Speakerplans.com Homepage
Forum Home Forum Home > General > Advanced Discussion
  New Posts New Posts RSS Feed - Delay settings soundsystem
  FAQ FAQ  Forum Search   Events   Register Register  Login Login

Delay settings soundsystem

 Post Reply Post Reply Page  <12
Author
Message
kipman725 View Drop Down
Registered User
Registered User


Joined: 02 September 2020
Location: Warrington
Status: Offline
Points: 231
Post Options Post Options   Thanks (1) Thanks(1)   Quote kipman725 Quote  Post ReplyReply Direct Link To This Post Posted: 18 August 2022 at 11:25am
If you perform the measurements with no crossover in place you get higher resolution data and data that is universally applicable; I.E it's the actual delay.  If you use these measured delays as a starting point for crossover design and don't vary them you should be able to ensure good summation if the drivers have roll-offs near the crossover points by using a mix of: Asymmetrical crossover slopes, all pass filters, filters with different HP and LP points.  A brute force approach (only works with floating point processors due to internal signal levels) is to flatten the drivers to an extra half octave either side of the crossover point using PEQs and then apply standard LR4/LR8 crossovers.  In most cases that should result in good summation.  A free tool to simulate your crossovers is VituixCAD

Back to Top
toastyghost View Drop Down
The 10,000 Points Club
The 10,000 Points Club
Avatar

Joined: 09 January 2007
Location: Manchester
Status: Offline
Points: 10920
Post Options Post Options   Thanks (1) Thanks(1)   Quote toastyghost Quote  Post ReplyReply Direct Link To This Post Posted: 18 August 2022 at 12:37pm
Originally posted by kipman725 kipman725 wrote:

If you perform the measurements with no crossover in place you get higher resolution data and data that is universally applicable; I.E it's the actual delay. 


That generally only holds true if you align using the group delay outside of the device passband, as shown in the Charlie Hughes guide. Experience users can do it other ways, but their process is established and they know what they’re looking for when making the measurements.

The problem with running devices with no filtering in place is that most people will align the IR peaks, which is not the true time of flight or first arrival. Intrinsically, the IR peak will be the high-frequency data - useless if you are trying to align a sub and full range cabinet.

Same for the ETC peak, although less so. The rise of the ETC is where you need to be looking - but this is difficult for seasoned folk, let alone beginners.

The other issue is that even if you get it correctly aligned without filters, the filters will then change the alignment through the acoustic crossover region. Unless they are carefully chosen — and parametric EQ or alternative filter shapes used to achieve a measured version of the actual target response of something like 4th order Linkwitz Riley — the delay values will cause crossover ripple at best, and cancellation at worst.

I should also point out that every DSP unit implements the same ‘standard’ filter shapes in varying different ways. Your simulated biquad filters aren’t always what comes out once the values are shoved into the actual hardware doing your processing.

Verification, verification, verification!

There is a series of 3 papers from the late 70s/early 80s by the illustrious Dick Heyser on determining the correct delay to observe the ‘true’ phase response of a loudspeaker if you’re interested in the deeper background.

Edited by toastyghost - 18 August 2022 at 12:39pm
Back to Top
kipman725 View Drop Down
Registered User
Registered User


Joined: 02 September 2020
Location: Warrington
Status: Offline
Points: 231
Post Options Post Options   Thanks (1) Thanks(1)   Quote kipman725 Quote  Post ReplyReply Direct Link To This Post Posted: 18 August 2022 at 2:18pm
REW doesn't use the IR peak if you use the estimate IR delay:
"Estimate IR delay calculates an estimate of the time delay in the measurement by comparing it with a minimum phase version. The delay it calculates can be removed from the impulse response by pressing the Shift IR button on the panel shown after the delay is calculated and can additionally be applied as a timing offset for subsequent measurements." https://www.roomeqwizard.com/help/help_en-GB/html/graph_impulse.html#top

Quote I should also point out that every DSP unit implements the same ‘standard’ filter shapes in varying different ways. Your simulated biquad filters aren’t always what comes out once the values are shoved into the actual hardware doing your processing
yes this is important, verify everything!

Quote he other issue is that even if you get it correctly aligned without filters, the filters will then change the alignment through the acoustic crossover region. Unless they are carefully chosen — and parametric EQ or alternative filter shapes used to achieve a measured version of the actual target response of something like 4th order Linkwitz Riley — the delay values will cause crossover ripple at best, and cancellation at worst.
To me this is more of a difference in methodology to achieve the same outcome.  There are many delay values that will work to achieve summation but only one that is the distance to the acoustic center (ignoring frequency variance) and this may not easily result in summation due to relative phase shifts between the boxes. so if tuning in the field you may want to perform a more iterative process with crossovers in place.


I am interested in the Dick Heyser papers if you have them?

