@jhollander said:
IIRC you need to keep the impulse window the same for all the measurements.
That is my understanding as well. Also, the amplifier volume level, distance from mic tip to speaker, and distance of mic tip to floor need to be held constant as well. So when I begin to do spinorama measurements, my plan will be to first measure the tweeter on axis from 0 to 180 degrees in 10 degree increments. Then I will raise the speaker up higher on the turntable so that the midrange is on axis with the microphone and then measure the mid on axis from 0-180 degrees in 10 degree increments. Then I would raise the speaker up higher on the turntable again so that the woofer is on axis with the microphone and then measure the woofer on axis from 0-180 degrees in 10 degree increments. The microphone tip would always stay the same distance from the driver and the floor for all measurments.
Ah I remember now. That message in your screenshot is not an error really, it's just the SE can't be bothered to include the sample rate in the exported data, so you have to select it when you load the IR into VituixCAD.
The error that you get is the next message, where it says "length is shorter that 512 samples". I've inspected the file, and even loaded up SE on my end, and the problem is with the old v20 SoundEasy. It would appear that SE corrected the MLSSA txt export format starting in v24, unfortunately it looks like you'll need to update SoundEasy to export an impulse response that VituixCAD can use. USD$25 for an update, just email Bohdan with your dongle serial number. Another option is to ask Kimmo to update VituixCAD to include support to the 2 column format of older SoundEasy versions.
It is indicated right at the top of the instruction for VituixCAD in fact, that SE version 24 or later is required, which I had overlooked.
You can fix the file yourself, the problem is that it includes 2 columns of data, and you actually want a file that includes only the second column. This isn't too much trouble for 1 file using a spreadsheet program and loading space separated data, but can be quite tedious for every measurement unless you create some automated macro or something. Might not be worth the effort when you can upgrade for $25.
I've attached your IR file, but fixed so it will load properly in VituixCAD.
... Then I will raise the speaker up higher on the turntable so that the midrange is on axis with the microphone and then measure the mid on axis from 0-180 degrees in 10 degree increments. Then I would raise the speaker up higher on the turntable again so that the woofer is on axis with the microphone and then measure the woofer on axis from 0-180 degrees in 10 degree increments. The microphone tip would always stay the same distance from the driver and the floor for all measurments.
This is where I have a problem with the instructions. There's no mathematical formula for converting the midrange or woofer "on axis" measurement to the tweeter (design axis). Jeff Bagby and I had the same discussion about PCD.
Once you get up an running you can test this yourself with a 3-way where the woofer is a significant distance from the tweeter- design axis.
@dcibel said:
Ah I remember now. That message in your screenshot is not an error really, it's just the SE can't be bothered to include the sample rate in the exported data, so you have to select it when you load the IR into VituixCAD.
The error that you get is the next message, where it says "length is shorter that 512 samples". I've inspected the file, and even loaded up SE on my end, and the problem is with the old v20 SoundEasy. It would appear that SE corrected the MLSSA txt export format starting in v24, unfortunately it looks like you'll need to update SoundEasy to export an impulse response that VituixCAD can use. USD$25 for an update, just email Bohdan with your dongle serial number. Another option is to ask Kimmo to update VituixCAD to include support to the 2 column format of older SoundEasy versions.
It is indicated right at the top of the instruction for VituixCAD in fact, that SE version 24 or later is required, which I had overlooked.
You can fix the file yourself, the problem is that it includes 2 columns of data, and you actually want a file that includes only the second column. This isn't too much trouble for 1 file using a spreadsheet program and loading space separated data, but can be quite tedious for every measurement unless you create some automated macro or something. Might not be worth the effort when you can upgrade for $25.
I've attached your IR file, but fixed so it will load properly in VituixCAD.
Thanks for the revised file. I tested it and it works on my computer as well. I've already decided to purchase a license for ARTA, so this will not be a problem as soon as I do this. Will also send an email to Bohdon to upgrade my copy of SE as well.
I was thinking that another possible workaround would be to set up the window gating markers and mic calibration file in SoundeasyV20 and then convert each IR to SPL by clicking "IR>SPL". Then I would "Export SPL" to FRD and name each file with the appropriate driver name and polar angle. These files could then be loaded into the appropirate VituixCAD driver folders for processing. Or am I missing something?
... Then I will raise the speaker up higher on the turntable so that the midrange is on axis with the microphone and then measure the mid on axis from 0-180 degrees in 10 degree increments. Then I would raise the speaker up higher on the turntable again so that the woofer is on axis with the microphone and then measure the woofer on axis from 0-180 degrees in 10 degree increments. The microphone tip would always stay the same distance from the driver and the floor for all measurments.
