I don't use (individual driver measurements for) xo simulation, so absolute delay timing reference is not needed. A single USB-mic, REW and multiway-dsp for xo/eq. For acoustic phase tracking/matching I measure two drivers simultaneously with xo, then study response, phase and step response. Adjust delay of higher driver by that info. Then inverting polarity for other driver is a sanity check when using symmetrical even order xo.
Last edited:
Since AP designs custom hardware, I never understood why they don't put that in the amplifiers feedback loop?
Now you have to deal with a 0.1 ohm error and find ways to compensate.
But yeah, in general that approach works much better than a fat resistor in series + you have the capability of doing current measurements as well.
Last edited:
Yes, I also think that measuring without timing reference is not a problem when using a DSP-based crossover.
But doing so to create a passive crossover would be a lot of IMHO unnecessary work and you have to own tons of different inductors, capacitors and resistors.
Kind regards
Michael
I've got opinion based on observation on the timing stuff, relating to Juhazi and Azrael comments about how important it is to measure timing:
Playing with my DSP controlled speakers, adjusting DSP while sitting other side of the room (where the DSP is), I cannot hear any difference adjusting delay of tweeter, until quite a long delay.
Starting to deviate from "optimal delay" first there is some frequency response dip forming, which is audible but quite hard to hear like any frequency response dip is. Then, many many milliseconds later, delay way way off, the tweeter and mid start to sound separate and this is clearly audible. This happens like 10ms delay, which equals three meters in distance, 10 feet.
My takeaway is this: if one listens far away in the room, or in other words uses strong early reflections to get "wide sound stage", one cannot simply hear differences with the timing, at least it's not high in list of importance. But, move your chair closer up, shrink the stereo listening triangle enough, so that early relfections get delayed and attenuated enough (compared to direct sound) so that stream separation happens in auditory system, clarity, now, even rather small delays seem perceivable.
What the close listening distance does it allows auditory system to lock in to the direct sound, and the lock in is basically due to good preservation of original harmonics (read Griesinger papers), which means group delay, edge diffraction, early reflections all kinds of secondary sounds kind of prevent, or reduce ability to lock into, reduce signal to noise ratio. What the preserved harmonics do they superimpose huge amplitude peaks for every fundamental cycle, making the sound stick out from all the sounds around us. Ruining the harmonics one way or another just reduces signal to noise ratio basically. When auditory system doesn't lock in it feels the sound is just noise, can't lock in, it's automatic process we cannot do nothing about except learn to listen for it and then take advantage of it, reason with it, gravitate to sound one likes to hear atm with the recording that happens to be playing.
Well, what ever it is eventually, delay mismatch with loud early reflections seem to make small delay / xo issues inaudible simply because most of the perceived sound is combination of direct sound and early reflections (6x boundaries on a cubicle combined is far louder than one direct sound), which are by far more scrambled than any small delay on xo region. Thus, agonizing about delays is irrelevant in my small experience as long as listening setup is like it usually is on audiophile photography, or in hifi shows, too big stereo triangle in reflective room, no toe-in, no directivity control and so on. So, if you plan to listen the "spacious hazy sound" at home then measuring with USB mike might be just fine, but if one want's to get accurate sound and is able and willing to set the system to get auditory system with it, then it pays off to measure and adjust delays accurately. I think it pays off to always measure accurately to be able to optimize for both perceptions, with or without auditory separation, which allows not just one good sound but two! It's just that a DIY setup could skip on it, as long as one always wants to enjoy only the spacious sound of loud early reflections, and never the accurate one, where brain pays attention to the sound.
This time delay stuff is relatively easy to experiment with active systems. I should make web app to allow testing this kind of stuff easily with any speakers: split signal, delay the other, assemble back, all in electric domain. While it's not too accurate it would still work to demonstrate this particular phenomenon I think, everyone could try.
Anyone tried? Did you hear a difference? We all have different listening skill/experience, so don't take my post as absolute truth, but just another perspective on things.
Btw, what's missing is how to measure auditory system. And it cannot be done with mic, or can it? Some calibration procedure somehow, hmm.
Playing with my DSP controlled speakers, adjusting DSP while sitting other side of the room (where the DSP is), I cannot hear any difference adjusting delay of tweeter, until quite a long delay.
