Fine tuning phase accuracy
In part 2 of my speaker setup series, I discussed setting up the accurate triangle for stereo reproduction, and how to adjust that triangle for correct bass reinforcement and tonal balance too. But there are two more tweaking subtleties that can bring extra rewards – and these are really worth trying out.
The first concerns the phase accuracy and tonality of central image sounds. Vocals of course are usually positioned in the centre of the soundstage, so that means the signal is fairly equally mixed to the left and right channels, and tightly in phase. What we hear at the listening position is that sound from both speakers arriving as intended so that it creates the illusion that the sound is between the speakers. But that illusion depends upon one very important factor - the time, and therefore the distance, of propagation from each speaker, to you. Now sure, at lower-mid frequencies the wavelength is fairly long, middle c is about 1.7m so it would be hard to get a significant discrepancy there. But go up three Cs on the piano and the wavelength is 16cm (f=2.093 kHz). Those three octaves are right in the range of most vocals, string instruments, guitars and so on. Plus there are the myriad of harmonics which stretch way above that. Go up to 5kHz and the wavelength is 6.8cm.
Right, let's consider a vocal, recorded dead centre, with lots of harmonics around 5kHz. If there is a 3.4cm discrepancy between the left and right propagation distance to your head, that 5kHz harmonic will arrive 180 degrees out of phase. Total cancellation in theory. Go back down to that very important 2kHz region and that 3.4 cm discrepancy means a phase error of around 76 degrees. So that smooth, sonorous tone at 2kHz arrives at your head with enough phase shift to generate significant intermodulation harmonics. Add to that the higher tones swinging in and out of phase and what you get is a significant 'rendering distortion' of all central vocal and instrumental tones. And yes, I mean you can hear that – clearly.
And don't think you can sort of compensate for it by moving your head around. For me, my brain clearly knows the errors are there no matter what I do. I can hear when the transmission distance is out by more than about 5mm, and whilst it's a pain in the arse, using a long tape measure and a buddy to help, getting the transmission distance from say a central point on your speaker's baffle, to your sternum when sitting in your listening seat, down to under 5mm, can be a revelation. [Remember to be very careful when using a steel tape around the front of your speakers – there might be a strong magnetic field there!]
And this then brings me on to the second fine tweak, that is an extension of the first one. This one is about speaker toe-in, and the on and off axis frequency response and on and off axis phase accuracy of your speakers.
Toe-in adjustment is often carried out at the point when we have our listening triangle set up to the level of my part 2 blog, ie fairly well adjusted but not to the accuracy we're talking about here. The toe-in discussion revolves around on and off axis frequency response and we think it is correct to avoid pointing the speakers directly at the listening position because they will be too bright, and we need to reduce the toe-in to reduce the brightness. But we were not including the propagation distance element in our setup at all and we were always getting an over bright presentation because of the phase discrepancy. Add to it, that in my experience a lot of speakers produce their most accurate phase response bang on axis and we can perhaps turn setup convention on its head here.
So, do the fine propagation distance setup as accurately as you can – get the distance matched to within 5mm. Then, try increasing the toe-in to point the speakers directly at the listening position. Try adjusting, and listening, and maybe move the speakers back and forward, and laterally as well, always keeping that 5mm accuracy intact. Bear in mind now that you are looking to resolve phase more accurately and you're not just thinking about frequency response – but juggling the two. You might not end up with firing the speakers directly at your listening seat, but some people do.
There are always tradeoffs of course, but there have been many occasions where these fine adjustments have been the absolute make or break on the final presentation capability of a system. Often, when dialled-in like this, the image springs out sideways with all width limitations suddenly seemingly gone, central images gain a new level of smoothness and richness, and depth is markedly improved too. This is brought about by a vast improvement in phase rendering capability and once achieved, is something you are almost certainly not going to want to throw away again.