Back to Top
303addict View Drop Down
Registered User
Registered User


Joined: 09 August 2022
Location: Belguim
Status: Offline
Points: 12
Post Options Post Options   Thanks (0) Thanks(0)   Quote 303addict Quote  Post ReplyReply Direct Link To This Post Posted: 18 August 2022 at 11:04pm



I am interested in the Dick Heyser papers if you have them?



also intresested in that how more info how better i will understand it beter and beter.

thx so mutch to ever one for all the posts!



Edited by 303addict - 18 August 2022 at 11:47pm
Back to Top
toastyghost View Drop Down
The 10,000 Points Club
The 10,000 Points Club
Avatar

Joined: 09 January 2007
Location: Manchester
Status: Offline
Points: 10920
Post Options Post Options   Thanks (1) Thanks(1)   Quote toastyghost Quote  Post ReplyReply Direct Link To This Post Posted: 19 August 2022 at 10:55am
Originally posted by kipman725 kipman725 wrote:

REW doesn't use the IR peak if you use the estimate IR delay:
"Estimate IR delay calculates an estimate of the time delay in the
        measurement by comparing it with a minimum phase version. The delay it
        calculates can be removed from the impulse response by pressing the
        Shift IR button on the panel shown after the delay is calculated and
        can additionally be applied as a timing offset for subsequent measurements." https://www.roomeqwizard.com/help/help_en-GB/html/graph_impulse.html#top



That’s definitely better than auto-aligning the energy peak, but it’s still not ideal in many real world situations. A minimum phase plot can only be accurately calculated from the Hilbert transform of the magnitude response if the data is anechoic.

That means you have to ensure you know the limitations of the measurement environment, make a large number of averages to improve the SnR and reduce the effects of stuff like wind, apply carefully chosen windowing to remove reflections, and then assess the results before you do the alignment auto-magic.

Using the group delay is much easier in my experience. Especially in many real world situations, where you might not have access to the DSP to remove the filters or be risking damage to components by running a high enough level log sine sweep to get a signal to noise ratio of >20dB at the absolute minimum.

When doing log sweeps, I use a local shelving filter at the audio interface outputs to protect HF drivers (and my ears). That doesn’t affect the results of the SnR is high, because the shelf is applied on the loop back too.

Quote

Quote I should also point out that every DSP unit implements the same
‘standard’ filter shapes in varying different ways. Your simulated
biquad filters aren’t always what comes out once the values are shoved
into the actual hardware doing your processing
yes this is important, verify everything!

Quote he other issue is that even if you get it correctly aligned without
filters, the filters will then change the alignment through the acoustic
crossover region. Unless they are carefully chosen — and parametric EQ
or alternative filter shapes used to achieve a measured version of the
actual target response of something like 4th order Linkwitz Riley — the
delay values will cause crossover ripple at best, and cancellation at
worst.
To me this is more of a difference in methodology to achieve the same outcome.  There are many delay values that will work to achieve summation but only one that is the distance to the acoustic center (ignoring frequency variance) and this may not easily result in summation due to relative phase shifts between the boxes. so if tuning in the field you may want to perform a more iterative process with crossovers in place.



Sure, there are different approaches when designing a system from scratch (or integrating mixed configurations of existing speakers) and working at a gig or venue.

There is always going to be some passband ripple, even if it’s only 0.2dB. Of course, moving off-axis does that more as the time of flight distance to each source changes.
Since the majority of listeners are off-axis — even for hifi — it’s also important to align or verify the results where people will actually be.

What is also often overlooked is the fact the choice of acoustic crossover slope affects the directivity. You can end up steering the main lobe up or down, or creating side lobes (vertically) that result in nasty reflections from the roof.

The ‘yellow bible’ book by Davis & Patronis Jr, Sound System Engineering, is a goldmine for this stuff with lots of nice diagrams. I got a second-hand copy of the third edition for a tenner on eBay.

Quote

I am interested in the Dick Heyser papers if you have them?



The three-part series is titled "Determination of Loudspeaker Signal Arrival Times" but unfortunately it doesn't seem to be open access on the AES e-Library.
Someone has uploaded the first part here, though it isn't the juiciest or most practical section:
https://sdlabo.jp/archives/Determination%20of%20Loudspeaker%20Signal%20Arrival%20Times.pdf

This four-page article isn’t about alignment explicitly, but it’s mostly plain English and has good diagrams regarding acoustic measurements in general. Start with that, before making too many choices based on what the computer shows:
https://www.prosoundweb.com/accuracy-applied-part-1-in-a-series-on-the-keys-of-loudspeaker-measurements/

Edited by toastyghost - 19 August 2022 at 11:05am
Back to Top
 Post Reply Post Reply Page  <12

Forum Jump Forum Permissions View Drop Down

Forum Software by Web Wiz Forums® version 12.06
Copyright ©2001-2023 Web Wiz Ltd.

This page was generated in 0.078 seconds.