Yes, but i am using many freq resp curves which i load many times while optimizing the crossover, so each response curve needs to have correct phase so i can keep up the work flow, otherwise i need to redo all settings for each time i load a response curve
May I ask why you need many freq resp curves and load them many times? What is the difference in "acoustic reality" between those responses?
I measure the system from many different angles, say at 0 deg and 45 deg up and to the sides. Then i load the responses from each angle and tune the crossover, then i keep doing this til the crossover works good at all these angles. I would like to see all curves at once but that is not possible as far as i know.
^Some simulators are able to calculate and show off-axis angles at the same time with axial. Few of those calculate also power response and directivity index to half of full space by simulated off-axis responses. Requirement in practice is that off-axis measurements of each driver include measured timing difference (compared to reference measurement) to different off-axis angles because minimum phase extraction normalizes=hides timing differences between mechanically different radiators and causes also polarity errors with dipole and leaking radiators. In addition, investigation of timing differences (not included in minimum phase responses) is extremely slow work and total pain in the a** if more than few off-axis measurements.
So your design philosophy and targets seem to be at different (higher) level than your measurement gear or method and selected XO simulator.
So your design philosophy and targets seem to be at different (higher) level than your measurement gear or method and selected XO simulator.
so i need to work this out by using minimum phase curves instead and post process each response to contain correct phase / time difference
Of course if timing is not locked with reference channel. Also minimum phase extraction causes drifting of timing especially with IR gating because LF response could vary quite randomly without changing measurement setup.
celef,
Xsim is really fantastic and also free, but for your specific task one can wonder if you also know about free VituixCad because its off axis features should cover what you are looking for.
links: Software and https://www.diyaudio.com/forums/software-tools/307910-vituixcad-152.html#post5930295
Xsim is really fantastic and also free, but for your specific task one can wonder if you also know about free VituixCad because its off axis features should cover what you are looking for.
links: Software and https://www.diyaudio.com/forums/software-tools/307910-vituixcad-152.html#post5930295
^I guess VituixCAD or compatible does not solve the problem if user is forced to use minimum phase extraction. Then actual problem is probably single channel measurement gear without I/O in the same box. For example USB microphone which is asynchronous random timing device by nature. Not recommended for this hobby.
Hm, I just find out that this export file did not contain phase, is there any way to activate phase info too?
Hi Bill,
I was wondering if you could add a feature in which the step response bandwidth can be modified? For example, I suppose the step response input has infinite bandwidth. Since our hearing is limited to about 20KHz, it would be nice to see the step response that is closer to our hearing bandwidth.
Thanks.
I was wondering if you could add a feature in which the step response bandwidth can be modified? For example, I suppose the step response input has infinite bandwidth. Since our hearing is limited to about 20KHz, it would be nice to see the step response that is closer to our hearing bandwidth.
Thanks.
Hi All,
I could use some help in trying to explain a paradox ( apologies up-front for the following ramble-on ).
Pre-ample & BackGrounding :
I've attached a zipped XSim file that encloses 5 of Zilch's 2010 "econowave" projects ( with .frd + .zma files included for each ) loaded from his classic Flex Your PCD Mettle over at TechTalk
Those here who use "curve-fitting" to arrive at a reasonable Z-Offset for crossover design will know ( I think ) how to use the included files ( they're all arranged vertically in order of groupings ) .
I've already determined that these files from Zilch would all have a Z-offset of about 4" ( + or - a bit of an inch ) between the de250 driver ( on a 152i horn ) and one of 5 different 12" woofers.
- The offsets were arrived at by the classic curve-fitting ( visual ) method.
- I focused on matching the curves above 1K for best fit ( since this is the area of intended crossover ).
- Zilch apparently focused on a different FR range since he typically arrived at a 2.5" figure for off-set.
- Please go through them to see if a more accurate Z-Offset is obtainable.
- Zilch used CLIO along with Jeff Bagby's PCD program ( back in the day ) to arrive at his offsets.
- Apparently CLIO exports a form of some minimum phase file since Zilch never went through the process of extracting minimum phase ( to then enter into PCD ).
- The acquisition software for the questionable frd's is REW.
- When REW exports an IR into a frd it typically "zeros" the beginning of the file ( removing all delay + therefore somewhat mimicking a minimum phase file).
Observations :
If a user "curve-fits" the last set of frd's + zma's ( which are for a Peerless DFM2535 driver on a 152i Clone horn and a modified JBL 2204h 12" woofer one ends up with @ zero offset for the Z-Axis ( as "best-fit" ) .
This makes no sense for this 12" woofer and horn/driver combo ( IME ).
- While the MMS is heavier than normal ( by 35 grams ) the Voice-Coils "Le" is in the ballpark of the stock 2204H, so I can't see those 2 factors combining to erase 4 inches of offset.
Question :
I'm wondering what's happened ( in the curve-fitting process ) to cause 4" of offset to go missing?
EarlK
PS: the "paradox" files aren't mine, but I'm assured mic positions ( and relative levels ) remained static for the three pertinent measurements that were used to create the .frds.
I could use some help in trying to explain a paradox ( apologies up-front for the following ramble-on ).
Pre-ample & BackGrounding :
I've attached a zipped XSim file that encloses 5 of Zilch's 2010 "econowave" projects ( with .frd + .zma files included for each ) loaded from his classic Flex Your PCD Mettle over at TechTalk
Those here who use "curve-fitting" to arrive at a reasonable Z-Offset for crossover design will know ( I think ) how to use the included files ( they're all arranged vertically in order of groupings ) .
I've already determined that these files from Zilch would all have a Z-offset of about 4" ( + or - a bit of an inch ) between the de250 driver ( on a 152i horn ) and one of 5 different 12" woofers.
- The offsets were arrived at by the classic curve-fitting ( visual ) method.
- I focused on matching the curves above 1K for best fit ( since this is the area of intended crossover ).
- Zilch apparently focused on a different FR range since he typically arrived at a 2.5" figure for off-set.
- Please go through them to see if a more accurate Z-Offset is obtainable.
- Zilch used CLIO along with Jeff Bagby's PCD program ( back in the day ) to arrive at his offsets.
- Apparently CLIO exports a form of some minimum phase file since Zilch never went through the process of extracting minimum phase ( to then enter into PCD ).
- The acquisition software for the questionable frd's is REW.
- When REW exports an IR into a frd it typically "zeros" the beginning of the file ( removing all delay + therefore somewhat mimicking a minimum phase file).
Observations :
If a user "curve-fits" the last set of frd's + zma's ( which are for a Peerless DFM2535 driver on a 152i Clone horn and a modified JBL 2204h 12" woofer one ends up with @ zero offset for the Z-Axis ( as "best-fit" ) .
This makes no sense for this 12" woofer and horn/driver combo ( IME ).
- While the MMS is heavier than normal ( by 35 grams ) the Voice-Coils "Le" is in the ballpark of the stock 2204H, so I can't see those 2 factors combining to erase 4 inches of offset.
Question :
I'm wondering what's happened ( in the curve-fitting process ) to cause 4" of offset to go missing?
EarlK
PS: the "paradox" files aren't mine, but I'm assured mic positions ( and relative levels ) remained static for the three pertinent measurements that were used to create the .frds.
Last edited:
Everything in Xsim is limited to 20kHz, no calculations or effects included beyond that (other than tails when doing Hilbert phase derivations).