Timing is everything…

[JonFo]

As they say, timing is everything, and in audio, it’s critical.

After having finished the install and baseline tuning of the new acoustic treatments, I spent last weekend re-tuning timing and EQ.

I’ll cover the details of why timing is important, and how I tuned my setup in this thread.

First, as I’ve noted several times, (and many a company has built their designs on) the goal of any speaker system is to deliver a cohesive, time-aligned wavefront to the listener at all frequencies. And while it’s important for speakers, it’s just as important for ALL speakers in the room to have correct temporal alignment. Otherwise, the soundfield will not be unified, frequency balance is impacted and sound localization destroyed.

In a well-tuned system, playing well-mastered multichannel audio (any Porcupine Tree DVD-A qualifies), the sounds move smoothly across the speakers and the room without ‘breaks’ or discontinuities in frequency, gain or timing.

BTW- listening to PT’s ‘In Absentia’ right now, and I can hardly write, as it keeps drawing me into the most enveloping and engaging mix. Steve Wilson is superb mixing engineer as well as an awesome musician.

[JonFo]

So, what does adjusting timing really mean?

It means that if you do active crossovers (and use speaker processors like the DriveRacks), one must time-align the panels to the woofers. This is to ensure that the impulse responses of the two are in synch and creating cancellations/peaks in the crossover regions. As an example, my Monoliths require 0.54ms of delay relative to the woofers to align the impulses.
This shows up as improvements to the frequency response and in clearer articulation of sounds in the crossover region.

But intra-speaker is just the foundation for the rest of the timing exercise, as aligning to the subwoofer and to each of the 5 (or 7) other speakers is vitally important to the soundstage cohesion.

First a bit of theory:

In general, the goal of delay setting is to ensure that the sound arrival from all speakers is synchronous at the main listening area.

Since we can rarely deploy 7.1 speakers in a neat equidistant circle from the listener, some form of digital delay must be employed to ‘reposition’ the speakers.
This so that the arrival times of speakers closer to the listener align to the one that is the furthest away.

Thus most processors ask for the distances of the speakers so they can go through a calculation roughly like this:

Find the max and minimum distances within the speaker array. Then, for each speaker, determine the distance offset from the maximum distance.
This distance offset is then converted to time (in milliseconds) to then apply as a delay to that channel.

So for my System, the calculus goes a bit like this in the image below.

Note that my furthest distance is the Monoliths, so everything works off those distances to figure the appropriate delay.

Columns F and G are there to provide extra insight into the correctness of the delay values.

[JonFo]

Now for most everyone, the items above are typically all it takes to correctly set up a Receiver or pre-processor that manages all your speakers.

But, since often we have DVD-A/SACD players that also need timing corrections (they bypass the processor in many cases), this distance to delay conversion is helpful for those units that want to know the values in ms.


One important thing to take into account is if your sub or other external processing, adds delay of its own.
For instance, a Behringer BFD EQ (commonly used to EQ subs) or the Velodyne SMS-1, each add about 1 ms of delay on their own. So that additional delay needs to be factored into the calculus.

For instance, in my setup I not only have a BFD for the sub, but since I do all speaker crossovers and bass management using the DriveRack processor, I need to factor in about 0.6ms of latency from those units as well.

So rather than post a pic, I’ll just share the spreadsheet I used to calculate all this, as it might come in handy for those with more complicated setups.

This spreadsheet also tracks the various gain settings (also important) of the various elements, as well as relative phase and polarity settings (note how the three front panels are set to negative polarity relative to their bass elements, I think that’s because ML passive x-overs invert polarity.

Note that I decided to do the fine-tuned alignment between the three front speakers in the speaker-processors, therefore I ‘lie’ to the surround procs or DVD-A player and put all three at the same distance. I just wanted to be able to fine-tune the center relative to L/R without having to reconfig three or four devices ;-)

The column ‘Computed total delay’ is the one that factors in the sum of all delays from the surround processor, the speaker settings, plus the sum of the baseline latency for each of the speaker processors the signal must travel through.

While Ideally all three elements of a channel (Panel, woofer, sub) should be at +/- 0.5ms of alignment, the sub typically is anywhere from 0.5 to 1ms off based on actual ETF measurements of relative impulse delays.

OK, I’ll stop here, as I’ve probably confused everyone with too much detail from an overly complex system

[Steve]

Jon,

This may be an ignorant question, but for those of us with the factory passive crossovers, wouldn't ML have adjusted these for proper time alignment between the woofer and panel at the factory (or as part of the design)? Is this something that requires an active crossover to do properly?

Steve

[JonFo]

Originally Posted by Steve

Jon,

This may be an ignorant question, but for those of us with the factory passive crossovers, wouldn't ML have adjusted these for proper time alignment between the woofer and panel at the factory (or as part of the design)? Is this something that requires an active crossover to do properly?

Steve

Steve, that's a good question actually.

While some delay may be introduced by passive crossover components, it's generally not enough to truly compensate for more than an inch or so.
But yes, most speaker designers do try and compensate for these issues as much as possible. But to me, a passive topology is a very limiting medium to work with. But actives add significant complexity and cost.

So when I see designs like the Prodigy, that have the panel 3" to 4" ahead of the woofer, I just know there are some timing alignment challenges there. Maybe masked by other parts of the design, but I've always wanted to actually measure one for sure.

In the models I have (SL3, Sequel, Monolith) even though the panels are right above the woofer, all require a slight delay in the panel relative to the woofer for perfect impulse alignment. And the factory passive x-over does not provide that.

This is one of the reasons I so admire vertically integrated solutions like the Meridian DSP speakers, that combine amps, x-over, EQ, delay, etc. all custom designed for that model and built right in. Just feed a digital signal, and voila.

When I first heard about the ML Purity, I was hoping for something with more vertical integration, but sadly, it’s just a passive design with an on-board amp ahead of it.

[Craig]

Originally Posted by JonFo

Steve, that's a good question actually.

While some delay may be introduced by passive crossover components, it's generally not enough to truly compensate for more than an inch or so.
But yes, most speaker designers do try and compensate for these issues as much as possible. But to me, a passive topology is a very limiting medium to work with. But actives add significant complexity and cost.

So when I see designs like the Prodigy, that have the panel 3" to 4" ahead of the woofer, I just know there are some timing alignment challenges there. Maybe masked by other parts of the design, but I've always wanted to actually measure one for sure.

Are the crossovers in the Summit and Vantage passive or active? I'm not sure since they have a couple of amps in each one and a circuit board or two.

Designs like the Prodigy, Odyssey have both front and rear firing woofers. The Summits have a front and down firing woofer. How do you account for time alignment in those models?

Isn't time alignment less important between the lows and mids than it is between the mids and highs? The wavelengths are much larger the lower you go right? I would think time alignment is more critical in the higher frequencies than the low freqs.

I'm thinking more in regards to 2 channel than surround audio.