December 3, 2010

Time alignment overview

Time alignment is often either ignored or done incorrectly. There are four critical points in this discussion. 

1. Time alignment is necessary

If you are serious about accurate reproduction, time alignment is critical. 

Why you should time align >


2. Time alignment is often done incorrectly

Stepped baffle arrangements will not necessarily get you there and they also introduce other problems.

3. Time alignment isn't difficult

Affordable digital active crossovers make time alignment easy to achieve.

4. Time alignment can only be done correctly with measurements

Measurements are essential and they are not difficult or expensive to perform. If you are smart enough to operate a PC and find this blog, you can learn how to take the measurements for time alignment. No excuses!

A common mistake

It's very common to assume that physical alignment of drivers will achieve time alignment. You will typically find instructions to measure the offset of tweeter dome to midrange driver dust cap. You can't calculate the necessary adjustment with a tape measure. The reason is simple - the crossover itself will cause misalignment as well that needs compensation. Even if you do align the acoustic centre of each driver, you won't have allowed for the impact of the crossover. 

So you can't judge time alignment by the baffle design. A speaker with a stepped or sloped baffle may or may not be time aligned. It depends on both the physical arrangement and phase shift. 

Are sloped or stepped baffles a bad idea?


A skilled designer can manipulate the baffle and along with a good crossover design achieve time alignment. The key is in getting the crossover and baffle to work together. One clever design is the Danley Unity horn, and it's later update the Synergy horn. In this design, a compression driver and cone midrange drivers share a common horn. The physical arrangement and the passive crossover together achieve alignment.




In this version, woofers are also included.

The easiest way to time align


Use a digital active crossover. Measure it and adjust the delays until you achieve time alignment. You will need a measurement setup to design your crossover, and the extra effort to time align is minimal. If you are using Behringer DCX, it's dead easy and takes minutes. If you are using another system, then you will have to do it manually, paying attention to phase as the crossover points. 


Subjective impressions

Don't expect your jaw to drop when you experience a time aligned system. It may not blow you away. The imaging will simply be more accurate, but the recording may also undermine evaluation. In some cases you might not like it as much. Sometimes time misalignment may create different imaging that you may in some cases prefer. The key point is that time alignment will present the most accurate imaging you can achieve, all other things being equal. 


A few months ago I experienced a system with a digital active crossover in which time alignment was treated as very important. The speakers were the "Franks" designed by Terry Jones in Bathurst Australia. If you get the chance to meet Terry and experience his system, it's likely it will challenge some of your beliefs. Many have described it as the best they have heard. Terry uses DEQX for the crossover and mostly PHL drivers. To date I have not heard a system that can match the imaging. Terry will tell you emphatically that time alignment is critical and his system proves his point. 
In my own system I have found that dramatic changes in the sound stage and imaging can happen as a result of time alignment (or misalignment). 

DEQX & Legend Speakers

It should be noted that there are levels of time alignment. Any digital active crossover can time align at the crossover points, but a more sophisticated system like DEQX can also correct phase shift within the bandwidth of each driver. 


Legend Speakers use DEQX in their top of the range Tikandi speaker.

As Dr Rod mentions in the comment below, DEQX can also correct existing passive speakers. This was demonstrated on another of his speakers, which was reviewed by Australian Hifi and is available online at AV hub.


Time alignment series

Why you should time align >
Is it really necessary?

Time alignment overview >
A bird's eye view of various ways that are used to time align speakers. Does physical offset really work?

Time alignment with Behringer DCX >
This is about as easy as it gets. You need a mic and a few minutes to run the auto routine.

Digital time alignment >
My preferred method. No need to build bizarre baffles that can introduce their own issues.

Subwoofer phase alignment
Yes, even subwoofers benefit, but not for the same reason

3 comments:

  1. As you mention, Terry uses the DEQX and not traditional DSP crossovers like the DCX and miniDSP, which unlike DEQX, do not correct frequency response in detail (about 4,000 points automatically from measurements) nor can they correct FREQUENCY RELATED TIME Delays WITHIN each driver (group-delay errors) as the DEQX does. Instead, systems like DCX only provide time alignment BETWEEN two drivers, which is a relatively small part of what Terry's DEQX DSP system does as far as I am aware.
    Another advantage Terry's DEQX based system has over traditional DSP crossovers is that DEQX uniquely provides steep LINEAR-PHASE crossovers from about 48dB/octave (8 poles). By comparison, normal DSP or analogue passive and active crossovers become NON-linear as soon as their slope increases above 6dB/octave (single pole).
    This may explain why "To date I have not heard a system that can match the imaging"... time alignment between drivers is certainly essential, but far from Terry whole story IMHO.

    ReplyDelete
  2. As you correctly point out Paul, crossover filters generally introduce phase changes – which is just another way of saying timing errors. For linear systems (such as analog filters) a change of amplitude (which is what crossover are doing for individual drivers at the crossover frequency) ALWAYS results in a change in phase i.e. a time shift (there is a fundamental mathematical theorem on this).

    At Linn, my first project was to upgrade the Sara crossover and I was asked to include three LC circuits in series with the treble filter that produced a total phase shift equivalent to about 0.15 ms of time delay – ie effectively moved the tweeter back about 5 cm to time align it with the bass/mid unit. It certainly improved the time alignment but I was not convinced that the signal degradation by the extra components was justified and so did not include them in subsequent circuits. I was also not convinced that the gains of timing correction of stepped baffles or sunken drivers were offset by the deterioration of sound quality due to diffraction effects etc from the resultant edges.

    As you also correctly state this time alignment is done much more easily & precisely in the digital domain – without degradation of the signal. These digital processes do not have to be linear – and so there does not have to be the same tied relation between amplitude and phase - although there may be if one wishes and one can compute perfectly flat amplitude AND phase/time response. With digital crossovers/correction this can be made true not just for the crossover filter itself but for the entire filter/driver system and so digital filters can correct for non-linearity/flatness of the drivers themselves that would otherwise produce phase/timing errors away from the crossover points (i.e. in band). As the above post points out, the DEQX system is the only one that also does this in-band correction and so are phase-linear (correct-timing) right across the frequency band. Which is why Legend Acoustics uses DEQX in its Tikandi loudspeaker system (www.legendspeakers.com.au/tikandi.html).

    It should perhaps also be noted that this ability of the DEQX to correct for phase/timing and amplitude errors of the drivers within band can also be applied to existing passive crossover systems. One measures the existing loudspeaker with its passive crossovers and then the DEQX “adds” the requisite corrections before the signal is fed to the speakers to create a finally (near) totally flat amplitude and phase/timing response for the overall system. We demonstrated this at RMAF with Legend’s passive Isobaric Small Red loudspeakers, with and without the digital correction. Because the ISRs are already amplitude very flat (as measured by Australian Hifi - see http://www.avhub.com.au/index.php?option=com_content&view=article&id=495&catid=6&Itemid=49&showall=1) the result was therefore not so much changes in tonal balance due to the small corrections in amplitude as the instruments etc snapping into better focus i.e. improved imaging/soundstage due to the small corrections in phase/timing. As Robert Harley, editor of The Absolute Sound magazine, said when he heard it “now I understand the need for digital crossovers/correction”!

    Dr Rod Crawford
    for Legend Acoustics
    www.legendspeakers.com.au

    ReplyDelete
  3. It's an honour to have someone of your caliber comment on my blog! Thanks for sharing. I have added the links from your comment.

    ReplyDelete

All comments are moderated.