Time Alignment of Speaker Drivers

[Keith_W]

Correct driver crossover phase correction (hence time alignment) lifts the phantom stage up if your tweeter is above your midrange driver. That's an easy to detect audible improvement if you can switch between calibrations quickly.

That is a good point, there are some things you don't hear with headphones. Maybe I will go and do that experiment, it should take me a couple of seconds to dial in the time misalignment in my convolver.

 

[dualazmak]

Correct driver crossover phase correction (hence time alignment) lifts the phantom stage up if your tweeter is above your midrange driver. That's an easy to detect audible improvement if you can switch between calibrations quickly.

I essentially agree with your point which seems supporting/rationalizing my very unique physical positioning (under the woofer) of my super-tweeter FOSTEX T925A singing (ca. 8.8 kHz - 22 kHz) together with my tweeter singing (ca. 6 kHz - 16 kHz) positioned slightly above the midrange driver; please refer to my post #27, #485 and #912 on my project thread.

As for the details of my latest total system setup, please refer to #931 and #1,009.

I believe, these unique physical positioning, i.e. tweeter at top and super-tweeter at bottom both having 0.1-msec-precision time-alignment with all the SP drivers, as well as the wide-3D reflective dispersion of super-tweeter sound (summary ref. #929), greatly contribute to stunning 3D stereo sound-image/phantom-stage (and hence amazing disappearance of SPs) without upper shift (i.e. lift) of sound-image (phantom-stage) as I illustrated in #926 and #927.

Spoiler

Ref. #926; Listening evaluation of sound stage (sound image) using excellent-recording-quality lute duet tracks:

Ref. #927; Listening evaluation of sound stage (sound image) using excellent-recording-quality jazz trio album:

 

[Waveform Fidelity]

Unfortunately, the acoustic center of the woofer is not where the dust cap is. It is some distance in front of the driver and is typically outside the loudspeaker enclosure. Therefore, moving the tweeter backward is going the wrong direction if you want to physically "time align" the drivers.
(Picture source: https://www.aes-media.org/sections/uk/meetings/AESUK_lecture_0604.pdf, slide 4)

View attachment 502613

Richard Heyer’s papers are an excellnt source of understanding time delay distortion.

Time Delay Spectrometry: An Anthology - AES

Time Delay Spectrometry: An Anthology An anthology of the works of Richard C. Heyser on measurement, analysis and perception On the advice of the Technical Council, the AES has decided to make this anthology available on an Open Access basis, in order to disseminate the work of Richard Heyser as...

aes.org

 

[OCA]

Richard Heyer’s papers are an excellnt source of understanding time delay distortion.

Time Delay Spectrometry: An Anthology - AES

Time Delay Spectrometry: An Anthology An anthology of the works of Richard C. Heyser on measurement, analysis and perception On the advice of the Technical Council, the AES has decided to make this anthology available on an Open Access basis, in order to disseminate the work of Richard Heyser as...

aes.org

This is a 279 pages long combination of ancient papers with no relevant or useful information on speaker driver time alignment. It's quite basic and is so old that there's a US patent in it for measuring frequency response of a speaker with a microphone!

 

[gnarly]

This is a 279 pages long combination of ancient papers with no relevant or useful information on speaker driver time alignment. It's quite basic and is so old that there's a US patent in it for measuring frequency response of a speaker with a microphone!

That's perhaps the greatest mis conclusion I think I've read in a long time...!

 

[Waveform Fidelity]

This is a 279 pages long combination of ancient papers with no relevant or useful information on speaker driver time alignment. It's quite basic and is so old that there's a US patent in it for measuring frequency response of a speaker with a microphone!

Buried in those papers, I think (but I didnt check) you will find the paper on time delay distortion between drivers, if not then you can google time delay distortion Richard Heyser. I would describe the theory as fundamental rather than “basic” and it applies just as much today as when it was written.

 

[Waveform Fidelity]

That's perhaps the greatest mis conclusion I think I've read in a long time...!

Care to be more specific?

 

[Waveform Fidelity]

That's perhaps the greatest mis conclusion I think I've read in a long time...!

Ok so I went through the papers. Starting page 23 Time Delay Distortion in Loudspeakers and in particular fig 1a, then explain to us why you think this is not relevant to the topic of this thread.

What is your alternative theory on time delay distortion? Care to share that with us?

 

[gnarly]

Care to be more specific?

I think Heyser's work is amazing ...foundational stuff as relevant today as ever. I just wish i could understand more of it...cause simple it ain't !

 

[JPA]

Nice and easy with active systems. Just invert the phase on the tweeter and move the tweeter "in time"(delay in DSP) - according to measurements - until the deepest cancellation is obtained together with the midrange. Re-invert the tweeter, and we're in-phase. Fact-check with a final measurement.

It's not quite that easy. Yes, this process ensures that the two drivers are in phase at that frequency. But two drivers can be in phase with one lagging behind the other. You also have to ensure that they are on the same cycle, i.e. that they both started their forward movement at the same time.

Edit: @KSTR explained it better in post #34.

 

[Waveform Fidelity]

It's not quite that easy. Yes, this process ensures that the two drivers are in phase at that frequency. But two drivers can be in phase with one lagging behind the other. You also have to ensure that they are on the same cycle, i.e. that they both started their forward movement at the same time.