Starting to deviate from "optimal delay" first there is some frequency response dip forming, which is audible but quite hard to hear like any frequency response dip is. Then, many many milliseconds later, delay way way off, the tweeter and mid start to sound separate and this is clearly audible. This happens like 10ms delay, which equals three meters in distance, 10 feet.
My takeaway is this: if one listens far away in the room, or in other words uses strong early reflections to get "wide sound stage", one cannot simply hear differences with the timing, at least it's not high in list of importance. But, move your chair closer up, shrink the stereo listening triangle enough, so that early relfections get delayed and attenuated enough (compared to direct sound) so that stream separation happens in auditory system, clarity, now, even rather small delays seem perceivable.
What the close listening distance does it allows auditory system to lock in to the direct sound, and the lock in is basically due to good preservation of original harmonics (read Griesinger papers), which means group delay, edge diffraction, early reflections all kinds of secondary sounds kind of prevent, or reduce ability to lock into, reduce signal to noise ratio. What the preserved harmonics do they superimpose huge amplitude peaks for every fundamental cycle, making the sound stick out from all the sounds around us. Ruining the harmonics one way or another just reduces signal to noise ratio basically. When auditory system doesn't lock in it feels the sound is just noise, can't lock in, it's automatic process we cannot do nothing about except learn to listen for it and then take advantage of it, reason with it, gravitate to sound one likes to hear atm with the recording that happens to be playing.
Well, what ever it is eventually, delay mismatch with loud early reflections seem to make small delay / xo issues inaudible simply because most of the perceived sound is combination of direct sound and early reflections (6x boundaries on a cubicle combined is far louder than one direct sound), which are by far more scrambled than any small delay on xo region. Thus, agonizing about delays is irrelevant in my small experience as long as listening setup is like it usually is on audiophile photography, or in hifi shows, too big stereo triangle in reflective room, no toe-in, no directivity control and so on. So, if you plan to listen the "spacious hazy sound" at home then measuring with USB mike might be just fine, but if one want's to get accurate sound and is able and willing to set the system to get auditory system with it, then it pays off to measure and adjust delays accurately. I think it pays off to always measure accurately to be able to optimize for both perceptions, with or without auditory separation, which allows not just one good sound but two! It's just that a DIY setup could skip on it, as long as one always wants to enjoy only the spacious sound of loud early reflections, and never the accurate one, where brain pays attention to the sound.
This time delay stuff is relatively easy to experiment with active systems. I should make web app to allow testing this kind of stuff easily with any speakers: split signal, delay the other, assemble back, all in electric domain. While it's not too accurate it would still work to demonstrate this particular phenomenon I think, everyone could try.
Anyone tried? Did you hear a difference? We all have different listening skill/experience, so don't take my post as absolute truth, but just another perspective on things.
Btw, what's missing is how to measure auditory system. And it cannot be done with mic, or can it? Some calibration procedure somehow, hmm.
Last edited:
I think even asynchronous D/A and AD could work, so you theoretically could use the other channel of the onboard sound card as timing reference. Provided latency on all converters is rock stable, that might be an issue though.
If timing is several cycles off, it comes audible.
Fractions of milliseconds accuracy is needed for MT xo, in my example xo is 3500Hz LR2 acoustic (outputs 3 and 4)
By the way I'm using two similar dsp units for L/R now and I haven't heard timing artefacts when listening to music.
Fractions of milliseconds accuracy is needed for MT xo, in my example xo is 3500Hz LR2 acoustic (outputs 3 and 4)
By the way I'm using two similar dsp units for L/R now and I haven't heard timing artefacts when listening to music.
Last edited:
I agree with what you're saying.
But measuring is not the same as listening.
Although maybe that's what you're trying to say as well!
For a measuring system reliability is important.
I don't mean reliability in accurate numbers, but how much you can trust your measuring setup.
You don't want weird unknown conditions happening randomly.
Even if that is like only 10% of the time.
(some people seem to be happy with 90% accuracy)
Especially if that can result in having to the measurements all over again.
Even for a very experienced user this makes you doubt everything, let alone for beginners.
Not to talk about the practical annoyances or complications down the road by not knowing something went wrong.
Since there are good alternatives available it 100% beats me to stretch this discussion?
