hi at all, I have a question about speaker measurements.
I built a stepped baffle speaker. It is a 3 way with mid and woofer 25mm forward regard to the tweeter.
How to get measurements? i explain:
normally, with no stepped baffle, I perform measurements leaving the mic at tweeter height or at middle height between mid and tw, distance 1000mm from baffle or less and get measurements of each driver individually without moving the mic. i perform dual channel measurements with ARTA for capturing also the phase.
but with stepped baffle, measuring this way, distance mic-baffle is not the same for the drivers because tw baffle will be at 1000mm from mic, and mid/woofer baffle will be at 750mm from mic.
need some suggestions. thank you
BR
I built a stepped baffle speaker. It is a 3 way with mid and woofer 25mm forward regard to the tweeter.
How to get measurements? i explain:
normally, with no stepped baffle, I perform measurements leaving the mic at tweeter height or at middle height between mid and tw, distance 1000mm from baffle or less and get measurements of each driver individually without moving the mic. i perform dual channel measurements with ARTA for capturing also the phase.
but with stepped baffle, measuring this way, distance mic-baffle is not the same for the drivers because tw baffle will be at 1000mm from mic, and mid/woofer baffle will be at 750mm from mic.
need some suggestions. thank you
BR
Preparation of response measurements for crossover simulation with VituixCAD
Page 3:
Rotation center on Z-axis while off-axis measurement sequence:
...
b) Rotation center on Z-axis varies with stepped baffle. Drivers on each baffle level has own rotation center on Z-axis. Distance from each baffle level to microphone must be constant (100 cm). Differences on Z-axis are entered to the simulator as Z mm of the driver, e.g. tweeter Z=0 mm, mid range Z= 20 mm, woofer Z= 100 mm.
Page 3:
Rotation center on Z-axis while off-axis measurement sequence:
...
b) Rotation center on Z-axis varies with stepped baffle. Drivers on each baffle level has own rotation center on Z-axis. Distance from each baffle level to microphone must be constant (100 cm). Differences on Z-axis are entered to the simulator as Z mm of the driver, e.g. tweeter Z=0 mm, mid range Z= 20 mm, woofer Z= 100 mm.
I do not think i understand your problem, cause in arta you set the measurement time window for the driver that has the longest path length to the mic, and leave this setting for all the drivers. In this way all driver measurements contains correct phase for crossover simulation. Then when you move the mic for another angle you just repeat the above
cucicu -
I see your problem, I think originally you were missing a step!
Gather FR, impedance and phase like you do now. But you need some way of measuring the acoustic diffences between the 3 drivers. Usually this is vai interferometry.
Measure the mid, measure the tweeter, then measure them both at the same time. Enter data into crossover simulation and adjust delay until the combined outputs match.
I see your problem, I think originally you were missing a step!
Gather FR, impedance and phase like you do now. But you need some way of measuring the acoustic diffences between the 3 drivers. Usually this is vai interferometry.
Measure the mid, measure the tweeter, then measure them both at the same time. Enter data into crossover simulation and adjust delay until the combined outputs match.
Difference between rotation center and acoustical center of driver is captured by dual channel measurement, and value is included in phase response. Requirements are that you keep distance between mic and rotation center constant (1000 mm), and keep Reference time constant in IR->FR conversion for all drivers while exporting frequency responses.
Everything will be taken care if you follow linked document. That method supports different types of speakers (dipoles, closed, cardioid, mixed) and drivers no matter are they minimum-phase or not, and how far acoustical center locates from baffle surface i.e. rotation center.
This is only accurate method to capture timing differences with turntable measurement also with box speakers if geometry simulation in crossover simulation does not calculate accurate path length from radiator(throat) via possible horn/wave guide to surface(mouth) and edge of front baffle. I don't know any XO simulator what could do that, so measuring distances to acoustical centers and entering difference to Z coordinate is bad method if target is to measure and simulate whole 0...90/180 deg to get also valid off-axis result, power response and DI. Other than conventional box speakers may need this method even more.
This is method is decent if drivers are "flat" and
a) Difference in acoustic centers is entered to delay (time) with unidirectional radiators such as box speakers.
b) Difference in acoustic centers is entered to Z coordinate (distance) with dipole radiators such as open baffle speakers.
Anyway, this method is not so good and accurate for off-axis simulation as dual channel method because requires accurate geometry simulation to support simulated rotation 0-180 deg with different speaker/driver types.
But it's totally okay if target is to measure and simulate just listening window and skip off-axis and all possibilities what it can offer.
So, cucicu did not miss any steps. He just needs some memory refreshing due to long time since the latest private discussion about correct measurement methods.
Last edited:
This method has been described by david L. RALPH.
Kimmo you know I never obtained decent listening results with that method ie for my taste.
