REW - Help Needed with Subwoofer Integration!

[maezun]

I'm trying to time / phase align my Arendal 1961S Subwoofer with my mains (Bowers & Wilkins 704 S3). I am using a standard 2 channel amplifier and the sub is connected via the pre-outs, so I am unable to apply any delay.

I've had a go with the "alignment tool" in REW to try and figure out the best phase setting (so far -100° with Signal Inverted looks best but I'm not sure). I'm also not sure about the optimal crossover. My speakers "officially" have a +/- 3dB of 48Hz but obviously the room response is pretty different and I have a massive mode peak at 55Hz, which is right at the crossover point.

Could somebody have a look at my MDAT and see if they can figure out the best crossover and phase setting for my subwoofer? (It has 'signal invert' option plus phase in 10° steps).

If @Keith_W could have a look, that would be amazing. Thanks!

 

[Keith_W]

Welcome to ASR!

I did have a look, but this is the problem: phase adjustment in a subwoofer is implemented as a variable all-pass filter (APF). These are usually (but not always!) first order APF's with a variable centre frequency. The phase rotation is frequency-dependent, and the maximum phase rotation occurs at the centre frequency. As the centre frequency changes, so does the phase slope.

What this means is that phase rotation is not uniform if you use phase control on a subwoofer. What this also means is that REW's alignment tool is the wrong tool for the job, since there are multiple unknowns. We don't know how your sub has implemented its APF. And the alignment tool adjusts delays, not phase. You can not model the behaviour of your phase control knob in REW. So I am sorry to say, this set of measurements is rather pointless given your goal.

This is what I suggest you do:

1. Set up the mic at the listening position.
2. Set up REW's generator to play pink noise from ALL the speakers.
3. Pull up REW's RTA view.
4. Adjust the phase of your subwoofer until the RTA is as flat as possible.

I do have some comments about your measurements though.

The first thing I did was vector sum your left and right speaker (purple) and compare it to your subwoofer. You can see that the pattern of peaks and dips is the same. I am guessing that you plonked your subwoofer in the middle, between your two main speakers. What this means is that you can not use your subwoofer to fill in the dip at about 80Hz.

To illustrate this point, I summed your L+R speakers with your subwoofer (yellow). I then inverted the polarity of the sub, and then summed that with your L+R speakers (green). You can see that all this does is shift the cancellation somewhere else.

I suggest you find a better position for your subwoofer.

And this is the step response of the L+R speakers (purple) compared to the subwoofer (blue). I know you said you have no means of adjusting the delay. But just so you know, your sub is about 8.6ms delayed compared to the main impulse. I can tell it's a rather small room because the subwoofer impulse looks a bit wavy early on, meaning that it is already being corrupted by reflections.

Your left speaker is located very close to a reflective surface. You can see a very early and very loud reflection in the left speaker, it's actually louder than the main impulse. Either this, or you accidentally swept both speakers by mistake. This is bad, it will smear the sound of your left speaker. Your right speaker does not have the same problem.

 

[maezun]

Thanks for getting back to me, appreciate you taking the time.

That's where I was going wrong then, I thought that the alignment tool would give some sort of indication as to the correct phase to use but obviously not.

The subwoofer is next to the left speaker. The left speaker is in free space (side to side) but is quite close to the rear wall. The right speaker is around 1 foot away from a side wall.

I did some separate measurements (0-200Hz sweeps of left speaker with sub) with multiple phase settings. Is phase 0 the best setting? I was using a 50Hz crossover. Maybe phase 0 is the best setting if, as you say, the delay is only 8.6ms.

I also measured differing crossover points (from 30Hz-90Hz). I think 50Hz was the best but I'm not sure.

 

[maezun]

These tests (taken another time) are non-inverted:

 

[Keith_W]

You are wasting a lot of time with those measurements. What matters is the summation of the subwoofer with the main speakers, not the individual phase of the subwoofer by itself. To gain anything meaningful from those measurements, I would have to take your new sub measurements, take your speaker measurements, and then sum every permutation and examine each sum to find the best result. I don't have time for that.

That RTA method I described is simple and quick. I suggest you do that.

Or if you don't want to do that, an alternative is this (similar in concept). This will take a heck of a lot of time compared to the RTA and its only advantage is that you can examine each graph individually.

1. Set up your mic at the listening position.
2. Set sub at phase 0
3. Sweep both speakers and subwoofers together.
4. Adjust phase of sub to the next increment and repeat steps 1-3.

 

[maezun]

OK I'll give your RTA method suggestion a go. What crossover setting do you think I should use? Or does that not matter?

 

[maezun]

I was going to say, the phase tests were actually done combined with the left speaker but I'm guessing what you mean by 'summation' with the main speakers, is that they are simultaneous?

 

[Keith_W]

OK I'll give your RTA method suggestion a go. What crossover setting do you think I should use? Or does that not matter?

Well that's up to you. I'll show you the considerations:

This is the subwoofer (blue) compared to the sum of your L+R speaker (purple). The question is whether you want to fill that dip at 80Hz or not. Your subwoofer certainly can do this, but not from where it is placed currently. You will need to move your subwoofer to another location and see if it fills the dip. If it does, then low-pass the sub at 90Hz.

If you don't want to fill that 80Hz dip, then it looks as if a 40Hz low-pass would be adequate.

I was going to say, the phase tests were actually done combined with the left speaker but I'm guessing what you mean by 'summation' with the main speakers, is that they are simultaneous?

You have only one sub, so your bass is mono. For mono bass, all bass sources in the room must be considered as a single unit. You can choose to either measure them separately with a time reference and sum them (my preference) or measure them together. The reason I prefer separate measurements is so that I can examine each speaker or sub individually.

 

[maezun]

I meant to mention that I will be equalizing the final response (<300 Hz only and to 0.8dB/8ve target) using PEQ filters (applied in Roon). I will be applying the EQ to each channel separately. Will this affect anything?

 

[sonitus mirus]

I meant to mention that I will be equalizing the final response (<300 Hz only and to 0.8dB/8ve target) using PEQ filters (applied in Roon). I will be applying the EQ to each channel separately. Will this affect anything?

Applying EQ below 300 Hz per channel shouldn’t be a problem in itself, and it can definitely help clean things up once you’re close. The main thing is that EQ works best after placement and basic integration are already cooperating. If the 80 Hz feature is coming from a boundary or cancellation issue, PEQ won’t fully fix it on its own, but it can smooth what’s left once you’ve landed on the best sub position, phase, and crossover. So I’d still treat EQ as the final refinement step rather than something that changes whether that dip is fundamentally addressable.