This is where I have a problem with the instructions. There's no mathematical formula for converting the midrange or woofer "on axis" measurement to the tweeter (design axis). Jeff Bagby and I had the same discussion about PCD.
Once you get up an running you can test this yourself with a 3-way where the woofer is a significant distance from the tweeter- design axis.
John, wouldn't this be mathematically accounted for with the x,y,z coordinates that you enter for each driver? In the case where I would measure each driver (tweeter, mid, woofer) on axis, I would enter the x,y,z coordinates as 0,0,0 for each driver with no driver delay value entered in VituixCAD. Each driver would be using the as measured phase data, not minimum phase data. In the case where I would measure each driver on the tweeter axis only, then I would have to enter x,y,z of 0,0,0 for the tweeter, but would enter x,y,z of 0,-y mm, 0 for the mid and x,y,z of 0,-y mm, 0 for the woofer. The -y mm offset for mid and woofer that I enter would correct for the fact that I made all my measurements on the tweeter's axis. That would be my understanding. Not sure about this; I could be completely wrong.
What you've got above ^ is actually a bit backwards. z offset stays at zero since the delay is included in the measured phase. X an Y coordinates entered as usual since the measurements entered are on-axis with the driver. Off axis at any angle is simply calculated as delay and angle using the measured data and interpolation.
@4thtry said:
I was thinking that another possible workaround would be to set up the window gating markers and mic calibration file in SoundeasyV20 and then convert each IR to SPL by clicking "IR>SPL". Then I would "Export SPL" to FRD and name each file with the appropriate driver name and polar angle. These files could then be loaded into the appropriate VituixCAD driver folders for processing. Or am I missing something?
Yes of course you can do all the IR to SPL processing in Soundeasy, VituixCAD just has a way cooler and convenient window for it.
For processing in SE I would still recommend saving each IR measurement, in this case you would use the "save IR" button. If you make an error in window or timing you don't have to remeasure, just reprocess the IR. Loading the saved IR in SoundEasy is a bit odd, somehow a "load IR" button didn't make sense here, so you check the "use saved buffer" checkbox, then when you hit "Run MLS" a file open window will appear.
Got it! So now I think I am at the point where I can set up one of my "Retro Speako" loudspeakers and generate my very first "spinorama" model. I'm currently finishing up the bass reflex ports, adding internal braces, and sealing up all the interior wood surfaces. Thanks again, dcibel, for all of your detailed replies and pdf tutorials.
John, wouldn't this be mathematically accounted for with the x,y,z coordinates that you enter for each driver? In the case where I would measure each driver (tweeter, mid, woofer) on axis, I would enter the x,y,z coordinates as 0,0,0 for each driver with no driver delay value entered in VituixCAD. Each driver would be using the as measured phase data, not minimum phase data. In the case where I would measure each driver on the tweeter axis only, then I would have to enter x,y,z of 0,0,0 for the tweeter, but would enter x,y,z of 0,-y mm, 0 for the mid and x,y,z of 0,-y mm, 0 for the woofer. The -y mm offset for mid and woofer that I enter would correct for the fact that I made all my measurements on the tweeter's axis. That would be my understanding. Not sure about this; I could be completely wrong.
Bill what you are saying is the "on the woofer axis measurement plus delay" is the same as the "woofer measurement on the design axis (tweeter) measurement". PCD and I imagine VituixCAD can predict directivity based driver diameter, but I've not seen inverse directivity applied to initial measurements (if that's even possible). Driver shape is a physical feature that should be accounted for in your measurements.
BMR woofers and dome mids would be a good test. You can easily take two measurements and prove it to yourself.
No prediction of of directivity is completed in VituixCAD, real measured data is used. The only place you will see any prediction using a piston model is within the diffraction modeller, and the intent for this is for low frequency only of off-axis of near field, as the prediction is inaccurate at high frequency since speaker cones are not pistons.
You would not have the off-axis problems presented in PCD if it used more than a single axis of measurement data. Piston model is not very accurate, so real measured data is the solution to capture all real world effects of cone breakup, driver shape, waveguides and horns, diffraction, etc. No models and calculated predictions are required if you have a complete 360 degree balloon of measured data.
Agreed, but the measurements need to be on a design axis for the horizontal 360. You mentioned balloon so how's VituixCAD handling the vertical measurements? Would this provide data for the design axis? That would be many measurements if each driver has 2-axis, 360 measurements.
@jhollander said:
Agreed, but the measurements need to be on a design axis for the horizontal 360.
No, they don't. I've explained above already, the simulation when you change y axis for example of a driver is quite simple. distance from driver to mic is calculated so the change in phase from the delay is incorporated. SPL change over distance is simply calculated as well, and the on-axis response is gathered from the interpolation of the "balloon" of measured data.
