Time alignment question

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

Crossover will be 'phase correct' (Linkwitz-Riley) at xover point. But there's not much point in sorting the filters phase-wise if the drivers aren't in phase. Might just as well stick with Butterworth or Bessel for all the good it will do.

Drivers that are time aligned in the vertical plane must be phase aligned (though the reverse isn't necessarily true). This of course assumes the drivers themselves have a good 'phase behaviour' - if they don't then the whole thing gets mangled and becomes a waste of time anyway.

Intuition tells me phase correctness is a good thing but to get it right I need to know where the centres are. Xovers are going to be active so some tweaking is allowable - however would like to get it as right by theoretical means as possible as my test & measurement equipment (and environment) is limited.

I suppose all these variables and what if's are what makes good crossover design a black art and why diy speakers with identical drivers can sound so different. Wouldn't be fun otherwise would it ????

I'm beginning to think you're right. Ultimately the only way to really know for sure is to measure it.

Does any one have any recommendations for cheap(ish) test & measurement packages for this kind of thing. I would assume these days it would all be PC based. Please excuse the ignorance but since my last foray into diy audio was in the days of the IBM XT (remember those things ?!!) I feel somewhat out of touch.


Vandersteen uses a narrow pulse response measurement and compares the FFT to the FFT of the original signal. The problem you have here is that you need to get a calibrated measuring mike. It would be an understatement if i told you they are not cheap (Bruel&Kjaer makes a whole slew of them http://www.bksv.com/bksv/). The next hurdle is to know what to adjust based on the FFT results. You will need to do a lot of analysis. I am not even certain if the FFT analyzers give you which frequency components represent sin or cosin phases. Plus a single pulse will have contigous spectrum. In reality you will perform the measurement with a signal with a fairly low repetition rate, say 20 cycles. A narrow pulse say 100 to 250 usec. If you are brave enough you can experiment visually by looking at the response on a scope and and just shift the drives beck and forth for the best looking impulse response. But it is not as scientific as the FFT based measurements may be. (Please do not take me for an expert, i just tinkered with this stuff for a little while.)

The PC based digital scopes and spectrum analyzers are not that terrible expensive nowadays.

Not sure if the project is still listed (the site is undergoing a redesign, but its a fairly new article), but http://www.speakerbuilder.net had a project in which you could build your own high-quality test mic for significantly less than you could purchase one for. He also showed how the measurements stacked up to another popular commercial version, and the results were most impressive, especially cost-considering.

Phase response?

Does anybody have a phase response diagram of the panasonic EC? I know for a fact that the 1/4" B&Ks have very linear phase response from about 25Hz to well beyond 20 KHz. If you know the phase response of the mike you can correct for it in software.

Okay, read the Linkwitz articles. Very interesting.
I can see this is not as straight forward as it looks.

However, there might be a 'cheat'.

Been looking at KEF drivers. The Uni-Q range are dual concentric full range drivers. This neatly sidesteps the seperation and time alignement problems (possibly).

Anybody have any experience with these drivers. Are they any good. How do they stack up (tonally) against a decent Scanspeak / Vifa driver set ??
Joined 2000
Paid Member
Dual concentric have their own problems. If they are not sturdy in construction, vibrations could smear the imaging. Tweeters are tide to the pole piece if I'm not mistaken. All freq will show up on time, however.

There are aproximations for time coherance, you'll just have to look around on the net. If you use a higher order crossover for both high and low frequency, you get better dispersion of upper and lower freq.

There is a good book by Radio Shack, it's the newer book. It's very mathimatical. It covers the use of high order xovers.

ckeck this out.


I have been using the KEF Q60 for about ten years. This has the original Uni-Q driver. Imaging and focus are superb, but the treble is a little bright though this can be aleviated by a small modification to the standard crossover. Later generations of the Uni-Q have an improved tweeter with a smoother response. So far as I am aware, KEF have stopped supplying Uni-Q drivers for DIY construction, though last time I checked Falcon Acoustics still had a few in stock. One alternative is to use the units designed for in-wall or in-ceiling mounting. Unfortunately, I have not been able to find any T-S data for these. Another option would be to buy the base model commercial speaker and to use the drive unit in a better cabinet with active filters. You would need to measure the T-S parameters to do this properly.

Other manufacturers have followed KEFs lead and now produce a bass/mid-range unit with a co-axial tweeter at the acoustic centre. Seas lists various models on their website and I am sure I have seen others elsewhere (but unfortunately can't remember where at the moment).

I have done some work on active filters for the original Uni-Q, though I have not actually built them yet, and I have the schematics for the original passive crossovers. If you decide to persue the Uni-Q route, let me have your email address and I'll send you details.

I work with a number of horn loaded systems in movie theatres. Here we always try to align the voice coils if possible. It is only really impossible to do this in 3-way systems though so this has to be done electronically. Lucasfilm THX took care of the time align problem in their THX systems by electronic methods, although there are also VERY specific positions for the horns in each instrallation.
Mark Gulbrandsen
Salt LAke City, Utah

Thanks for the info on the UniQ drivers.
I checked the Wilmslow audio site. The only UniQ's currently for sale are the Custom Install items. The KEF website describing these as 'general purpose - suitable for 'airport lounges' type of thing'. The specs don't look too hot and the plastic driver surrounds look a bit disappointing too. Unless I can get my hands on a proper set of UniQ's i'll probably give it a miss - otherwise i'll give you a shout for help.

Wilmslow also still sell T27, B110 (actually there's a thought i've always fancied a pair of LS3/5s) & B200 drivers though I suspect they are as stocked - once they're gone thats it. KEF definately seem to have deserted the diy market. Its a shame considering that a few years ago they were THE diy suppliers in the UK.

It's unfortunate that KEF has quit selling drivers to the public. I own a pair of LS3/5As, and have on hand a pair of T-27s and B-110s that I had always intended to use to build another pair, just for fun. My intention was to use some B-139s for (sub)woofers since the B-110 doesn't do well below about 80-100Hz. I've got scads of B-139s on hand--I just need to follow through.
Not to mention the B-139 always did a spendid job in a transmission line.


You lucky sod :)

KEF did a 3way kit using the t27, b110, b139 and an ABR - can't remember what it was called. A friend of my fathers built a pair - I must have been about 13 when I heard them & I remember thinking 'wow' (first taste of REAL hifi ) they were the dogs doo-dahs.

If you're thinking of building LS3/5 have you checked out the Unofficial LS3-5a website:


There's a fair ammount of constructional stuff. Plus (if you can get your hands on another couple of B110's) an AB1 project - would go nicely with your LS3's.


PS: I think Rogers is still making the ls3/5 and ab1 for the Japanese market.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.