Member
Joined 2003
That was exactly my point though. Not so much that you will receive much more income through this process, but by removing the "free support" as an option, your efforts are focused on professionals who can provide more valuable input to drive development, and as well for newcomers who are at least serious enough about the hobby to be willing to pay for support or who feel the free support available by the community other than yourself is not sufficient. The repeat simple questions can receive repeat simple answers through the DIY community, as has been the way for many years, not just with software support but all aspects of loudspeaker design.
Anyway, it was just a suggestion, with the intent to dissuade you from removing VituixCAD as a "free for anyone" option by providing some alternative path that removes your need to constantly provide "first aid".
Member
Joined 2003
Yes, I would. However, I would also just move the mic a few inches and place on-axis with each driver regardless, it's not that hard to do so just do it anyway regardless of driver proximity.
Or rather enter +2 / -2 inches respectively for each driver (for Y).
But agree with Db, do try to measure each one separately on its own axis. Ten more minutes or so...
Yes, I'm going to remeasure. I chose not to mention (admit) that my tweeter is not round (it's the PTmini-6) AND I did do vertical measurements too. Thanks all.
RTF* 
This paragraph does not contain any exceptions, though it's possible to enter 0,0,0 to all drivers if also vertical plane is measured and measurement distance is close to simulation distance. In your case it's not measured if you followed instructions.
So no need to redo measurements. Just follow both measurement instructions and user manual.
Driver instances
Each driver instance added in the crossover can be provided with location relative to "design origin". Location is entered to Parameters grid. Design origin is typically perpendicular endpoint of listening axis on front baffle surface. X [mm] is horizontal coordinate of center point; negative to left and positive to right (mic/listener view). Y [mm] is vertical coordinate; negative down and positive up (mic/listener view). Z [mm] is horizontal distance coordinate; negative closer to mic and positive further from mic.This paragraph does not contain any exceptions, though it's possible to enter 0,0,0 to all drivers if also vertical plane is measured and measurement distance is close to simulation distance. In your case it's not measured if you followed instructions.
So no need to redo measurements. Just follow both measurement instructions and user manual.
Last edited:
Member
Joined 2003
Hmm..thinking about this, initial thought would be that including y axis offset for a driver that is not measured on-axis is actually introducing phase error to the result, but it is a small error, and including y axis offset would provide a more accurate vertical polar representation. I guess the thought process here is that with small drivers the difference in phase at listening distance of 2m+ given the mic location is insignificant.
In any case, my recommendation is to avoid the "exception" noted in the measurement document, and just follow the same measurement process with mic on axis with the driver regardless of driver size and proximity.
you still need to enter per driver the xyz , it is just the mic height that is not changed. you still measure 2 discrete drivers one at a time.
the xyz define the geometrical relationship of the drivers. so relative to your chosen design origin, which can be the half way between center points.
I did not see other mails addressing the same question, my mistake. Never the less my response is in line.
I was contemplating on this, triggered by my granson, who has developed a habit of asking while already knowing the answer or too lazy to figure it out himself. What we do is exactly that, instead of solving, ask him to think and read and figure out himself. It works.
And in occastion a small change in the text could add a lot of clarification. So a win-win.
Member
Joined 2003
New preliminary revision of Linux & Wine instructions is on the server: VituixCAD_in_Linux_v0.2.pdf
That is tested with Mint Mate and Cinnamon 64-bit, but not with other distros and 32-bit wine prefix. Should work on Ubuntu too. Please comment if there's something to fix.
I really cannot recommend this combination because curve antialiasing/smoothing does not work so all graphs and XO schematic looks just trash compared to Windows in real or virtual machine. Other problems are low speed and differences in font rendering. I know close to nothing about Linux so this problem can be on this side of the keyboard.
That is tested with Mint Mate and Cinnamon 64-bit, but not with other distros and 32-bit wine prefix. Should work on Ubuntu too. Please comment if there's something to fix.
I really cannot recommend this combination because curve antialiasing/smoothing does not work so all graphs and XO schematic looks just trash compared to Windows in real or virtual machine. Other problems are low speed and differences in font rendering. I know close to nothing about Linux so this problem can be on this side of the keyboard.
