Building Monitors - Time Alignment?

[bino_5150]

Ok, so I have decided to take the plunge and build my own studio monitors...

The one part of the design that I'm seriously lacking in is the time alignment of the drivers. This is a very crucial thing, and I am hoping that someone out there has some experience with this...

Let the games begin...

[tINY]

Once you decide on drivers and develop an initial cross-over, you then have to look at the phase and/or group delay in the transition band of the filters. Don't neglect the delay through the horn if you are using one.

Then, you need to offset the acoustic centers of the drivers to find the best match or use all-pass filters to move the group delay around (or both).



-tINY

[bino_5150]

The drivers that I'm using are:

Vifa XT25TG30 dual concentric diaphragm dome tweeter

Scan-speak 18W/4531G-00 Revelator 7"

Madisound Speaker Components, Inc. is doing the crossover with their LEAP-5 software.

At which point, I guess I'm in trial and error mode as far as time alignment?

Thanks for the response... That's good shizznit

[tINY]

Once you get them together, you could measure them....




-tINY

[bino_5150]

So correct me if I'm wrong, but if I recess the tweet and maybe make the front of the enclosure bubbled out to bring the woofer out a bit further...

The closer I get to aligning the voice coils on vertical axis, the closer I am to time alignment?

[tINY]

Only if the interaction between the cross-over and the drivers work out that way.




-tINY

[mikeg]

I'm going to assume that Adam at Madisound is doing the xover. Don't be afraid to ask him for help, advice. He's very very good.

Or, you can send him one unit, after you get it built, and they will measure and allign for you. It isn't real spendy. They've done work like this for me. I'm a happy customer.

[bino_5150]

Thanks mike. I have been emailing back and forth with Josh lately, and he suggested that I get with Tom about the details. I'm not sure exactly who is going to build the actual crossover.

[Wavebourn]

Originally Posted by bino_5150

The closer I get to aligning the voice coils on vertical axis, the closer I am to time alignment?

No. In my current monitors I have 2 vertical rows: of tweeters, and of midrangers.
The distance between radiating surfaces and a measurement mic matters. A triangle.
The close are drivers to each other, the wide listening angle you may afford. But you can't put more than one driver on the same place, however...

Differences between distances from one driver to the measurement mic and from another driver to the same mic divided by speed of sound give you time difference.

Knowing time difference you may calculate phase shift on a crossover frequency. Crossovers give phase shifts as well, so now you may play with distances and phase shifts to align drivers in complex, i.e. electrically and physically. If the system is bi-amped you may use an APF phase corrector. If not, sometimes it is easier to swap polarity of a tweeter than move it physically. As Tiny suggested, build and measure, then adjust electrically: anyway you can't listen always from the same point, and drivers themselves have deep and sharp bumps that can't be corrected by equalizers.

[Jim Williams]

One trick a speaker designer told me was to use a length of 18~20 awg insulated wire and wrap it in a coil around a pencil. Put it in series with the hot lead to the tweeter. Feed a click to the speaker. Unwrap the coil slowly until the click times up. Then slide it off the pencil and wire it in place.

Jim Williams
Audio Upgrades

[tonymite]

[barefoot]

Originally Posted by bino_5150

Madisound Speaker Components, Inc. is doing the crossover with their LEAP-5 software.

At which point, I guess I'm in trial and error mode as far as time alignment?

They will take care of this. You need to tell them if the drivers are mounted on a flat baffle or physically time aligned with a stepped baffle. They will alter the crossover parameters in such a way that the drivers are phase aligned on-axis. This can include using all-pass filters, although they are usually not necessary, especially with 2nd, 3rd and 4th order crossovers. The phase manipulation can readily be 'built in' to higher order crossovers. Personally, I'm not a fan of most stepped baffle designs. The reflection and diffraction anomalies produced by those surfaces and edges being so close to the tweeter typically cause many more offenses than the minor benefit the physical time alignment yields.

You want to make sure they are building baffle-step compensation into the crossovers as well. I recall having Madisound LEAP design me a crossover back in the early 90's when I was just a hobbyist. They didn't account for the baffle step and the result was an overly bright speaker. Maybe they're more sophisticated now. They should ask you for the width of your speaker cabinet. The baffle-step compensation can also be built into the low-pass section of the crossover.

Hope this helps!
Thomas

[bino_5150]

Hey good looking out everybody! Thanks for all the replies and good information.


I'll keep everyone posted on the progress of the project.

[Wavebourn]

Originally Posted by Jim Williams

One trick a speaker designer told me was to use a length of 18~20 awg insulated wire and wrap it in a coil around a pencil. Put it in series with the hot lead to the tweeter. Feed a click to the speaker. Unwrap the coil slowly until the click times up. Then slide it off the pencil and wire it in place.

Hmmm.... It sounds strange: if a coil has enough of inductivity to affect phase shift on a crossover frequency it should introduce roll-off on higher frequencies.
And he should need couple of thousand of turns on a pencil to get half of milliHenry.

[barefoot]

Originally Posted by Wavebourn

Hmmm.... It sounds strange: if a coil has enough of inductivity to affect phase shift on a crossover frequency it should introduce roll-off on higher frequencies.

Yeah, that didn't quite make sense to me either. But I thought it might be better to just step over that sleeping dog. You're braver than I am.

[Wavebourn]

Originally Posted by barefoot

Yeah, that didn't quite make sense to me either. But I thought it might be better to just step over that sleeping dog. You're braver than I am.

By the way, speaking of sleeping dogs, is 38 Hz - 20 kHz (+/- 1.5 dB) electrically, or acoustically? (I just clicked on your site link)
Can you show a burst decay graph with 1/64 Oct smoothing?

[barefoot]

That's +/- 1.5 dB acoustic response with 1/12 octave smoothing for production data. Our mics are calibrated to +/- 0.5 dB. The actual anechoic data is close to +/- 1 dB with no smoothing. However, I don't have that kind of facility for production testing. And our background noise is far from ideal. 1/3 octave smoothing is probably the closest thing to an industry "standard" for published response data. But it's really all over the map. So I just try to be more conservative than most in the data we publish, yet still give numbers that are reasonable to compare.

Achieving a flat on-axis response is a relatively trivial matter these days. So I don't consider these data to be particularly illuminating. They're really just there to make the customer feel confident that you're not way off the mark. It's a thousand other things that separate great speakers from mediocre ones. And those are better described in a short book rather a few lines in a spec sheet. Or, a simpler method is to simply listen.

[Wavebourn]

It is really impressive.
What kind of EQ do you use, digital or analog?