OK, but as the tweeter is now 20mm closer at the listening position than it was at the measurement position, is the tweeter's delay -20mm(-58us), or +20mm(+58us)?
Last edited:
Member
Joined 2003
Negative values move things closer to the listener. Sorry for bad advice above, if you're trying to fix delay so drivers represent correct delay if measured on-axis, correct for all drivers, not just the tweeter. If mic was centred in between woofer and tweeter, process can be avoided altogether since delay for each driver is the same value.
Yes, but my mic was 272mm below the tweeter axis, but only 46mm above the woofer axis.
As I said, I will probably bite the bullet and redo the measurements again in the New Year.
But again, this is where mixed information confuses guys like me who are just trying to learn the best way to do something.
One school says to locate mic on the tweeter's axis and take ALL far field measuring from there.
And the other says to locate the mic on the axis of each driver being measured.
I know that the mic would have to be precisely located each time (along with the acoustic reference speaker).
Moving the speaker being measured vertically onto the net driver's axis is an alternative, but I'm not sure how to easily move the speaker vertically only.
Sorry intervening, confusing it is until you make some measurements and mess with it few times and think everything through, no way around I think. You are well on your way so keep on keeping on
Acoustic timing reference seems a hustle but if dual channel measurement is not an option then that is it you need to manage with. If you cannot reposition speaker instead of mic to measure each driver individually on its on-axis (moving the mic would mess timing with the acoustic reference) then measure everything from the tweeter axis. Tweeter axis in order not to miss the beaming highs as explained earlier by DcibeL. Measuring every driver from one mic location will yield some error for short wavelengths for drivers other than tweeter but its probably okay, best you can do. By best I mean most accurate, least error. If you have dual channel measurement setup you could measure every driver on-axis with the baffle as common reference and get least error for each driver, procedure explained in the measurement manual I think. No need to adjust delays or anything, most accurate data with least amount of human error as long as distance to baffle was constant while measuring, amplifier and preamp settings not touched between measurements, same windowing (processing as in the manual) for all measurements. Note, there would still be error on low frequencies with small windowing we need to use in home semi-anechoic measurements so its never accurate to measure far field at home. End results are as good as the data, so one should concentrate effort to get the measurements right and processing of the data right and load it all into the simulator right and then know what error there is for even better results.
That said, if acoustic reference was all I had I'd conduct measurements outside ground plane from listening distance each driver without moving the mic or dut. This would be nice with two channel measurements as well why not, longer window and all. If you have to manage inside you could measure impulse from listening distance and estimate timing error with 1m single location measurements. Multiple things one could try do get more accurate data. Just think it through what is it you are looking at, wheres the error, how to validate magnitude of error if there is some.
Hope it helps you to think about thinking about it
Acoustic timing reference seems a hustle but if dual channel measurement is not an option then that is it you need to manage with. If you cannot reposition speaker instead of mic to measure each driver individually on its on-axis (moving the mic would mess timing with the acoustic reference) then measure everything from the tweeter axis. Tweeter axis in order not to miss the beaming highs as explained earlier by DcibeL. Measuring every driver from one mic location will yield some error for short wavelengths for drivers other than tweeter but its probably okay, best you can do. By best I mean most accurate, least error. If you have dual channel measurement setup you could measure every driver on-axis with the baffle as common reference and get least error for each driver, procedure explained in the measurement manual I think. No need to adjust delays or anything, most accurate data with least amount of human error as long as distance to baffle was constant while measuring, amplifier and preamp settings not touched between measurements, same windowing (processing as in the manual) for all measurements. Note, there would still be error on low frequencies with small windowing we need to use in home semi-anechoic measurements so its never accurate to measure far field at home. End results are as good as the data, so one should concentrate effort to get the measurements right and processing of the data right and load it all into the simulator right and then know what error there is for even better results.
That said, if acoustic reference was all I had I'd conduct measurements outside ground plane from listening distance each driver without moving the mic or dut. This would be nice with two channel measurements as well why not, longer window and all. If you have to manage inside you could measure impulse from listening distance and estimate timing error with 1m single location measurements. Multiple things one could try do get more accurate data. Just think it through what is it you are looking at, wheres the error, how to validate magnitude of error if there is some.
Hope it helps you to think about thinking about it
Last edited:
Yeah, tweeter is too far in the 1m measurements data compared to situation where you listen say 2.5m away. This starts to make significant error here in your case, as measurement distance is only 1m and c-c spacing seems to be ~30cm, quite large angle. For smaller speaker with smaller c-c this wouldn't be much of an issue. To fix the 20mm error in your case, you need to get tweeter closer to observation point so negative delay to tweeter or positive delay to woofer to better reflect reality at your listening position. As tweeter was measured off-axis at least top octave is probably more down than in reality so ignore some of the highs, its louder than the graphs show, adjust by ear.Yes, but my mic was 272mm below the tweeter axis, but only 46mm above the woofer axis.