It's looking for a solution for an introduced problem that isn't necessary to begin with.
The only difference is one XLR cable instead of a integrated audio interface in your microphone.
🤷♂️🤷♂️ 🤷♂️
Me too ! What a deal breaker a short mic cable is.
I have a measurement tower off the back of my deck. It's takes about 100ft of XLR for each mic, to get to my measurement station indoors...
Hello,
QuantAsylum has some nice kit that you can put in you backpack and take it outside and down the road or take out on the deck.
The QA461 is a test power amplifier with a internal 0.01R current sense resistor.
The QA472 is a low noise mic pre-amp with mic power supply.
Thanks DT
Thx,
but don't ever think it comes with quiet. A big motorboat lake is one noisy mother just from boats and PWCs !!
Plus, lot's of folks (like me for example) love blasting music outdoors, either from deck, dock, or boat/PWC.
I set up my audio interface in the "Semi Dual Channel Setup". I use ARTA.
For several years I used a Behringer UMC202HD. It was perfectly serviceable, and provided good quality polar frequency response data. Last year I decided I wanted to something which had better than a 80 dB SNR and better than 0.5% THD+N, so I bought a MOTU M2. Using a loop-back test, then internal noise in the unit was about -120 dB, and THD was less than 0.01%.
https://www.diyaudio.com/community/threads/arta.76977/post-7436346
For several years I used a Behringer UMC202HD. It was perfectly serviceable, and provided good quality polar frequency response data. Last year I decided I wanted to something which had better than a 80 dB SNR and better than 0.5% THD+N, so I bought a MOTU M2. Using a loop-back test, then internal noise in the unit was about -120 dB, and THD was less than 0.01%.
https://www.diyaudio.com/community/threads/arta.76977/post-7436346
@tmuikku: I'm not sure if I did understand you correctly.
In a DSP-based configuration i do also have a look at the impulse responses of the drivers in a multi-way-construction via overlay (which is no problem i.e. in REW): it is not very complicated to achieve phasecoherence. But the impulse responses have to be near to each other too.
I hope I'm understandable, since english is not my mother language and I'm trying to use translation programs as less as possible.....
Kind regards
Michael
edit:
It has just occurred to me that a correct display of the impulse responses of the drivers involved in a multi-way loudspeaker via Overlay is not possible without Timing Reference.
So I change my mind: using a timing reference is probably more useful than you think, even with DSP.
In a DSP-based configuration i do also have a look at the impulse responses of the drivers in a multi-way-construction via overlay (which is no problem i.e. in REW): it is not very complicated to achieve phasecoherence. But the impulse responses have to be near to each other too.
I hope I'm understandable, since english is not my mother language and I'm trying to use translation programs as less as possible.....
Kind regards
Michael
edit:
It has just occurred to me that a correct display of the impulse responses of the drivers involved in a multi-way loudspeaker via Overlay is not possible without Timing Reference.
So I change my mind: using a timing reference is probably more useful than you think, even with DSP.
Last edited:
Your power amplifier and mic-preamp is inside the loop with that measurement. Probably not a big deal, but you'd need to understand if, maybe, there's a latency variable, or polarity invert, or some other aspect to consider.
Also, some sound-cards don't have the same signal on the left and right outputs.
But, differential measurements are a powerful scheme to employ. For measurements these days I almost always use the FR2 mode of ARTA.
Dave.
Any modern audio interface can be just hooked up to the output of the power amplifier.
Since they have 48Vdc phantom power, this means they can withstand at least that DC voltage (in practice that's either 50Vdc or 63Vdc).
It depends a bit, but often 7Vrms or so is also not a big deal anymore.
This way, even bridged amplifiers can be used, since the input of any audio interface is balanced in.
This also works for impedance measurements btw
(the schematic ARTA shows is kinda old, from WinXP times when they had onboard single ended inputs)
Btw, SNR is never really an issue an can be improved by averaging the measurements.
Although it goes with a 10log, so I wouldn't bother with more than 8 or 16 averages.
Which gives 9dB or 12dB better SNR.
Also always use a swepped sine wave, there is a ARTA paper that describes pretty well how that gives you the best acoustic SNR.
I was planning to make a small voltage divider network to decrease the amplifier gain by ~20 dB... So you think this would be unnecessary? If so, that would be convenient.