I think that issue is keeping costant the distance. Mic to baffle but also the measurement chain of an hobbyst can have latency issues. So, it means that also in dual channel mode, IR may be generated/kept in a different time and also keeping the same window for all drivers can bring some latency.
Kimmo you know I never obtained decent listening results with that method ie for my taste.
I think that issue is keeping costant the distance. Mic to baffle but also the measurement chain of an hobbyst can have latency issues. So, it means that also in dual channel mode, IR may be generated/kept in a different time and also keeping the same window for all drivers can bring some latency.
^Dual channel measurement connection and mode eliminates latency issues except changes in speed of sound i.e. changes in temperature or "wind"/A.C. in your room between/during different measurements.
Measuring of distance from mic to rotation center is not always easy. I usually rotate speaker 90 deg, align mic and baffle surface, measure distance from mic to baffle edge and add distance from edge to center of driver to previous value. Then rotate back to 0 deg and start measuring sequence.
We've had discussion with dlr at parts express about differences of these two methods (with addition that single channel measurements are processed with minimum-phase extraction). I suppose we did not get clear consensus, but it doesn't change my confidence that dual channel method is universal and accurate. Single channel method with three measurements is more or less limited and inaccurate depending on speaker construction, driver types, target of simulation (are we simulating full space or narrower window close to listening axis) and how XO simulator calculates geometry of driver and baffle.
This has nothing to do with taste.
As long as you use VituixCAD 2 and simulate at least up to 90 deg, you better use dual channel connection and mode, measure physical distances from mic to rotation center of each driver by tape and enter differences to Z mm.
Exporting of frequency responses with Convert IR to FR tool takes few minutes per project, also with hundreds of measurements. This is no less than unbeatable system at the moment
Measuring of distance from mic to rotation center is not always easy. I usually rotate speaker 90 deg, align mic and baffle surface, measure distance from mic to baffle edge and add distance from edge to center of driver to previous value. Then rotate back to 0 deg and start measuring sequence.
We've had discussion with dlr at parts express about differences of these two methods (with addition that single channel measurements are processed with minimum-phase extraction). I suppose we did not get clear consensus, but it doesn't change my confidence that dual channel method is universal and accurate. Single channel method with three measurements is more or less limited and inaccurate depending on speaker construction, driver types, target of simulation (are we simulating full space or narrower window close to listening axis) and how XO simulator calculates geometry of driver and baffle.
This has nothing to do with taste.
As long as you use VituixCAD 2 and simulate at least up to 90 deg, you better use dual channel connection and mode, measure physical distances from mic to rotation center of each driver by tape and enter differences to Z mm.
Exporting of frequency responses with Convert IR to FR tool takes few minutes per project, also with hundreds of measurements. This is no less than unbeatable system at the moment
Last edited:
No. There is no latency of sound card. Just flying time and possible processing delay if you have semi-dual connection and reference channel is taken before DSP device, typically from output of sound card. Just flying time with full dual connection and mode when reference channel is taken from output of power amplifier.
You're probably confused by start margin of ARTA. ARTA locates IR so that measurement starts at sample #300. You have to ask from iMat why he has added 300 samples extra in the beginning of IR also while dual channel mode. That is not necessary while acoustical measurements, but some electrical cases might be different.
Sample #300 is also zero time point of single channel measurement.
Yes. Typically tweeter is 0 mm and woofer is -25 mm (if horizontal distance from W to listener is shorter).
^Ok. All measurements at 1000 mm from driver's baffle which is also Z-coordinate of rotation center. To simulator: tweeter Z=0mm, mid Z=-25mm and woofer Z=-25mm.
and all this was already explained
and all this was already explained
^That's okay (if you don't measure vertical plane) because angle error is typically no more that 4 deg at 100 cm.
But if you have planar/ribbon tweeter, you should measure also vertical plane to get correct power & DI. Then you better measure tweeter and mid from different elevation to get common axial response for tweeter in hor and ver planes.
But if you have planar/ribbon tweeter, you should measure also vertical plane to get correct power & DI. Then you better measure tweeter and mid from different elevation to get common axial response for tweeter in hor and ver planes.
I try to explain.
When measuring each driver individually, distance baffle to mic must be the same on axis with driver ie when measuring stepped baflle woofer, the z for it have to be - 25mm.
When measuring instead at middle height between M and T, not on driver axis but at middle height, and Mid is 25mm stepped forward, z for the mid has to be - 25mm again?
When measuring each driver individually, distance baffle to mic must be the same on axis with driver ie when measuring stepped baflle woofer, the z for it have to be - 25mm.
When measuring instead at middle height between M and T, not on driver axis but at middle height, and Mid is 25mm stepped forward, z for the mid has to be - 25mm again?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.