This does make a little bit more work for the user though, as there is no way to visually display or keep track of which part of the resistance belongs to the coil and which part belongs to an actual physical series resistor to make up the total value.
Doesn't matter for simulating but matters when you go to order the parts.
In my crossovers which I recently designed and built using vituixcad (yes they are up and running now.
) the way I dealt with this is when I went to order the parts I created a spreadsheet from the component list export, and for all coils I first decided on which exact coil I was going to buy, noted its series resistance on the spreadsheet, subtracted this from the resistance needed in the simulation to get the value for the physical resistor which I actually needed to order to go with it.As I used a number of notch filters all of which have physical LC and R this was necessary, and even with a spreadsheet it did get a bit confusing to keep track of the distinction between total simulated R required (coil + R) and physical R (alone) required.
Actually in reality before this I also did a second pass revision of the design - with "real" resistance of practical coils that I can buy plugged back into the simulation then re-optimised to get the original target responses again.
This is mainly in the case of the all-pass, lowpass and highpass sections where I did not want any deliberate resistance in the coils but some must realistically exist and does alter the response, vs notch filters where series resistance was desired and the desired value was easy to reach by just adding additional series resistance to the coils own resistance.
Last edited:
Doesn't matter for simulating but matters when you go to order the parts.
Yes. That is the latest moment to decide how to split resistance to internal and external. Availability and price of coils are helpful information. Sometimes coil selection with known internal resistance is obvious during simulation, but deciding of internal resistance before crossover optimizing and shopping is not always the best idea. Therefore separate parameter for internal R is not mandatory at all.
^Probably filenames are not valid. Unfortunately this isn't explained directly in Frequency responses section of user manual. Options section contains:
Measurement preparations.pdf document contains more detailed information, especially for ARTA with (manual) turning table:
If you measure axial responses only without off-axis with manual or automatic turning table, you can name exported frequency response files as Drivername_hor 0.txt or .frd. "deg" is not required by VituixCAD.
Measurement preparations.pdf document contains more detailed information, especially for ARTA with (manual) turning table:
If you measure axial responses only without off-axis with manual or automatic turning table, you can name exported frequency response files as Drivername_hor 0.txt or .frd. "deg" is not required by VituixCAD.
Some commercial speakers designed with VituixCAD:
Taipuu Speakers
Simulation data @kimmosaunisto.net
2-ways and 3-way are active with IIR filters. FIR option was also available but we're removing it from preset menu.
Passive crossover of 4-way will be assembled soon.
Taipuu Speakers
Simulation data @kimmosaunisto.net
2-ways and 3-way are active with IIR filters. FIR option was also available but we're removing it from preset menu.
Passive crossover of 4-way will be assembled soon.
I thought so. In my opinion, BMS has the best available coaxial speakers on the market. I currently have 5-inch coaxial monitors.
Very nice work Kimmo.
Whole AES E-Library has been available, Acoustics: Sound Fields and Transducers by Beranek & Mellow and everything I can found from the internet. For example some MIT lessons and JOS @ccrma.stanford.edu.
Few practical books such as Loudspeaker Design Cookbook and our national reference TEE ITSE Hifikaiuttimia by P. Tuomela.
Few practical books such as Loudspeaker Design Cookbook and our national reference TEE ITSE Hifikaiuttimia by P. Tuomela.
I have a question on how best to go about measuring a speaker with multiple drivers per way, the design is basically a WMTW like
is it better to measure the ways together or separately?
I was thinking that if you measure each driver separately then you would enter the physical location of each individual driver and enter it as a 2 drivers whereas if you were to measure the drivers together then you would estimate the physical location as the mid point between the 2 drivers and enter it as a single driver
Obviously the latter is less work because you cut the no of measurements in half so that would be nice but I'm not sure if it will still produce a good result (I'd guess it should).
Code:
T
W W
M Mis it better to measure the ways together or separately?
I was thinking that if you measure each driver separately then you would enter the physical location of each individual driver and enter it as a 2 drivers whereas if you were to measure the drivers together then you would estimate the physical location as the mid point between the 2 drivers and enter it as a single driver
Obviously the latter is less work because you cut the no of measurements in half so that would be nice but I'm not sure if it will still produce a good result (I'd guess it should).