Edit: @KSTR explained it better in post #34.

Yes. Don’t confuse phase shift with time delay. Not so easy if the crossover filter is IIR Biquad and phase of each driver not known precisely. With FIR filters, one can completely compensate the phase first to be flat and the same for each driver - assuming you that extract the minimum phase of the driver correctly first. Then setting the delay so that the signal from each driver reaches the microphone at the same time is straight-forward. But in this process ^ you have to do the phase compensation before time alignment first.

 

[Waveform Fidelity]

I think Heyser's work is amazing ...foundational stuff as relevant today as ever. I just wish i could understand more of it...cause simple it ain't !

Sorry - I think I misunderstood your original comment. Yes, Richard Heyser’s life long work on TDS and time distortion are a major contribution to Audio Engineering. Did you know that the Acoustic center of a driver…. The central point where the the wave appear to originate, can change with frequency?

 

[KSTR]

Sorry - I think I misunderstood your original comment. Yes, Richard Heyser’s life long work on TDS and time distortion are a major contribution to Audio Engineering. Did you know that the Acoustic center of a driver…. The central point where the the wave appear to originate, can change with frequency?

When you have a driver with a slightly flexing but well-damped cone and dustcap this is the expected behavior, the HF acoustic center shifting towards the cone/dustcap junction

 

[RegularStereoGuy]

driver with a slightly flexing but well-damped cone

Ahh, the dreaded flapping woofer, the fifteen-incher in those giant Realistic speakers I used to marvel at in the Radio Shack back in 1987!

 

[Waveform Fidelity]

When you have a driver with a slightly flexing but well-damped cone and dustcap this is the expected behavior, the HF acoustic center shifting towards the cone/dustcap junction

I have actually detected the AC change slightly on the KEF UniQ mid-range driver. Its a metal AL cone with the coil driving point roughly midway up the cone (at the node postion) - so it acts pretty much as a rigid piston over its intended operating range. It does not have dust cap because thats where the tweeter fits. Anyway, its not obvious to me how the AC is formed in the near field. Conceptually yes, but a detailed theoretical understanding - no.

 

[OCA]

That's perhaps the greatest mis conclusion I think I've read in a long time...!

You were right and I was a bit unfair but it seems to have triggered and evolved in a very useful and informational discussion here so I will not regret

 

[OCA]

explain to us why you think this is not relevant to the topic of this thread.

This is essentially what that section explains: different frequencies emerge from a speaker at different times, with woofers inherently slower than tweeters. Ignoring this time alignment in multi-driver speakers degrades the sound, which can be diagnosed in the phase response. The paper buries this straightforward concept under ten pages of unnecessary jargon like "anomalous dispersion" and "minimum phase network".

What is your alternative theory on time delay distortion? Care to share that with us?

Getting time alignment right is simple: just measure each driver at the listening position with a timing reference. For phase alignment, use inverted allpass filters (1st or 2nd order) at the crossover frequencies to counter the phase shift from the filter slopes. Tools like rePhase make this process intuitive and far easier to understand than wading through impractical, academic papers IMO.

 

[gnarly]

You were right and I was a bit unfair but it seems to have triggered and evolved in a very useful and informational discussion here so I will not regret

Thank you OCA.
It kind of blew my mind that you looked down on Heyser's work, as I know how interested and competent you are with measurements.

Heyser was what I'd call an applied mathematician. He seemed most interested in determining how to make useful/valid acoustic measurements with available analog gear, that still complied with all known relevant waveform / signal theory math.
Do wish I could understand more of his work ...but I've long forgotten my college math minor.

 

[gnarly]

Did you know that the Acoustic center of a driver…. The central point where the the wave appear to originate, can change with frequency?

Yep, it's something I attempt to measure fairly often with my DIY speaker hobby.
Damn difficult.... so much so, I've come to think of it as 'the apparent acoustic center'. Partly because I think acoustic center can take on different meanings for different purposes.

Formally, mathematically, I see it as the point of adhering to 1/r distance attenuation.
From a speaker tuning, time alignment point of view, I see it as the distance measurement software provides via impulse arrivals.

And both of those viewpoints appear to have frequency dependency. yikes...
There's also the issue of how do you even determine the arrival time of a particular frequency? Scope a sine wave's time-of-flight from initiation, with a mic ? Hope not lol.
I like using a few cycles of a Gaussian wavelet as stimulus with our FFT's ...as an acceptable mixture of a defined frequency with time info.

 

[OCA]

Yep, it's something I attempt to measure fairly often with my DIY speaker hobby.
Damn difficult.... so much so, I've come to think of it as 'the apparent acoustic center'. Partly because I think acoustic center can take on different meanings for different purposes.

Formally, mathematically, I see it as the point of adhering to 1/r distance attenuation.
From a speaker tuning, time alignment point of view, I see it as the distance measurement software provides via impulse arrivals.

And both of those viewpoints appear to have frequency dependency. yikes...
There's also the issue of how do you even determine the arrival time of a particular frequency? Scope a sine wave's time-of-flight from initiation, with a mic ? Hope not lol.
I like using a few cycles of a Gaussian wavelet as stimulus with our FFT's ...as an acceptable mixture of a defined frequency with time info.

Maybe this can help you:

https://www.nature.com/articles/s43588-021-00183-z.pdf