Driver time alignment

[Tony_J]

An interesting article from the miniDSP forum on using REW to sort out time alignment in multi driver speakers:

https://www.minidsp.com/applications/auto-eq-with-rew/measuring-time-delay

 

[TheFlash]

An interesting article from the miniDSP forum on using REW to sort out time alignment in multi driver speakers:

https://www.minidsp.com/applications/auto-eq-with-rew/measuring-time-delay

Don't you just use a sloping baffle like on the Thiels...? :minikev:
 

 

[Tony_J]

Don't you just use a sloping baffle like on the Thiels...? :minikev:
 

No 

 

[AudioTechnology]

I just saw this email from them.

Multi-way louspeakers can be more involved than just aligning the crests of the impulses like shown in their diagram. If the DUT is a 3-way or 4-way with a very low woofer crossover merge frequency, the designer has to choose where along the woofer's impulse they want to set the midrange and tweeter impulse. Using vector autoregression the summed responses should resemble a Dirac step, but this can result in a very long filter. Personally, this is why I wouldn't worry about time alignment in the lower frequencies, and when I design linear-phase loudspeakers I don't place a lot of weight on what is going on in that respect below 200Hz. The latency that results from trying to correct it and phase delineation causes audio to lose sync with video in real-time applications.

 

[greybeard]

Don't you just use a sloping baffle like on the Thiels...? :minikev:
 

Nigel, I do a very simplistic version of time alignment on my Behringer xover, that is my take on a sloping baffle. I measure the depth of the bass and mid drivers, from their throat to baffle, and then enter this figure into the time delay section on the Behringer for the respective drivers(this then converts the distance into millisecond time delays, for the drivers), and leaving the Bass at nil. Very basic I know, but to my simple brain, should work. I obviously wait to be corrected 

Sorry, to make this clearer, I delay the tweeter by the distance of the bass measurement, and then the mid, by the difference between bass and mid. Hope that makes sense  :rofl:

 

[Tony_J]

I hate to admit that I haven't bothered with time alignment so far, but plan to give it a whirl.

 

[greybeard]

I hate to admit that I haven't bothered with time alignment so far, but plan to give it a whirl.

To my ears it improves separation and placement of instruments, voices ect.

 

[MartinC]

I hate to admit that I haven't bothered with time alignment so far, but plan to give it a whirl.

So long as you're at least close to begin with, won't Dirac sort this for you?

My Shearwater speakers have a sloping baffle, which combined with the first order crossover filters meant that the uncorrected impulse response was reasonable, and not vastly changed by Dirac. There are compromises in other regards from this approach having been used though.

 

[Tony_J]

So long as you're at least close to begin with, won't Dirac sort this for you?

To be fair, it probably is pretty close to start with, as the Tannoy DC drivers should be pretty well time aligned anyway 

 

[Shadders]

To my ears it improves separation and placement of instruments, voices ect.

HI,

The established smallest interaural time difference that can be heard is 10us.

https://asa.scitation.org/doi/10.1121/1.5087566

It does seem strange that people worry about a few 10's of picoseconds jitter, yet most speakers have zero time alignment, and this is never questioned by people or reviewers. I suppose it is out of sight, out of mind.

Regards,

Shadders.

 

[andrew s]

I came across this yesterday https://www.audioholics.com/room-acoustics/human-hearing-phase-distortion-audibility-part-2

It seems to be a good review of what we know on the topic.

Regards Andrew 

 

[andrew s]

It does seem strange that people worry about a few 10's of picoseconds jitter, yet most speakers have zero time alignment, and this is never questioned by people or reviewers. I suppose it is out of sight, out of mind.

Regards,

Shadders.

I think jitter has other impacts in that it creates side bands not simple phase differences. 

This is from Benchmark 

"It can be easily demonstrated that most people have the ability to detect and identify tones which are buried 25 dB or more below white noise (A-Weighted). Therefore, it is important to keep jitter induced side bands nearly 25 dB below the A-Weighted THD+N of the converter, otherwise the jitter may become audible."

From this paper https://benchmarkmedia.com/blogs/application_notes/12142221-jitter-and-its-effects

Regards Andrew 

 

[Shadders]

I think jitter has other impacts in that it creates side bands not simple phase differences. 

This is from Benchmark 

"It can be easily demonstrated that most people have the ability to detect and identify tones which are buried 25 dB or more below white noise (A-Weighted). Therefore, it is important to keep jitter induced side bands nearly 25 dB below the A-Weighted THD+N of the converter, otherwise the jitter may become audible."

From this paper https://benchmarkmedia.com/blogs/application_notes/12142221-jitter-and-its-effects

Regards Andrew 

Hi,

I read the link - from 2002. There was a proposal by Neil Young stating that you don't need 24bit audio since 16bit audio can contain a tone at -105dB compared to the signal to noise capable by 16bit audio (96dB).

(https://web.archive.org/web/20120307015559/http://people.xiph.org/~xiphmont/demo/neil-young.html)

When you examine the audio file provided, to encode that sine wave at -105dB you need 5bits peak from the 16bits available, and you can actually see the sine wave, albeit blocky, in the signal using audacity.

Even the benchmark caveat the statement about jitter being audible in the same article :

"Jitter can only be considered totally inaudible if the worst case jitter induced sidebands are at least 23 dB below the A-weighted system noise. Above this level jitter may be audible or it may be masked by the program audio"

I am always sceptical when there are claims of signals being heard much lower than than the best hifi S/N of approximately 110dB. Frequencies seen on a spectrum analyser are not always representative of real world hearing and as usual, if someone is selling something, there is always poetic licence in their text to some extent.

Regards,

Shadders.

 

[andrew s]

Hi,

I read the link - from 2002. There was a proposal by Neil Young stating that you don't need 24bit audio since 16bit audio can contain a tone at -105dB compared to the signal to noise capable by 16bit audio (96dB).

(https://web.archive.org/web/20120307015559/http://people.xiph.org/~xiphmont/demo/neil-young.html)

When you examine the audio file provided, to encode that sine wave at -105dB you need 5bits peak from the 16bits available, and you can actually see the sine wave, albeit blocky, in the signal using audacity.

Even the benchmark caveat the statement about jitter being audible in the same article :

"Jitter can only be considered totally inaudible if the worst case jitter induced sidebands are at least 23 dB below the A-weighted system noise. Above this level jitter may be audible or it may be masked by the program audio"

I am always sceptical when there are claims of signals being heard much lower than than the best hifi S/N of approximately 110dB. Frequencies seen on a spectrum analyser are not always representative of real world hearing and as usual, if someone is selling something, there is always poetic licence in their text to some extent.

Regards,

Shadders.

Me too, I am normally sceptical but Benchmark ABX listen to their kit and I doubt they would have claimed audibility without testing it. But, that's just an opinion based on their track record. I like to find references based on measurements so will see what I can dig up.

Regards Andrew 

PS Here are some test so you can try for yourself how much below the noise floor you can hear. https://ethanwiner.com/audibility.html

 

[tuga]

To be fair, it probably is pretty close to start with, as the Tannoy DC drivers should be pretty well time aligned anyway 

It's also easier to time-align with coaxial drivers.

 

[Tony_J]

Indeed.

 

[Bokke]

An interesting article from the miniDSP forum on using REW to sort out time alignment in multi driver speakers:

https://www.minidsp.com/applications/auto-eq-with-rew/measuring-time-delay

This is cool 

Thanks