Member
Joined 2003
I used winetricks and got this message:
warning: This package (dotnet452) is broken in wine-6.0.3. Broken since 5.18. Upgrade to >=6.6. See https://bugs.winehq.org/show_bug.cgi?id=49897 for more info. Use --force to try anyway.
I was able to get it working by installing the latest Wine version from here:
https://wiki.winehq.org/Debian
warning: This package (dotnet452) is broken in wine-6.0.3. Broken since 5.18. Upgrade to >=6.6. See https://bugs.winehq.org/show_bug.cgi?id=49897 for more info. Use --force to try anyway.
I was able to get it working by installing the latest Wine version from here:
https://wiki.winehq.org/Debian
Few details related to fonts and mouse hovering adjusted. Refresh to see the latest version.
is this intended behavior or a bug, when we have a resistor out of the main circuit (ie not connected anywhere) and if you short a wire across it then suddenly all of the crossover graphs disappear from view...until you unshort it or delete it. Accidently did it and couldnt figure out why everything vanished from view.
^This was discussed few weeks ago. See: https://www.diyaudio.com/community/threads/vituixcad.307910/page-187#post-6995240
I will not sacrifice network parsong performance for detecting illegal circuits, but user can control this feature by following those tips.
I will not sacrifice network parsong performance for detecting illegal circuits, but user can control this feature by following those tips.
Maybe something interesting to look at.
I recently bumped into someone's Master Thesis for T/S parameter estimation by curve fitting, using method of steepest descent. For impedance as well phase of the impedance (real and im)
Meaning, no added mass or different compliance (volume) method is needed.
See; https://projekter.aau.dk/projekter/...elling_and_Crossover_Network_Optimization.pdf
(also attached as .pdf)
Page 49, chapter 4 (pdf page 57 of 141) as well as page 57 chapter 5.3 (pdf page 65 of 141)
Basically all is needed is the impedance measurement as well as the Sd (aka diameter)
Unfortunately in this report this has not been tested against other methods.
Also for T/S parameters we could lower the bandwidth drastically by taking max freq 4x Fs and min freq Fs/4 (or there abouts)
Instead of going all the way up to 5kHz, which is not a such a good idea since the model can be very tricky above 300Hz or so depending on demodulation rings and such. (which results in errors).
Personally I think there are some other additional things that can be used for optimizing the results.
With some simple math (or even just using data points) the max impedance value can be determined as well as the Fs.
Depending on what method/standard one is using, the Re can be given as well.
Which makes curve fitting a lot easier as well.
I have no idea how easy this is to implement, but since VituixCAD already uses some kind of curve fitting possibilities, I thought you might find it interesting?
*btw this is only about the curve fitting method, the rest of the report is a bit meh.
I recently bumped into someone's Master Thesis for T/S parameter estimation by curve fitting, using method of steepest descent. For impedance as well phase of the impedance (real and im)
Meaning, no added mass or different compliance (volume) method is needed.
See; https://projekter.aau.dk/projekter/...elling_and_Crossover_Network_Optimization.pdf
(also attached as .pdf)
Page 49, chapter 4 (pdf page 57 of 141) as well as page 57 chapter 5.3 (pdf page 65 of 141)
Basically all is needed is the impedance measurement as well as the Sd (aka diameter)
Unfortunately in this report this has not been tested against other methods.
Also for T/S parameters we could lower the bandwidth drastically by taking max freq 4x Fs and min freq Fs/4 (or there abouts)
Instead of going all the way up to 5kHz, which is not a such a good idea since the model can be very tricky above 300Hz or so depending on demodulation rings and such. (which results in errors).
Personally I think there are some other additional things that can be used for optimizing the results.
With some simple math (or even just using data points) the max impedance value can be determined as well as the Fs.
Depending on what method/standard one is using, the Re can be given as well.
Which makes curve fitting a lot easier as well.
I have no idea how easy this is to implement, but since VituixCAD already uses some kind of curve fitting possibilities, I thought you might find it interesting?
*btw this is only about the curve fitting method, the rest of the report is a bit meh.
Last edited:
