Yup, and it's exactly what I, and many others, do. Though I actually use eight amps, one for each bass driver parralleled, and one each for mid and highs.
I don't quite follow, do you mean you have a passive crossover for mid to tweeter sharing one output of your active crossover system ?
If it is active, do you have the drivers acoustic centres physically aligned, or are you adding digital delay to align them ?
What I was suggesting is that you have the ideal test setup to evaluate the importance (or lack of) acoustic centre alignment in the areas of coherence and imaging, as I proposed earlier in the thread.
If you start with a time aligned (whether physical or digital) idealised L/R 4th order crossover, you will have a flat excess group delay curve apart from the small bump near the crossover frequency.
If you then add delay to your mid (and woofers...) to bring the tweeter ahead by a delay equivalent to say 50mm, and then change your crossover slopes to regain phase tracking and flat amplitude summing through the overlap region, you now have a large excess group delay step, but with hopefully still a flat amplitude response.
Since its digital you should be able to do relatively easy A/B comparisons of the possibly subtle differences.
I know if I had an active system like that I would be performing such experiments to try to determine how important time alignment is and what the thresholds of audibility might be.
I've been meaning to get myself a DCX2496 for some time to allow me to do tests like this, but haven't got around to it...(I have a DEQ2496 but of course it's only an equalizer not a crossover...)
Last edited:
There's a passive between mid and tweeter. Sorry. 
It seems to me that the most direct way to test audibility of delay is to allpass the source files, then use headphones. It removes the room variable from the equation and allows very simple ABX confirmation of any perceived differences. That may not be perfect for imaging, but certainly for coherence.
It seems to me that the most direct way to test audibility of delay is to allpass the source files, then use headphones. It removes the room variable from the equation and allows very simple ABX confirmation of any perceived differences. That may not be perfect for imaging, but certainly for coherence.
Drat.There's a passive between mid and tweeter. Sorry.
But you've missed the whole point of the test though. Blauert and Laws have already done studies into the audibility of group delay of all pass filters. An all pass filter only introduces delay around the crossover frequency - far enough on either side the delay is zero, so in the middle of each drivers bandwidth there is no relative delay, only a small delay through the overlap region.
Acoustic centre misalignment doesn't introduce a hump, it introduces a step. So if your tweeter is ahead of the woofer by 0.2ms then there would be a flat delay of 0.2ms through most of the midrange drivers range, then a further increase at the crossover due to the allpass function then it would drop away to zero.
I don't think the audibility of this type of group delay step has been studied in the same way that allpass delays have been, and I don't think its safe to simply take the threshold figures from one and apply it to the other.
If it hasn't been researched its somewhat surprising given that most speakers do indeed have this type of response due to non alignment of acoustic centres.
Headphones might be a useful starting point, but I have to wonder whether the 3D nature of driver interference patterns in a room would introduce additional complexity, and whether the pinna response might also play a role.
If the physical speaker is not moved or changed in any way and the A/B testing is done via DSP delay of one driver and switching between different DSP filter slopes, I don't see why it couldn't be done in a valid way in a room like any other ABX testing.
Here's an example of what I'm talking about. Tweeter is 20mm ahead of the midrange, 3rd order crossover, you can see the bump from the crossover and the shelf from the acoustic centre offset. (The measurement isn't valid much below about 1Khz due to the FFT window time I used for that particular measurement, it should remain flat a lot lower than that)
An error as small as 20mm introduces nearly as much group delay as the peak delay from a typical crossover all pass response, but over many many more octaves. There are certainly plenty of speakers with more than a 20mm error between midrange and tweeter...just saying...
An error as small as 20mm introduces nearly as much group delay as the peak delay from a typical crossover all pass response, but over many many more octaves. There are certainly plenty of speakers with more than a 20mm error between midrange and tweeter...just saying...
Attachments
Wow 6 or 8 tube amps would be quite a home system. What kind of sound card are you using that has that many outputs? Some kind of expander board? Or are you using professional rack gear like was posted earlier?
Anyhow a more practical idea, from my point of view, might be to measure the delays, then re-encode the music to account for the delays and advances, then play in a normal fashion. You could have the alignment, but would still have the shifts if using crossovers, hope fully it would not sound computer generated.
Feed the speaker an error it can naturally correct.
Anyhow a more practical idea, from my point of view, might be to measure the delays, then re-encode the music to account for the delays and advances, then play in a normal fashion. You could have the alignment, but would still have the shifts if using crossovers, hope fully it would not sound computer generated.
Feed the speaker an error it can naturally correct.
You don't need that many. There's no reason whatsoever to ask an amp with an output transformer to produce 20Hz at hundreds of watts. In my setup, tubes drive midrange and tweeter, an old Adcom 555 drives midbass, and Sunfires drive the lower bass. The soundcard only outputs SPDIF, allowing six channels of DCX2496 to do their thang.
Getting a little more back on topic, earlier on in the thread it was mentioned that a 20KHZ tube would pass down a copper pipe wrapped around a coffee mug with little impedance.
Let's say you have a horn with an arbitrary length of 15' (4.5m). You also have a tweeter crossed over at 10,000hz.
The wavelength would fit comfortably inside a 1 3/8" (3.5cm) diameter tubing.
Assuming there are not too many losses down the tube, adding a 15'(4.5m) tube onto the tweeter would create a mechanical delay device. Lengthen or shorten to fine tune.
You could coil it up or perhaps run it inside the horn.
Let's say you have a horn with an arbitrary length of 15' (4.5m). You also have a tweeter crossed over at 10,000hz.
The wavelength would fit comfortably inside a 1 3/8" (3.5cm) diameter tubing.
Assuming there are not too many losses down the tube, adding a 15'(4.5m) tube onto the tweeter would create a mechanical delay device. Lengthen or shorten to fine tune.
You could coil it up or perhaps run it inside the horn.
Unfortunately for me, this does not work.
The tube used for testing is very lossy, packed with absorbent material.
Dr. Geddes relates, when discussing this same idea some time back, that there would be excess distortion created as the wave travels down the very long tube without any losses internally, to an exit... I darn sure wish it was not so, as that would solve a ton of problems. Drat.
_-_-bear
I have started a new thread - branched off to a new thread on the construction and mechanical/electrical aspects of the 555 driver.
It is here:western-electric-1928-we-555-dissect-design-diy.html#post3046788
C'mon down!
_-_-bear
It is here:western-electric-1928-we-555-dissect-design-diy.html#post3046788
C'mon down!
_-_-bear
I wouldn't accept them either, even though I used to use them at work. I want my analog chain to be all-analog.
I'm not sure what you mean by "...and part of a much larger and elaborate system", since any of the devices I listed has the freedom of adjustment necessary to be universal, not just components of one specific bespoke system.
But for experimental purposes, one could take one of these devices, and accomplish your stated goal. Re-combining the outputs could be done with a small mixer. If there's a specific purpose for this though experiment, it could be done, in the real physical world.
"and part of a much larger and elaborate system" as compared to the average or above average home HiFi, maybe the odd Bryston etc.
I assumed most of use here were not using rack mounted gear to overcome time delays at home.
Not so much an experiment, is anyone doing this on a day to day basis?
I assumed most of use here were not using rack mounted gear to overcome time delays at home.
Not so much an experiment, is anyone doing this on a day to day basis?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.