The procedure describes with details how this can be done. It uses REW instead of ARTA, where you can use "Acoustic timing reference" for timing .
Thanks for the suggestions - to answer the questions
The time alignment is to be used to time align a mid and tweeter possibly with a waveguide for the tweeter, it is a new project that will not use only passive crossovers and no dsp.
My test was using an old speaker with woofer-tweeter-woofer configuration this may change results ? , I am just learning how to do this type of measurement.
The mini dsp instruction suggests that 1meter is too close to test this - interesting and will read all the detail thanks ; but they make the test using REW a and a umk USB mic . I could use REW but was after advice on using ARTA that I have used longer .
ARTA does make dual channel measurements to achieve time reference and also compensates for the amplifier used in testing without the use of an audible reference . My mic is not a USB type .
My test method is simple and results in a direct readout of offset distance from a curser to marker time, someone who has ARTA may be able to say if this method is invalid or an over simplification .
Note photo of general 'Learning" setup , the mic would positioned 1m from front face in line with tweeter, it is closer here just to fit it in the picture . Normally I would measure in my rooms longest dimension and raise the box higher off floor . Or ideally measure outside - not often possible
Stephen
The time alignment is to be used to time align a mid and tweeter possibly with a waveguide for the tweeter, it is a new project that will not use only passive crossovers and no dsp.
My test was using an old speaker with woofer-tweeter-woofer configuration this may change results ? , I am just learning how to do this type of measurement.
The mini dsp instruction suggests that 1meter is too close to test this - interesting and will read all the detail thanks ; but they make the test using REW a and a umk USB mic . I could use REW but was after advice on using ARTA that I have used longer .
ARTA does make dual channel measurements to achieve time reference and also compensates for the amplifier used in testing without the use of an audible reference . My mic is not a USB type .
My test method is simple and results in a direct readout of offset distance from a curser to marker time, someone who has ARTA may be able to say if this method is invalid or an over simplification .
Note photo of general 'Learning" setup , the mic would positioned 1m from front face in line with tweeter, it is closer here just to fit it in the picture . Normally I would measure in my rooms longest dimension and raise the box higher off floor . Or ideally measure outside - not often possible
Stephen
Attachments
First of all, with dual channel in Arta, it is custom to use the " home" button, that would put the timing "0" at 300 samples. Secondly, its timing is quite accurate. So if you measure like the picture shows, the measurment for tweeter and for mid/woofer can be made at a distance that suits you better like 50cm. Important is not to move the box or the mic in between measurements.
The geometric isssue is not trivial at 50cm, but at increasing listening distance the difference reduces obviously.
So make the measurements, save them both of course.
Then load one and set it as overlay. then load the other, and zoom in, an exmple:
When zoomin in at a point you will see the measurement points (sampling rate), use gain to increase the vertical dimension if needed.
Now you can measure easily the horizontal distance between the peaks (these are impulse, so peaks) .
You can of course use left and right cursor if you want, like so:
In this example the time difference is 0.052msec, easily convertable to distance ;-)
The geometric isssue is not trivial at 50cm, but at increasing listening distance the difference reduces obviously.
So make the measurements, save them both of course.
Then load one and set it as overlay. then load the other, and zoom in, an exmple:
When zoomin in at a point you will see the measurement points (sampling rate), use gain to increase the vertical dimension if needed.
Now you can measure easily the horizontal distance between the peaks (these are impulse, so peaks) .
You can of course use left and right cursor if you want, like so:
In this example the time difference is 0.052msec, easily convertable to distance ;-)
The time difference between the peak maxima is not the whole story. The rise of the midrange signal is slower than with the tweeter, which amounts to a fraction of of the delay between the peak maxima. Therefore, even if the acoustic centersof bith drivers were at the same distance from the mic, the peak maximum of the mid would show up later than that of the tweeter.
In this example the time difference is 0.052msec, easily convertable to distance ;-)
In principle with a pure dirac signal, no the peak is the point. In reality this is not a pure dirac but a sweep converted into a pulse with the obvious noise.
When looking at a step then it is the point with the maximum slope, in theory the middle of the step flank.
What i showed was sample rate of 96k or 192k ( not at my desk right now)
What accuracy would that be in this case?
How stable is normally the acoustic centre over the frequency range in use?
When looking at a step then it is the point with the maximum slope, in theory the middle of the step flank.
What i showed was sample rate of 96k or 192k ( not at my desk right now)
What accuracy would that be in this case?
How stable is normally the acoustic centre over the frequency range in use?