As I said, I will probably bite the bullet and redo the measurements again in the New Year.
But again, this is where mixed information confuses guys like me who are just trying to learn the best way to do something.
One school says to locate mic on the tweeter's axis and take ALL far field measuring from there.
And the other says to locate the mic on the axis of each driver being measured.
I know that the mic would have to be precisely located each time (along with the acoustic reference speaker).
Moving the speaker being measured vertically onto the net driver's axis is an alternative, but I'm not sure how to easily move the speaker vertically only.
Lets guestimate error if you do not compensate the delay due to 1m measurement distance and you plan to listen at measurement height 2.5m away: Your c-c distance is quite large so crossover is somewhere 1-2kHz? If it was at 1500Hz, 20mm error would mean roughly 1/10wl so error around crossover is roughly 30 degrees in phase observed at the measurement height and 2.5m away. 30 degrees means about no difference in frequency response. Thinking timing, similar "error" is introduced back in if your head was ~15cm above or below design height there at listening position.
Reasoning from this the speaker will probably sound about the same if you compensate the delay or not.
Scale of things is this: while there might be better "time alignment" little above the measurement height (assuming tweeter is above woofer) because of the error, head ~15cm above measurement height so straighten your back while sitting on sofa for better sound. I'm quite positive it would be very hard to hear the difference though.
It is good to know what the error is about and you might as well fix it, assuming you will listen at that height, ear around woofer height. Use time delay to fix it, do not change the Z coordinate as it would make error off-axis.
Other lessons from all this might be that if one needs absolutely perfect "timing" make c-c much less so it varies less with change in ear elevation, point source would be best in this regard. Or always listen at design height. Always make sure there is no error in measurements, or if there is know what it is.
Last edited:
Member
Joined 2003
Let's be perfectly clear, no one says that for VituixCAD. Instruction has been with single focus and source of information from the start, "follow measurement guide for the software".
This problem you have made for yourself by continuing with USB mic and acoustic timing reference. Process can be very easy with the correct hardware, that is the "best way".
@tmuikku @DcibeL Thanks guys.
Given the cone "tweeter" I'm using, I'm probably trying to make a "silk purse out of a sow' s ear".
But I wanted to at least make the best effort, and maybe learn something at the same time.
Also this was meant to be a one- off project, hence not wanting to spend money on testing equipment such as a proper Mic and DATs, etc.
Running VituixCAD with the measurements I have, and given that I have a driver c/c of 318mm, my best crossover appears to be at 1kHz, using 2nd order Bessel filters on both.
The interesting thing is that adjusting tweeter delay by +20mm seems to bring the phases closer together than using a -20mm delay (in the GD and Phases graph).
Given the cone "tweeter" I'm using, I'm probably trying to make a "silk purse out of a sow' s ear".
But I wanted to at least make the best effort, and maybe learn something at the same time.
Also this was meant to be a one- off project, hence not wanting to spend money on testing equipment such as a proper Mic and DATs, etc.
Running VituixCAD with the measurements I have, and given that I have a driver c/c of 318mm, my best crossover appears to be at 1kHz, using 2nd order Bessel filters on both.
The interesting thing is that adjusting tweeter delay by +20mm seems to bring the phases closer together than using a -20mm delay (in the GD and Phases graph).
Unless you have the intention of adding this extra delay in your crossover, then keep the delay true and work with it.
If this gets your woofer a longer reflection free time in a way that could be helpful then it's worth considering.
I wouldn't assume that. Some prefer the presentation of a cone tweeter. For them it has a better directivity than a dome.
Thanks Allen.
My measurements of my cone tweeter 15deg off axis, probably accounts for a lot of what I'm seeing above 10kHz.
I think I'll take DcibeL's advice and re-do my far field measurements.
I'll just have to be very careful re-aligning my reference speaker with the re-located mic.
I know that everything else I've re-done in this project has led me to learn a bit more, and usually get a better/more accurate result.
After all, it's only taken me 6 months to get this far. What's another week or so.
I'm not sure how you mean.
I thought the reference speaker's tweeter would need to face the mic.
Also as the reference speaker is a bookshelf sized speaker, I didn't want it in a location where the sound from the speaker under test could reflect off it into the mic.
I could see your idea working if the reference speaker was just an isolated tweeter. But even then I think it would attract some early reflections.
I was always thought that locating the reference speaker adjacent to the speaker being measured, but just behind the baffle, would be the best solution.
I thought the reference speaker's tweeter would need to face the mic.
Also as the reference speaker is a bookshelf sized speaker, I didn't want it in a location where the sound from the speaker under test could reflect off it into the mic.
I could see your idea working if the reference speaker was just an isolated tweeter. But even then I think it would attract some early reflections.
I was always thought that locating the reference speaker adjacent to the speaker being measured, but just behind the baffle, would be the best solution.
Member
Joined 2003
Measurement mics are "omnidirectional" so it can face the mic from behind or below or where ever really, as long as it's loud enough for the mic to pick it up. Adjacent to speaker being measured is possibly the worst location.
