You're welcome!
In my usual haste I didn't take the time to work through it, just answered your Q, but using a 1 kHz XO led you to make a false assumption. For 333 Hz the delay is much longer since the WLs expand exponentially [1/f], so..........
~13543"-sec/333 = ~40.67"
~13543"/1000 = ~13.54"
~40.67"/~13.54" = ~3 ms ............. and ~0.3 ms for the tweeter.
Or simply 1/f = ms.
This assumes a coincident [point source] system, so for a typical multi-way there's the phase angle to consider.
Back when I learned this stuff you either did mass quantities of distance measurements and worked it out with a slide rule or then very expensive electronic calculators for arrays or normally for individual speakers just did the physical calcs, locations, then reversed the HF polarity to find the deepest null, screwed it down tight and switched it back to positive and it was what it was and since it was usually well inside the pioneer's 1/4 WL rule it was seamless at the measurement/listening area.
Re modern DSP, I've only just been exposed to it recently and was mightily impressed with what it could do to convert a horrible room, multi-way system into a 'HIFI' reproduction system, though with the downside that it was by far the least entertaining Altec horn based system I've ever auditioned, but the DIY and HIFI oriented 'hornies' really liked it to the point that some declared it maybe the best Altec based system to date, so for all its trickery, I lost a lot of interest in it for personal use.
That said, a local DIYer with a very large all horn 20-20 kHz multi-way system in a fairly large, open format room TD'd and EQ'd flat claimed it impressive but in discussions it sounded like the various delays were off, so he removed all EQ, unwrapped the phase of each, TD'd them till he wound up with a flat phase and amplitude response from 20-20 kHz, i.e. can near enough replicate a square wave = accurately reproduce a sound signal, with the most interesting part [to me] being that he didn't have to use any EQ, where initially he needed mass quantities of it, so back liking/wanting DSP.
Now he's trying to integrate two separate REL infra-bass sub systems, which I fear will be his undoing even with all the electronic didgeridoo. Guess I need to get an update and hopefully he'll publish his 'adventures' in audio system design once 'close enough' to perfection.
GM
Ah, I see where the problem is. This statement is plain wrong, where did you get that from?
A LR4 XO has the same identical phase (and thus group delay) response for woofer and tweeter, which means the drivers are in "total sync"... and my experiment will just show this and confirms theory.
But you don't want to time align the drivers because of crossover delay (1 period for LR4), you want to time align because the acoustical center of the drivers are ususally in different locations hence different distance to the listeing/measuring spot.
Phisically moving the higher frequency driver to compensate the delay caused by the crossover can work at one specific frequency but causes problems at other frequencies.
Last edited:
360 degree out phase (LR4) is not equal to 0 phase difference in the time domain to me (in the phase domain it is equal ofc), because the 360 degree out of phase response needs time to be "completed" , hence one cycle delay. Correct me i am wrong.
Last edited:
From Hypex HFD manual for example:
"Linkwitz-Riley
For best crossover filter results, Linkwitz-Riley filters are a good choice. They come
2nd, 4thand 8th
only in even-order, so like
etc. The 2nd
order filter will give you a phase
shift of 180 degrees. Best is to use the 4th
order Linkwitz-Riley, which will result in 360
degrees phase shift. That way, the low and high output appear in phase (although the
low output will have a one-period delay)."
Btw i think it's nothing wrong with LR2 where one driver needs reversed polarity compared to the other driver for phase matching.
Moreover LR2 have better time domain response compared to LR4 because it has only half cycle delay, therefore, the polarity of one drive must be reversed.
"Linkwitz-Riley
For best crossover filter results, Linkwitz-Riley filters are a good choice. They come
2nd, 4thand 8th
only in even-order, so like
etc. The 2nd
order filter will give you a phase
shift of 180 degrees. Best is to use the 4th
order Linkwitz-Riley, which will result in 360
degrees phase shift. That way, the low and high output appear in phase (although the
low output will have a one-period delay)."
Btw i think it's nothing wrong with LR2 where one driver needs reversed polarity compared to the other driver for phase matching.
Moreover LR2 have better time domain response compared to LR4 because it has only half cycle delay, therefore, the polarity of one drive must be reversed.
Last edited:
It is better that you don't use the term 'delay' when talking about phase. They are not the same.
This is audiophile conjecture. If I read this in the manual, I'd begin regretting buying the product.
Yes, I may have used the term 'delay' incorrectly.
And if the manual trying to suggest what's the "best" choice of crossover that doesn't mean that's for a bad product. That means it's not the most correct manual and nothing more.
Thank you for the illuminating discussion.
My question arose when I discovered that, especially with one particular piece of music, it sounded like the band members were deliberately trying play off-beat. It was horrible. However, both my DSP delays were well under 1 ms, so I was trying to figure out what was going on by understanding why those particular empirical delays had produced nulls when the mid was inverted and yet the music was so off-beat.
To me, the most likely answer was that there are an infinite number of delays that will produce nulls, but that they are infinite integers of periods apart. So how do I know that I have the right one? It now seems that this isn't the case, although I could still be off to a limited extent.
Rod Elliott alludes to the phenomenon of "slow bass", but doesn't elaborate much. I'm wondering whether that was what I heard from my ported box.
Since posing the question, I have been able to measure the acoustic delays to a reasonable first approximation. However, these produce only shallow nulls with the mid inverted, and tweaking them by a few hundredths of a millisecond either side makes the nulls even shallower. Clearly there's something else going on that I need to account for. I hope to get back to this later today or tomorrow.
I appreciate the stimulating discussion!
My question arose when I discovered that, especially with one particular piece of music, it sounded like the band members were deliberately trying play off-beat. It was horrible. However, both my DSP delays were well under 1 ms, so I was trying to figure out what was going on by understanding why those particular empirical delays had produced nulls when the mid was inverted and yet the music was so off-beat.
To me, the most likely answer was that there are an infinite number of delays that will produce nulls, but that they are infinite integers of periods apart. So how do I know that I have the right one? It now seems that this isn't the case, although I could still be off to a limited extent.
Rod Elliott alludes to the phenomenon of "slow bass", but doesn't elaborate much. I'm wondering whether that was what I heard from my ported box.
Since posing the question, I have been able to measure the acoustic delays to a reasonable first approximation. However, these produce only shallow nulls with the mid inverted, and tweaking them by a few hundredths of a millisecond either side makes the nulls even shallower. Clearly there's something else going on that I need to account for. I hope to get back to this later today or tomorrow.
I appreciate the stimulating discussion!
...meaning that the driver's natural roll-off is steeper than the filter's roll-off?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.