Last edited:
Drivers and ways are measured separately. At least one woofer in a group should be measured separately. The other can be skipped (and use same measurements) if difference in baffle diffraction is not significant. Near field measurement is trickier if both woofers are in the same cabinet volume because both woofers should be connected to power amp to get correct response at low freq. I've damped the other woofer with thick pillow to prevent excessive sound leak.
This document was originally written for WWMTMWW. Should be quite valid to your project.
Last edited:
Thanks for the link to the doc
I got my words mixed up in my previous post, what I meant to say was "drivers" not "ways". i.e. if measuring say 0-90 degrees horizontally, is it better to measure each driver separately and enter individual x-y offsets or measure them both playing together and estimate the x-y position as the midpoint? I take it from your doc that, given the baffle effects are the same in this case, measuring a single driver should be sufficient.
I got my words mixed up in my previous post, what I meant to say was "drivers" not "ways". i.e. if measuring say 0-90 degrees horizontally, is it better to measure each driver separately and enter individual x-y offsets or measure them both playing together and estimate the x-y position as the midpoint? I take it from your doc that, given the baffle effects are the same in this case, measuring a single driver should be sufficient.
Noticed a minor bug - I think it has been there for a while, but I keep forgetting to report it.
On the crossover tab at the bottom you have "Way X settings" that include Gain, Delay, Invert and Enabled.
Of these, I've noticed the check boxes for Gain and Delay do not work - the value entered in the adjacent value entry box always applies to the results regardless of the checkbox state on or off.
On the crossover tab at the bottom you have "Way X settings" that include Gain, Delay, Invert and Enabled.
Of these, I've noticed the check boxes for Gain and Delay do not work - the value entered in the adjacent value entry box always applies to the results regardless of the checkbox state on or off.
Those two checkboxes are for Optimizer only - as tooltip tries to tell. Active Gain and Delay can be optimized automatically just like filter block parameters.
The latest addition (rev 1.1.29.0) is power dissipation analysis. Graph shows power spectrum of all (enabled) resistors in passive filter. Cursor shows part# and resistance value. See changelog.
Spectrum of crest factor is flat. Power drops to 1/4 by entering 6 dB.
Pink noise option emulates spectrum of music. Response is square root of 1st order low pass, giving slope of -3 dB/oct. Average corner frequency is about 2 kHz, but significant variation exists. Initial/default value is 3 kHz.
An externally hosted image should be here but it was not working when we last tested it.
Spectrum of crest factor is flat. Power drops to 1/4 by entering 6 dB.
Pink noise option emulates spectrum of music. Response is square root of 1st order low pass, giving slope of -3 dB/oct. Average corner frequency is about 2 kHz, but significant variation exists. Initial/default value is 3 kHz.
Many thanks for continuing to add more features to your great software.
It seems a bit cheeky to ask but having seen your power dissipation display above, I wonder if it is possible to give a similar estimation for the drivers. I quite often end up doing calculations on the power dissipation in tweeters which are often padded down quite a lot to make sure they are OK with the system power.
Many thanks,
Andrew
It seems a bit cheeky to ask but having seen your power dissipation display above, I wonder if it is possible to give a similar estimation for the drivers. I quite often end up doing calculations on the power dissipation in tweeters which are often padded down quite a lot to make sure they are OK with the system power.
Many thanks,
Andrew
^Power dissipation of drivers wouldn't be a problem to add but I suppose resistances and drivers can't be visible at the same time due to different Y scaling.
How padding down and system power are related? Is your typical system semi-active, and padding is for reducing noise of amplifiers due to very high sensitivity of tweeter? Or do you use shunt resistor to reduce and equalize impedance? Level tuning possibilities with fully passive crossover are very limited - within less than 1 dB.
How padding down and system power are related? Is your typical system semi-active, and padding is for reducing noise of amplifiers due to very high sensitivity of tweeter? Or do you use shunt resistor to reduce and equalize impedance? Level tuning possibilities with fully passive crossover are very limited - within less than 1 dB.