Investigating impact of reference speaker location is easy. Just take a measurement without acoustic reference in place, then locate the reference speaker and re-measure and compare. Timing is not required for this type of comparison.
Investigating impact of reference speaker location is easy. Just take a measurement without acoustic reference in place, then locate the reference speaker and re-measure and compare. Timing is not required for this type of comparison.
Thanks so much for that explanation.
I am finding more and more that my logic is NOT necessarily correct, or the best way to do something.
I'm trying to talk myself into doing my measurements again before Christmas (get it over and done with), so I'm glad you replied quickly (don't you ever sleep????).
So, assuming I'm still correct regarding avoiding early reflections from my reference speaker cabinet, where would be best place to locate the reference speaker?
If it's next to the mic, it would cause early reflections.
Maybe off to the side of the mic? Say 90deg to the DUT/mic axis and 3m or more from the mic, so that any reflections off it's cabinet would be late/small enough to still allow a decent Impulse window?
In REW, I can check if the reference signal is being effectively picked up.
I am finding more and more that my logic is NOT necessarily correct, or the best way to do something.
I'm trying to talk myself into doing my measurements again before Christmas (get it over and done with), so I'm glad you replied quickly (don't you ever sleep????).
So, assuming I'm still correct regarding avoiding early reflections from my reference speaker cabinet, where would be best place to locate the reference speaker?
If it's next to the mic, it would cause early reflections.
Maybe off to the side of the mic? Say 90deg to the DUT/mic axis and 3m or more from the mic, so that any reflections off it's cabinet would be late/small enough to still allow a decent Impulse window?
In REW, I can check if the reference signal is being effectively picked up.
Last edited:
OK.
I re-did all my Far Field measurements at 1000mm, this time moving the mic onto each driver's horizontal axis.
The acoustic reference speaker was 1805mm off to the side of the mic.
I was very careful to keep this distance for both mic locations.
I then applied gating to all measurements before merging with my Near Field measurements.
Then exported as text files into VituixCAD.
I've located the two drivers on the baffle, relative to the listening height (X and Z are zero).
Do I need to make any timing/delay adjustments, or is that now all taken care of?
With no crossovers, just the drivers active, the phases are spinning like a top (don't remember that with the first measurements).
Haven't looked at adding crossovers yet.
Thanks Allen.
At least relative to each other, especially at the proposed crossover area, both driver's phase appear to be in unison.
Member
Joined 2003
Window start reference time in REW should be moved near start of impulse, sounds like you have lots of excess delay included from time of window start to impulse.
Alternative is to use timing offset when measuring to move impulse to t=0 when measuring so default window reference time location of zero can be used. This is the instruction in the measurement guide.
You can “fix the phase” as Allen suggests, but better to learn how to apply windowing properly from the start.
Alternative is to use timing offset when measuring to move impulse to t=0 when measuring so default window reference time location of zero can be used. This is the instruction in the measurement guide.
You can “fix the phase” as Allen suggests, but better to learn how to apply windowing properly from the start.
Last edited:
Yep, that made all the difference.
I also adjusted the delay of both drivers by -10us, which both the minimum phase a bit closer to the normal phase at the proposed crossover area.
Member
Joined 2003
The funny thing about REW, it that it doesn't treat phase relative to the window reference, but to t=0, so the delay between the acoustic reference and measured speaker is included, always. This is different from the way other software will process phase, where phase is represented relative to the window start time. In order to remove the excess delay in REW, either use a timing offset when measuring, or t=0 can be moved in the impulse response tab after measuring, but make sure same adjustment is made to all measurements so relative differences are maintained.
Alternative is to process measurements through VituixCAD IR -> FR tool. This can be done with REW, but some steps are required to export the impulse response from REW correctly. I've made a quick instruction on this, attached.
(Since you are using an acoustic timing reference, it will be important that the acoustic timing reference is closer to the mic than the speaker being measured for this instruction, since I am effectively deleting everything prior to t=0 in the export).
Member
Joined 2003
My Google Drive document collection has been updated with a new addition on the usefullness (or uselessness) of minimum phase. This document in combination with my other on single vs dual channel measurements should highlight the reason why dual channel measurements and using measured data including phase should be done where ever possible when designing with VituixCAD.
https://drive.google.com/file/d/1dq7mLeQMint8V95oJ1i6GXxyYRFfCgXG/view?usp=share_link
Complete set of documents is found here:
https://drive.google.com/drive/folders/1KDyECkAIVAuNEtKVobKmsfvtFczGqOQw?usp=share_link
https://drive.google.com/file/d/1dq7mLeQMint8V95oJ1i6GXxyYRFfCgXG/view?usp=share_link
Complete set of documents is found here:
https://drive.google.com/drive/folders/1KDyECkAIVAuNEtKVobKmsfvtFczGqOQw?usp=share_link