it aligns the phase" but it is meaningless if the time reference of the mains and subs are wrong in the first place. How do you know it's not 3.5ms + (unknown delay) caused by inconsistent delay measurements?
@Arianoxx In your opinion, do you think you have achieved consistency with your timing measurements?
Sweeping L speaker with R timing reference should give you 0ms or close to it. Look at what you are measuring: -740mm, 1.2mm, 247mm, -5.5mm. I have given the reasons for REW getting it so wrong. Another reason is because low frequencies are time stretched, so it is difficult for an algorithm to automatically determine the delays. If you were using a tweeter impulse, the time measurements would be far more consistent.
Your thread made me ask John Mulcahy whether there is a way to manually determine the delay by looking at the tweeter timing chirp impulse, but unfortunately the answer is "no". It would be very nice if he had designed REW to have the timing chirp play at exactly 500ms before the main sweep, that way we can compare the impulse of the DUT to the timing chirp, do some subtraction, and you're done.
What you need to do is to be able to sweep the DUT together with a tweeter reference with a known delay. With your system, this would require creation of a special crossover with tweeter delays built-in. The measurement looks something like this:
As you can see, the DUT and tweeter are part of the same impulse. But because the tweeter is delayed, the impulse of the DUT and tweeter are separated in time. This allows you to read the impulse of the DUT and compare it to the tweeter. You can have absolute confidence that the reading is correct, since both are anchored in time to the same impulse.
Is it possible to create a special filter like that in REW? Yes, but it is going to be very complicated and you will need third party software. I feel a bit sorry for you, so i'll talk you through it. Send me a PM with your phone number, and we can set up a WhatsApp chat and we'll go through it together.
@Arianoxx In your opinion, do you think you have achieved consistency with your timing measurements?
Sweeping L speaker with R timing reference should give you 0ms or close to it. Look at what you are measuring: -740mm, 1.2mm, 247mm, -5.5mm. I have given the reasons for REW getting it so wrong. Another reason is because low frequencies are time stretched, so it is difficult for an algorithm to automatically determine the delays. If you were using a tweeter impulse, the time measurements would be far more consistent.
Your thread made me ask John Mulcahy whether there is a way to manually determine the delay by looking at the tweeter timing chirp impulse, but unfortunately the answer is "no". It would be very nice if he had designed REW to have the timing chirp play at exactly 500ms before the main sweep, that way we can compare the impulse of the DUT to the timing chirp, do some subtraction, and you're done.
What you need to do is to be able to sweep the DUT together with a tweeter reference with a known delay. With your system, this would require creation of a special crossover with tweeter delays built-in. The measurement looks something like this:
As you can see, the DUT and tweeter are part of the same impulse. But because the tweeter is delayed, the impulse of the DUT and tweeter are separated in time. This allows you to read the impulse of the DUT and compare it to the tweeter. You can have absolute confidence that the reading is correct, since both are anchored in time to the same impulse.
Is it possible to create a special filter like that in REW? Yes, but it is going to be very complicated and you will need third party software. I feel a bit sorry for you, so i'll talk you through it. Send me a PM with your phone number, and we can set up a WhatsApp chat and we'll go through it together.
alankila
Member
- Joined
- Mar 28, 2025
- Messages
- 81
- Likes
- 64
@Keith_W I just assumed that the software works, I guess. When I use acoustical timing reference, I get reference delay around 0.05 ms or less for the main channels and about 2.4 ms for sub. Subwoofer impulse response is a big series of swings and I don't really understand those things intuitively, so I can't say what is going in there, except that the ref seems to be centered at the biggest lobe. However, it is not the first lobe but the second one in the impulse and it's a negative lobe rather a positive one. I don't really know why that happens.
I didn't look into what timing values @Arianoxx was getting. I now see that these values are all over the place for some reason. I think I did accidentally establish that the sub needs to be inverted for phase match, even if the delay timing value itself is possibly wrong.
I didn't look into what timing values @Arianoxx was getting. I now see that these values are all over the place for some reason. I think I did accidentally establish that the sub needs to be inverted for phase match, even if the delay timing value itself is possibly wrong.
Keith_W
Major Contributor
In case there are others who may want to time align subs with Wiim and REW, this is how we solved the problem.
First I realised that he has the ability to set delays in his Wiim. So I asked him to delay the main speakers by 50ms and take a sweep of the sub with the main speakers.
If you study the impulse response, you can see the 50ms delay that we imposed has separated the subwoofer from the main impulse. It is difficult to see since it is low amplitude and time stretched. This is why automatic time alignment readings sometimes fail. The first thing we need to do is make the impulse peak of the main speakers our time reference. Right click on the graph and "Set t=0 at IR peak".
We need to zoom in a special way so that we can see the impulse clearly. I drew a rectangular box (Ctrl-right mouse click and drag) that is wider than it is tall. Do it a few times and you can clearly see the subwoofer impulse.
Zoom in a bit and place the marker at the start of the impulse, then read the time. Here it is -47.7ms.
Since the mains was delayed by 50ms, we expect the subwoofer to be at -50ms if it was time aligned. So the subwoofer is 2.3ms delayed. The correction is to delay the main speakers by 2.3ms.
Once this was determined, I helped him simulate the effect of the crossover and used the alignment tool to confirm the findings, fine-tune the volume, and so on.
This is the step response before (yellow) and after (green).
Frequency response before (yellow) and after (green). Turns out @alankila was right, if the polarity is inverted, the summation is much better. I think the bass is still too hot, but OP said he likes hot bass and wanted to listen a bit, so we stopped there. I would personally throw in some PEQ's but that's up to him.
