I figured out that PVA does not stick to Epoxy...I am visualizing a 2 part mold from wood, using PVA glue for its non stick qualities to line the mold.
Does this make sense? Am I supposed to set the Ref Time to the first Peak?
Or maybe this?
Or maybe this?
Last edited:
The window needs to start just ahead of the first rise in amplitude. How much will only change the phase and you can find the absolutely correct time by adjusting it for the flattest phase response. What you show should be about right place to start. Definitely not the top picture.
This is about an inch and a half from mouth. I think the Dip from the previous measurement is due to the vertical reflections in the slot. The first little notch approximately 187hz is from lack of internal damping. The Dip at 250-260hz is the one I think is due to the horizontal planes of the slot. I say this because of how the carpet seems to have gotten rid of the null. The above is in the basement on carpet, the below is from when upstairs on hard floor.
Last edited:
Does this make sense? Am I supposed to set the Ref Time to the first Peak?
Good question.
For minimum phase time alignments, which most folks use.....you want to use the initial rise times of the impulse responses, to align the driver sections.
For the VHF section, initial rise time and peak impulse are very close. Not so, once you go lower in frequency. (Peak and initial rise are 1/4 apart for any given freq.)
For linear phase time alignments, yes you do want to align peaks.....which makes alignment much easier.
Absolutely correct time simply becomes distance between acoustic centers.
This condition exists, like Earl is saying, when phase is totally flat. (or you use nothing beyond 1st order xovers)
I think thats what was confusing me. Time alignment and setting the ref time for the IR window are not the same thing.
I didn't take it that way. "Flattest" gave me the impression of simply tuning out delay.This condition exists, like Earl is saying, when phase is totally flat.
You may be right. As I'm not sure what Earl meant.I didn't take it that way. "Flattest" gave me the impression of simply tuning out delay.
I forget that most folks use delay (fixed time) and phase as substitutes for each other.
I see them as completely separate, non-exchangeable, and need to remember that when commenting.
Phase and delay are linked together mathematically such that knowing one determines the other. Hence they cannot be thought of as independent.I forget that most folks use delay (fixed time) and phase as substitutes for each other.
I see them as completely separate, non-exchangeable, and need to remember that when commenting.
Perfectly flat phase can only be achieved with perfectly flat FR - never going to happen for a loudspeaker. When I say "flattest", I simply mean the deviations about the mean are the lowest. This is not the "minimum phase" as "flattest" could involve having a negative group delay. Minimum phase would not have any negative group delay.
Phase and delay are linked together mathematically such that knowing one determines the other. Hence they cannot be thought of as independent.
Thanks for the explanation of what you meant earlier.
I see phase and delay as linked together mathematically too, but at only any given frequency. Otherwise there is no strict relationship.
One varies by frequency, one doesn't
So functionally, I see them as independent, and not really interchangeable for multi-way speaker tuning purposes.
Perfectly flat phase can only be achieved with perfectly flat FR - never going to happen for a loudspeaker. When I say "flattest", I simply mean the deviations about the mean are the lowest.
I guess it depends on what we accept as perfectly flat FR. How much deviation we are willing to accept.
For me, the transfer function of the speaker below works as "reasonably perfectly flat FR". Both for magnitude and phase.
I think such "reasonably flat response" is pretty easy to obtain on-ax and in listening window with tody's processing options.
Tougher to obtain more off-ax of course, but still very decent.
Here's a 5-way.
(Red trace in magnitude window is Coherence, of the 650Hz to 4kHz driver section is this graph)
Oh Camplo, what i was saying earlier about impulse response alignments....the above is obviously tuned with linear phase xovers.
Which means impulse peaks need to be aligned, not initial rise like with minimum phase xovers.
Check out the top pane below, with all 5 of the impulse responses and how their peaks are time aligned, via constant Delay only.
And note the resultant mag and phase in bottom pane.
You can't change delay without changing phase, that's pretty strict. Yeah you can go back and move solely phase around to compensate but the initial moving of the start up transient will change delay and phase accordingly, isn't that strict?I see phase and delay as linked together mathematically too, but at only any given frequency. Otherwise there is no strict relationship.
One varies by frequency, one doesn't
Yes because you aligned phase using the peak of the start up transients as reference.5 of the impulse responses and how their peaks are time aligned, via constant Delay only.
And note the resultant mag and phase in bottom pane.
Last edited:
What you are missing is that a valid, useful phase measurement, is one in which all constant delay has been removed.You can't change delay without changing phase, that's pretty strict. Yeah you can go back and move solely phase around to compensate but the initial moving of the start up transient will change delay and phase accordingly, isn't that strict?
Measure any one of the 5-sections of the multi-way I posted at 1m. Remove the constant delay. Phase will be flat.
Move mic to 2m, remeasure and remove the constant delay(which has increased by 1m additional time travel. Phase will be flat.
So yes, you can change delay without changing phase.
Those aren't startup transients. Those are impulse responses of the drivers' entire bandwidths.Yes because you aligned phase using the peak of the start up transients as reference.
The thing is that they are both functions of frequency and as functions they are locked together by the fact that the delay at any given frequency is equal to the negative of the phase slope at that frequency. They simply are not independent and either can be used in any situation.One varies by frequency, one doesn't
So functionally, I see them as independent, and not really interchangeable for multi-way speaker tuning purposes.
Only "relative phase", not absolute phase (i.e. including and constant delay.)So yes, you can change delay without changing phase.
Impulses are transients and transients have a start but the start up transient is called that because it is the initial movement from Origin.Those aren't startup transients. Those are impulse responses of the drivers' entire bandwidths.
The thing is that they are both functions of frequency and as functions they are locked together by the fact that the delay at any given frequency is equal to the negative of the phase slope at that frequency. They simply are not independent and either can be used in any situation.
Disagree.
You are still focusing on the mathematical relationship at any ONE specific frequency.
Big whoop it exists...what does it mean?....why does it vary vs freq?
Remove fixed delay from a transfer function...then tell me how it has anything to do with phase.
Imso, a transfer function without fixed delay removed is near meaningless.
Please tell me why this train of thought in incorrect.
I'm adding a reply to help explain my position....Only "relative phase", not absolute phase (i.e. including and constant delay.)
Which is....only "relative phase" is Phase.
Absolute phase is meaningless and worthless for any kind of precise xover development.
Constant delay must be removed to get rid of the idea of absolute phase, and make phase have any real meaning.
You seem to be saying that delay and phase are different in more ways than simply by their relationship.
YES !You seem to be saying that delay and phase are different in more ways than simply by their relationship.
They are 100% different.
Their relationship is purely mathematical, and is good for only a single point on the frequency axis.
Their relationship has no practical value other than robbing Peter to pay Paul, when juggling them one for another.
- Home
- Loudspeakers
- Multi-Way
- Is it possible to cover the whole spectrum, high SPL, low distortion with a 2-way?