Again, that depends on the rise time, which in turn corresponds to the frequency response. A tweeter (with a frequency response that extends far up into the spectrum) will have a fast rise time, while a woofer will have a slow rise time. Midrange drivers will be somewhere in between. The time of the peak maximum therefore depends on the frequency response and is therefore misleading to determine the acoustic center.
It would be (much!) better to look at excess phase in the frequency range where the drivers operate. Joe D'Appolito described this pretty nicely in his book.
Thanks for the improved method for measuring time alignment with ARTA
The idea was to time align tweeter and mid then measure all drivers directly to then create crossover with a simulator xsim or vitixcad.
The method described by Jan worked on my test box giving much the same results as my original measure but a much better presentation.
I had forgotten to use overlays, the example below I used 192K sampling rate improving accuracy over my lower rate original tests .This shows about 25 mm between drivers using the peaks - note the gate time is also shown as a distance.
If I try measure part way up the impules its not really possible , limited by the number of sample points avalable even with 192k sample rate
To check my distance results are in reality area, made a distances to tweeter face plate from time 0 (home) Then moved mic back 20mm measured with tape measure - and measured again . ARTA reported a change of change of 197mm seems within error of my rough tape measurement.
Then repeated the test with a DSP created delay of 0.08ms to the tweeter only using an old ver1 minidsp 2x4 that had 0.02 ms steps .
The Minidsp will not be used on the new speaker . It looks like the impules start about the same time, but peaks are different - please advice if this is inaccurate or irelevant.
Do these results look typical, maybe this is near enough?
I see there are more accurate methods and may attempt this when the project is using the real drivers for the project. I wonder how accurate it needs to be concidering most speakers do not physically time align at all and use crossovers to add delay or ignore it . It seems phase alignment may be more important.
On my test box the tweeter is offset horizontally to the two woofers, can I assume this will make a mess of any time alignment off axis. Would vertically in line drivers maintain time alignment off axis ?
The whole subject is much more complex than expected, with many opinions
Thanks for the help
The idea was to time align tweeter and mid then measure all drivers directly to then create crossover with a simulator xsim or vitixcad.
The method described by Jan worked on my test box giving much the same results as my original measure but a much better presentation.
I had forgotten to use overlays, the example below I used 192K sampling rate improving accuracy over my lower rate original tests .This shows about 25 mm between drivers using the peaks - note the gate time is also shown as a distance.
If I try measure part way up the impules its not really possible , limited by the number of sample points avalable even with 192k sample rate
To check my distance results are in reality area, made a distances to tweeter face plate from time 0 (home) Then moved mic back 20mm measured with tape measure - and measured again . ARTA reported a change of change of 197mm seems within error of my rough tape measurement.
Then repeated the test with a DSP created delay of 0.08ms to the tweeter only using an old ver1 minidsp 2x4 that had 0.02 ms steps .
The Minidsp will not be used on the new speaker . It looks like the impules start about the same time, but peaks are different - please advice if this is inaccurate or irelevant.
Do these results look typical, maybe this is near enough?
I see there are more accurate methods and may attempt this when the project is using the real drivers for the project. I wonder how accurate it needs to be concidering most speakers do not physically time align at all and use crossovers to add delay or ignore it . It seems phase alignment may be more important.
On my test box the tweeter is offset horizontally to the two woofers, can I assume this will make a mess of any time alignment off axis. Would vertically in line drivers maintain time alignment off axis ?
The whole subject is much more complex than expected, with many opinions
Thanks for the help
Attachments
For arta/steps/limp users, the last update 1.9.8 is totally free, in 2025 it will stop making other versions, enjoy it free. https://artalabs.hr/download.htm
Attachments
Does anyone know where I can find pcb source files for ARTA test jig? Kicad or whatever else. Maybe there is a pcb with a mic preamp as well ?
The ARTA jig is so simple, you can build it on a piece of wood or even free air wired in a small case. IMO the best option for a pre amp is an external sound card with XLR and 48V phantom power. If you want good, reproduceable measurements, sooner or later you will get such a thing anyway. You can prevent yourself from loosing time and money plus quite some frustration, if you go this route from the beginning. USB microphones are not useable when timing of signals get's critical. So a simple, maybe calibrated XLR measuring mike, is the best and cheapest way to go. Ask me how I know...
I was gonna also ask what extrnal soundcard to buy. I am just an occasional speaker builder. I know it would save a lot of hustle buying professional equipment but it costs too much money. Behringer 202HD is 200 Euro plus calibrated mic ..... eehhhh. I build speakers once every few years.
In the past I was using internal soundcard plus a calibration mic from my amplituner with a mic preamp. It was enough for me.