The problem with anyone using PCD with on-axis driver information is that everyone using PCD is also using a USB mic, so delay between drivers is not easily determined if the mic isn't completely stationary across multiple measurements.
You mentioned balloon so how's VituixCAD handling the vertical measurements?
You can certainly measure both horizontal and vertical, for many speakers the horizontal provides information that is "good enough". The influence of the variance in diffraction vertically often isn't enough to have significant effect on overall power response for the "spinorama". For non-round drivers such as ribbon, AMT, horns, horizontal and vertical is a requirement.
If only horizontal data is provided to VituixCAD, it is assumed that vertical = horizontal.
For a comparison of how much effect the inclusion of the vertical axis has on a system, simply measure both horizontal and vertical axis, then run your simulations with and without the inclusion of the vertical measurements.
That would be many measurements if each driver has 2-axis, 360 measurements.
Yes, lots of measurements. I do 15 deg increments, so 24 measurements per speaker for horizontal, 46 if I want to include the vertical axis. Making the process simple as possible, in ARTA just hit record, save, move speaker and repeat. Then all measured data can be processed in bulk in VCAD IR - FR tool. ARTA includes some functionality for automated motorized measurement process if you want to get creative.
@dcibel said:
Yes, lots of measurements. I do 15 deg increments, so 24 measurements per speaker for horizontal, 46 if I want to include the vertical axis.
If the speakers are laid out symmetrically on the baffle don't you just need 0 - 180 horizontally (12 measurements) and tell VituixCad to mirror the other 180 degrees?
@4thtry said:
That was fast! I downloaded the new version and tested an "Export IR" text file from SE. It now opens with VituixCAD's conversion tool! Thanks much!
Kimmo provides great support, usually only a day or two for anything I've asked for provided he is in agreement with the suggestion. Donate to the cause under the help menu
Comments
That is my understanding as well. Also, the amplifier volume level, distance from mic tip to speaker, and distance of mic tip to floor need to be held constant as well. So when I begin to do spinorama measurements, my plan will be to first measure the tweeter on axis from 0 to 180 degrees in 10 degree increments. Then I will raise the speaker up higher on the turntable so that the midrange is on axis with the microphone and then measure the mid on axis from 0-180 degrees in 10 degree increments. Then I would raise the speaker up higher on the turntable again so that the woofer is on axis with the microphone and then measure the woofer on axis from 0-180 degrees in 10 degree increments. The microphone tip would always stay the same distance from the driver and the floor for all measurments.
Ah I remember now. That message in your screenshot is not an error really, it's just the SE can't be bothered to include the sample rate in the exported data, so you have to select it when you load the IR into VituixCAD.
The error that you get is the next message, where it says "length is shorter that 512 samples". I've inspected the file, and even loaded up SE on my end, and the problem is with the old v20 SoundEasy. It would appear that SE corrected the MLSSA txt export format starting in v24, unfortunately it looks like you'll need to update SoundEasy to export an impulse response that VituixCAD can use. USD$25 for an update, just email Bohdan with your dongle serial number. Another option is to ask Kimmo to update VituixCAD to include support to the 2 column format of older SoundEasy versions.
It is indicated right at the top of the instruction for VituixCAD in fact, that SE version 24 or later is required, which I had overlooked.
You can fix the file yourself, the problem is that it includes 2 columns of data, and you actually want a file that includes only the second column. This isn't too much trouble for 1 file using a spreadsheet program and loading space separated data, but can be quite tedious for every measurement unless you create some automated macro or something. Might not be worth the effort when you can upgrade for $25.
I've attached your IR file, but fixed so it will load properly in VituixCAD.
This is where I have a problem with the instructions. There's no mathematical formula for converting the midrange or woofer "on axis" measurement to the tweeter (design axis). Jeff Bagby and I had the same discussion about PCD.
Once you get up an running you can test this yourself with a 3-way where the woofer is a significant distance from the tweeter- design axis.
Thanks for the revised file. I tested it and it works on my computer as well. I've already decided to purchase a license for ARTA, so this will not be a problem as soon as I do this. Will also send an email to Bohdon to upgrade my copy of SE as well.
I was thinking that another possible workaround would be to set up the window gating markers and mic calibration file in SoundeasyV20 and then convert each IR to SPL by clicking "IR>SPL". Then I would "Export SPL" to FRD and name each file with the appropriate driver name and polar angle. These files could then be loaded into the appropirate VituixCAD driver folders for processing. Or am I missing something?