First I realised that he has the ability to set delays in his Wiim. So I asked him to delay the main speakers by 50ms and take a sweep of the sub with the main speakers.
If you study the impulse response, you can see the 50ms delay that we imposed has separated the subwoofer from the main impulse. It is difficult to see since it is low amplitude and time stretched. This is why automatic time alignment readings sometimes fail. The first thing we need to do is make the impulse peak of the main speakers our time reference. Right click on the graph and "Set t=0 at IR peak".
We need to zoom in a special way so that we can see the impulse clearly. I drew a rectangular box (Ctrl-right mouse click and drag) that is wider than it is tall. Do it a few times and you can clearly see the subwoofer impulse.
Zoom in a bit and place the marker at the start of the impulse, then read the time. Here it is -47.7ms.
Since the mains was delayed by 50ms, we expect the subwoofer to be at -50ms if it was time aligned. So the subwoofer is 2.3ms delayed. The correction is to delay the main speakers by 2.3ms.
Once this was determined, I helped him simulate the effect of the crossover and used the alignment tool to confirm the findings, fine-tune the volume, and so on.
This is the step response before (yellow) and after (green).
Frequency response before (yellow) and after (green). Turns out @alankila was right, if the polarity is inverted, the summation is much better. I think the bass is still too hot, but OP said he likes hot bass and wanted to listen a bit, so we stopped there. I would personally throw in some PEQ's but that's up to him.
alankila
Member
- Joined
- Mar 28, 2025
- Messages
- 81
- Likes
- 64
I wonder about setting the timing at the start of the sub impulse. When I use acoustical timing ref, the reference point is at about 14 ms, it seems:
And the actual alignment using the time align tool with the main speaker impulse being ref'd on the big peak very near 0 ms:
Showing that if sub played about 5 ms earlier -- or mains 5 ms later, the phase overlap is good between 40-100 Hz. So I don't really understand how these impulses work and what the ref times mean...
And the actual alignment using the time align tool with the main speaker impulse being ref'd on the big peak very near 0 ms:
Showing that if sub played about 5 ms earlier -- or mains 5 ms later, the phase overlap is good between 40-100 Hz. So I don't really understand how these impulses work and what the ref times mean...
- Thread Starter
- #30
Hey guys!
I just found another used sub like mine (B&W ASW610) for the same price I paid for mine a year ago, 300€.
I was wondering: would adding a second sub make a noticeable difference, enough to justify spending another 300€?
Consider that WiiM Amp only has one sub out, so I’d have to use a Y-splitter to run both subs. That means crossover and delay settings would apply to both together, would that be a problem?
Just a reminder: the positioning is far from ideal, I can only place them inside an IKEA shelving unit.
Let me know what you think!
I just found another used sub like mine (B&W ASW610) for the same price I paid for mine a year ago, 300€.
I was wondering: would adding a second sub make a noticeable difference, enough to justify spending another 300€?
Consider that WiiM Amp only has one sub out, so I’d have to use a Y-splitter to run both subs. That means crossover and delay settings would apply to both together, would that be a problem?
Just a reminder: the positioning is far from ideal, I can only place them inside an IKEA shelving unit.
Let me know what you think!
Keith_W
Major Contributor
Ciao! Adding another sub might help in fixing these problem areas:
However, you would definitely need independent sub control. i.e. NO Y splitter. And the limitation of only being able to put it inside your shelving unit likely means you won't be able to position your second sub to fill in those dips. I'm not saying that it won't, i'm saying that nobody knows until there's a subwoofer in there. If you want a preview, you could put your existing subwoofer in there and take a measurement. Then compare that measurement with this graph (the one we worked on together) and see whether it will fill in the holes.
Right now the measurements suggest you have enough bass. The problem with the bass in your room is not that you don't have enough, it is that your bass is not even. Fixing that means getting another sub, something like a MiniDSP, and being able to put your second sub where it needs to be.
Otherwise, my suggestion is to keep the 300 Euro in your pocket.
alankila
Member
- Joined
- Mar 28, 2025
- Messages
- 81
- Likes
- 64
I am not convinced those are that much of problem areas. Consider that system's audio output level seems to be about 73 dB. Bass tends to trend upwards due to room gain type of effects and that kind of rise seems to start around 2 kHz, but it's probably relatively modest rise. I am mentalling drawing a gradual rise from about 73 dB level at 2 kHz to something like 77 dB near 20 Hz.
Thus, I'd rather invert your point of view and consider that room modes are causing the peaking around 25 Hz, 55 Hz, 150 Hz, 200 Hz, etc, whereas the regions you circled in red are likely showing a response that isn't strongly assisted by mode, and thus these could be the bass at proper level. There is probably some narrow cancellation also going on in some frequencies, but the smoothing of 1/6 hides that kind of detail from us. My guess is that 90 Hz is strongly cancelled, but I can't say if anything else is.
Keith_W
Major Contributor
Sorry. In case it wasn't clear, the yellow line is before DSP, and the green line is after. The dips remain after removing the peaks.
alankila
Member
- Joined
- Mar 28, 2025
- Messages
- 81
- Likes
- 64
Oh, I see. I didn't know what I was looking at -- I assumed these were not corrected responses mostly because that 25 Hz was so high. I typically allow REW to design filters that bring the peaks down. I most commonly do flat in REW, and then shape actual tonality with couple of very wide peaking band filters on top. In my case, green is uncorrected and orange is corrected. The correction is probably not the best you've ever seen, but I don't chase perfection in this sort of thing myself, either. I try to stay within about 3 dB of the target line, except in those regions where response is cancelled and which probably require placement changes or something such to solve.
Still, I'd imagine that REW could do more precise equalization optimizing for the 10 filters that Wiim has.