I just saw tha people use nice looking pcbs for ARTA jig. I was hoping I would make one too. I would last longer than the "spyder" type of build.
Can you recommend an external sound card that does the job and does not cost a fortune? It can be XLR or an ordinary one. I really do not need to go pro here. Impedance measurements are more important to me.
BigArtoo wrote: "Can you recommend an external sound card that does the job and does not cost a fortune?"
Yes, that is what I also like to know because I am intended also to build an test jig with an external sound card.
Yes, that is what I also like to know because I am intended also to build an test jig with an external sound card.
Second hand focusrite scarlett 2i2 second gen or thirth gen, used it for years and only this year not used.
There are plenty of used and cheap studio interfaces that will work great with Arta. It is important that it is not a Solo interface, and that it has a sampling rate of at least 96 khz.
You need a sound card with two channels and XLR phantom power. That is the best option. The Focusrite 2i2 Gen3 is about 119€ including postage and tax.
https://www.thomann.pl/interfejsy_audio1.html?__cTr=5c6e0c61-9246-4ff6-8a87--b289f0c08726&shp=eyJjb3VudHJ5IjoicGwiLCJjdXJyZW5jeSI6MTAsImxhbmd1YWdlIjoxMn0=&reload=1
If you are honest, it doesn't make sense to buy this one used. Anyway, even the Focusrite 2i2 Gen 1 or 2 is perfectly OK.
There are some other candidates, starting at 30€, but so far nothing I know that is more solid, percise and has a good software support. Software is something that can get you in trouble easily with used and even cheap new hardware. A Focusrite will keep a reasonable resale price for a long time.
Thomann has about 130 different USB interfaces, two channel + phantom XLR. Read the comments there, some are not what you expect from 2025 Plug&Play.
On the other hand, even a 48 kHz interface has enough resolution for speaker building, no need to go with 192 kHz. This will not change in the future, only the software support is a problem. Behringer, Swissonic, ESI are not the best in support. I have some fine interfaces on the shelf, that never worked well even with WIN7 because of driver trouble. Don't expect any Mac or Win 11 software...
PS No potentiometers on the interface is an advantage for ARTA IMO. If you get allong with the driver, the 39 € ESI Maya 22 USB is a good option.
https://www.thomann.pl/interfejsy_audio1.html?__cTr=5c6e0c61-9246-4ff6-8a87--b289f0c08726&shp=eyJjb3VudHJ5IjoicGwiLCJjdXJyZW5jeSI6MTAsImxhbmd1YWdlIjoxMn0=&reload=1
If you are honest, it doesn't make sense to buy this one used. Anyway, even the Focusrite 2i2 Gen 1 or 2 is perfectly OK.
There are some other candidates, starting at 30€, but so far nothing I know that is more solid, percise and has a good software support. Software is something that can get you in trouble easily with used and even cheap new hardware. A Focusrite will keep a reasonable resale price for a long time.
Thomann has about 130 different USB interfaces, two channel + phantom XLR. Read the comments there, some are not what you expect from 2025 Plug&Play.
On the other hand, even a 48 kHz interface has enough resolution for speaker building, no need to go with 192 kHz. This will not change in the future, only the software support is a problem. Behringer, Swissonic, ESI are not the best in support. I have some fine interfaces on the shelf, that never worked well even with WIN7 because of driver trouble. Don't expect any Mac or Win 11 software...
PS No potentiometers on the interface is an advantage for ARTA IMO. If you get allong with the driver, the 39 € ESI Maya 22 USB is a good option.
Last edited:
Has anyone tried the Arturia MiniFuse 1? It has virtual loopback, I wonder if that would work and how accurate it would be.
Depending on your needs, but for just some simple and occasional loudspeaker/acoustics measurements, any USB audio interface with phantom power will do with two inputs and outputs (ideally similar).
Ideally 96kHz, but it's not strictly necessary.
These days that will be around 50-70 bucks new price, or a lot less 2nd hand.
Which I don't think is a fortune?
If there is enough interest, I might be able to help with a test jig, or maybe even a complete test amplifier.
Ideally 96kHz, but it's not strictly necessary.
These days that will be around 50-70 bucks new price, or a lot less 2nd hand.
Which I don't think is a fortune?
If there is enough interest, I might be able to help with a test jig, or maybe even a complete test amplifier.
Please post in English.Translation:
Sorry, I'm breaking into this thread because I got the message that I can't send private messages for 24 hours. Hopefully you see this and can give me an email address ouzo. Otherwise I'll respond tomorrow afternoon. Greetings, Ed