John, wouldn't this be mathematically accounted for with the x,y,z coordinates that you enter for each driver? In the case where I would measure each driver (tweeter, mid, woofer) on axis, I would enter the x,y,z coordinates as 0,0,0 for each driver with no driver delay value entered in VituixCAD. Each driver would be using the as measured phase data, not minimum phase data. In the case where I would measure each driver on the tweeter axis only, then I would have to enter x,y,z of 0,0,0 for the tweeter, but would enter x,y,z of 0,-y mm, 0 for the mid and x,y,z of 0,-y mm, 0 for the woofer. The -y mm offset for mid and woofer that I enter would correct for the fact that I made all my measurements on the tweeter's axis. That would be my understanding. Not sure about this; I could be completely wrong.
What you've got above ^ is actually a bit backwards. z offset stays at zero since the delay is included in the measured phase. X an Y coordinates entered as usual since the measurements entered are on-axis with the driver. Off axis at any angle is simply calculated as delay and angle using the measured data and interpolation.
Yes of course you can do all the IR to SPL processing in Soundeasy, VituixCAD just has a way cooler and convenient window for it.
For processing in SE I would still recommend saving each IR measurement, in this case you would use the "save IR" button. If you make an error in window or timing you don't have to remeasure, just reprocess the IR. Loading the saved IR in SoundEasy is a bit odd, somehow a "load IR" button didn't make sense here, so you check the "use saved buffer" checkbox, then when you hit "Run MLS" a file open window will appear.
Got it! So now I think I am at the point where I can set up one of my "Retro Speako" loudspeakers and generate my very first "spinorama" model. I'm currently finishing up the bass reflex ports, adding internal braces, and sealing up all the interior wood surfaces. Thanks again, dcibel, for all of your detailed replies and pdf tutorials.
Bill what you are saying is the "on the woofer axis measurement plus delay" is the same as the "woofer measurement on the design axis (tweeter) measurement". PCD and I imagine VituixCAD can predict directivity based driver diameter, but I've not seen inverse directivity applied to initial measurements (if that's even possible). Driver shape is a physical feature that should be accounted for in your measurements.
BMR woofers and dome mids would be a good test. You can easily take two measurements and prove it to yourself.
No prediction of of directivity is completed in VituixCAD, real measured data is used. The only place you will see any prediction using a piston model is within the diffraction modeller, and the intent for this is for low frequency only of off-axis of near field, as the prediction is inaccurate at high frequency since speaker cones are not pistons.
You would not have the off-axis problems presented in PCD if it used more than a single axis of measurement data. Piston model is not very accurate, so real measured data is the solution to capture all real world effects of cone breakup, driver shape, waveguides and horns, diffraction, etc. No models and calculated predictions are required if you have a complete 360 degree balloon of measured data.
Agreed, but the measurements need to be on a design axis for the horizontal 360. You mentioned balloon so how's VituixCAD handling the vertical measurements? Would this provide data for the design axis? That would be many measurements if each driver has 2-axis, 360 measurements.
No, they don't. I've explained above already, the simulation when you change y axis for example of a driver is quite simple. distance from driver to mic is calculated so the change in phase from the delay is incorporated. SPL change over distance is simply calculated as well, and the on-axis response is gathered from the interpolation of the "balloon" of measured data.
The problem with anyone using PCD with on-axis driver information is that everyone using PCD is also using a USB mic, so delay between drivers is not easily determined if the mic isn't completely stationary across multiple measurements.
You can certainly measure both horizontal and vertical, for many speakers the horizontal provides information that is "good enough". The influence of the variance in diffraction vertically often isn't enough to have significant effect on overall power response for the "spinorama". For non-round drivers such as ribbon, AMT, horns, horizontal and vertical is a requirement.
If only horizontal data is provided to VituixCAD, it is assumed that vertical = horizontal.
For a comparison of how much effect the inclusion of the vertical axis has on a system, simply measure both horizontal and vertical axis, then run your simulations with and without the inclusion of the vertical measurements.
Yes, lots of measurements. I do 15 deg increments, so 24 measurements per speaker for horizontal, 46 if I want to include the vertical axis. Making the process simple as possible, in ARTA just hit record, save, move speaker and repeat. Then all measured data can be processed in bulk in VCAD IR - FR tool. ARTA includes some functionality for automated motorized measurement process if you want to get creative.
I'll stop derailing Bill's thread as I've also answered why change in delay not equal to a vertical off axis measurement.
No problem, John. You are not derailing. I learn quite a bit from following detailed discussions such as this.
If the speakers are laid out symmetrically on the baffle don't you just need 0 - 180 horizontally (12 measurements) and tell VituixCad to mirror the other 180 degrees?
Yup.
@4thtry Today's VituixCAD update to 2.0.82.1 should fix your problem with loading SE v20 impulse response files.
That was fast! I downloaded the new version and tested an "Export IR" text file from SE. It now opens with VituixCAD's conversion tool! Thanks much!
Kimmo provides great support, usually only a day or two for anything I've asked for provided he is in agreement with the suggestion. Donate to the cause under the help menu