TOPIC:

Hi Team,

I’m going to try dual subs with my SHD. One sub will be just to the left of the left main speaker the other will be just to the right of the right.

I’m leaning towards setting them up in stereo but with some bleed from the opposite channel.

If I do this, I’m thinking I’ll get the best of both traditional mono and stereo setups. I’ll reduce the effect of any peaks and troughs by having all bass coming from two location and I’ll support the stereo imaging and avoid pulling the leading edge of bass notes into the centre.

My question: on the matrix for the each sub I’ll have 0db set for the main channel, let’s say Left for this example. When I mix in some of the right channel, have you any thoughts on how many dBs down I should set it to pull the image into the centre by about 50cm to line up with the main. Vice versa obviously for the other side. -12db? -18db? Is there a rule of thumb formula?

Thanks
Alex

Depending on what frequency range the subs go up to* it is unlikely you can tell where the bass is coming from, we're just not sensitive to imaging localisation at lower frequencies. That's pretty much what allows you to put subs in any mad places, corners, behind you, anywhere really.
So the only thing that tends to matter is time alignment of subs to mains.
But don't let me dissuade you from an interesting experiment

*The exception would be if your mains are tiny and only go down to 200Hz or so and your subs are really acting like woofers.

I would personally always place subs in corners. They are then a minimum distance from 3 walls in all axes which maximises those SBIR frequences, and the maximum distance from the 3 opposite walls in all 3 axes, which minimises those SBIR frequencies. The overall effect is the SBIR frequencies are moved both above and below critical frequencies and don't all collide in the critical frequency range. Your mains will generally always suffer this SBIR collision causing very deep notches which Dirac *cannot* correct. This is one of the few problems that Dirac can't correct so it is generally wise not to make your subs suffer the same uncorrectable problem the mains have.

Hi,

I was reading on KenRockwell about dual subs. He wrote: “...A 12 db/octave 80 Hz crossover attenuates 160 Hz by only 12 dB, and attenuates 320 Hz, which is a very localizable E above middle C, by only 24 dB. 24 dB is equivalent to turning your volume only halfway down, and you can hear that very well.

In my own tests on myself, in order not to be able to localize a subwoofer, I can cross it over at no more than 60 Hz at 18 dB/octave. I can hear exactly where it is when crossed over at 80 Hz at 18 dB/octave.

In order truly not to be able to hear the location of a non-distorting subwoofer, cross-over at 60 Hz or below or 18 dB/octave or more.

At 80 Hz with a steeper slope of 24 dB/octave, a THX standard, ought to prevent localization....”

Article here: www.kenrockwell.com/audio/stereo-subwoofers.htm

I’ll be using 24dB/Octave but I do normally have my crossover set high because I’m using valve monoblocks so best to leave the true bass to solid state amps in my subs.

Wrt position, my subs will be in the corners with the main speakers just inboard of them. My listening room isn’t very big.

I think I will start with true stereo and after Dirac, measure with REW. If it’s successfully dealt with peaks and troughs I won’t worry about mixing in some of the opposite channel. If I’ve still got nulls then it’s mixing time.

With the mains being just inboard of the subs and ~2m apart, how much do you think I should cut the opposite sub channel? Do you think -12dB is too much or too little? The aim would be to ‘move’ the subs’s image right on top of its accompanying main speaker.

OK, with relatively high crossover frequencies it may be worth doing, my subs rarely get set above 50Hz.

One thing to watch out for is that as you add bleed from the opposite sub the overall bass level will increase. It will increase slowly to begin with but as the bleed gets louder you may be suckered in to thinking more bass is better, regardless of where the virtual image is.
-12dB shouldn't affect things too badly but if you need to go much higher watch out for that. At -6dB you would actually have to reduce the direct sub levels to -6dB as well to keep the correct bass level, and at that extreme point you obtain mono bass. That's because the difference in levels is 0dB, you have half left and half right coming from both subs so pure mono with unity left and right overall giving the proper bass levels.

This adjustment of direct sub level is only required if you adjust bleed after calibrating Dirac, which is valid. If you can't be bothered to correct the direct sub level manually run Dirac again after setting bleed and it will then also correct for excess bass due to bleed.

So somewhere around -12dB but much less than -6db is probably your range to explore.

It doesn’t make any difference but the site wouldn’t let me edit my previous post. I wanted to make it clearer that my speakers are only about 2m apart with subs just outboard of them and in the corners.

Yes I would plan to set the bleed and then leave it to Dirac to set levels.

My speaker delays however will only be correct for Dirac measurement position 1. Will Dirac struggle with the other measured positions given the same signal will arrive at slightly different times or does it not matter for bass?

Sound travels around 1 foot (30cm) per millisecond in air (at typical temperatures) so in theory the small lateral movement of the apparent sub position is unlikely to have a significant effect at the listening position at the long wavelengths of low frequencies due to trigonometry. However in practice I generally find very accurate time alignment of subs works wonders at integrating them optimally.

I guess what I'm saying is that I would perhaps set up another SHD preset for this. Adjust the bleed, recalculate time delays, recalibrate with Dirac, Iterate until happy with result. In principle when you have the correct bleed all the speakers should have the same delay because they are then spatially coincident. However the trigonometry will cause the delay to be somewhat insensitive to postion I fear.

If you do it in a different preset you will be able to A/B unbled and bled subs with all time delays and Dirac calibration optimised for each situation. That's always a great way to get confidence that a huge change is actually working well.

And yes, I did get that your speakers were around 2m apart with the subs just outboard from them in the corners from your previous post. All my handwaving was based on that. I did try to spell some maths out but had to remove it from the post because I have ear infections from hell at the moment and I can't seem to be confident about any maths.

Hi entripy?
Double thanks if you’re providing your advice whilst poorly. I’d definitely be interested in understanding the calculation when your head is in the right place. I’ll digest the feedback you’ve provided and save any further questions until the end of the week when hopefully you’ll feel better.
Take it easy.
Alex

Oh believe me, it's far safer for me to comment here than climb ladders and drill holes to fit more acoustic diffusers to celings

I'm not even convinced the correct maths will work in such a small room. The maths generally assumes point sources in free space. Given speakers radiate omnidirectionally below the baffle diffraction limit as far as the subs go they will probably see the walls as almost perfect mirrors. And the floor and ceiling. Just try to imagine sitting in a room with surfaces totally covered with mirrors and even knowing which of the multiple infinitely reflecting images is the actual object you are trying to position*

So I would still suggest starting bleed at -12dB, going up a bit and down a bit and trying to hear the effect.

*there is actually a sub positioning technique which relies on the image reflections to create a 2D infinite array of equally spaced sub images with the object of creating a near perfect plane wave down the length of the room. You then place similar subs at the other end of the room and get them to drive an antiphase signal with an appropriate time delay such that they perfectly cancel out the reflection from the rear walls. I'd love to try something mad like that but it requires a perfectly rectangular room. My main room is more like the top 2/3 of a wonky octagonal tube so I am, perhaps fortunately, unable to test the technique.