I thought I'd post an update. I decided to export minimum phase for my off axis measurements, and scale the woofer down by 6db and impedance doubled. I then put in the two M's and single tweeter with the correct offsets.
Now my 0 - 90deg measurements model correctly
though my power response and di now look a bit crappy compared to just zero deg modeling, but I guess that is the point of using the extra measurements
Tony.
Now my 0 - 90deg measurements model correctly
though my power response and di now look a bit crappy compared to just zero deg modeling, but I guess that is the point of using the extra measurements Tony.
Attachments
...though my power response and di now look a bit crappy compared to just zero deg modeling, but I guess that is the point of using the extra measurements
It's quite educational to notice that design by axial response only is fundamentally very sh*t method which relies on luck or long experience how to avoid e.g. stuffy but harsh sound. One of the main reasons why VituixCAD exists is to offer these shocking experiences
^Project files are not compatible because free form and block ladder crossover networks are totally different. I'll keep most of my 100+ old projects in version 1 only, and create new projects to version 2. Exceptions are active commercial projects and demo projects which I've converted manually for version 2.
Hi,
is there any change to implement the use of relative paths (beside from absolute) for measurement files - or is this already possible?
I frequent rename or/and move old project directories (or send project directories to friends), so the use of absolute paths for the measurement files isn't the best option.
Would be really nice if VC could detect (or if there is an option "Save project with relative path" to activate) ./ and ../ and VC then use the current VC-project file directory as starting point for a relative path search.
regards, Armin
is there any change to implement the use of relative paths (beside from absolute) for measurement files - or is this already possible?
I frequent rename or/and move old project directories (or send project directories to friends), so the use of absolute paths for the measurement files isn't the best option.
Would be really nice if VC could detect (or if there is an option "Save project with relative path" to activate) ./ and ../ and VC then use the current VC-project file directory as starting point for a relative path search.
regards, Armin
You probably mean cousin directories such as "../.." because support for relative paths is already implemented below main directory i.e. where project file (vxp) is located.
Rooted full path is saved to project file (vxp) only if auxiliary files such as frequency responses, impedance response or transfer function file is not located in the main directory or below.
I'm using this kind of directory system:
Documents\VituixCAD\Projects\ProjectID vxp, vxb, vxm files are located here
ProjectID\Far Far field measurements and txt/frd exports
ProjectID\Near Near field measurements and txt/frd exports
ProjectID\Merged Merged near+far saved here
ProjectID\Impedance Impedance measurement are here
few other directories such as Export, Images, Drawings are usually needed for complete project design and publish.
Cabinet impact response can be exported to Merger or Near directory.
This hierarchy is easy to zip and backup/archive, restore and transfer to other users.
thanks for the quick reply...
So it is impossible to start a project from a stick? First copy the project-directory from the stick to the VC main directory on the other PC/Notebook then it will work?
regards, Armin
This only works if I use the VC main directory? My loudspeaker projects are on a separate partition. Think I could use windows symlinks to redirect the VC main directory.
So it is impossible to start a project from a stick? First copy the project-directory from the stick to the VC main directory on the other PC/Notebook then it will work?
regards, Armin
^ProjectID (where vxp file is located) is "main directory" in my previous list. It could be on any partition or drive, for example USB drive works okay.
Edit: Just remember that Save as -command saves only vxp file to different location or name. Response files stay in their original places and paths in vxp file change to absolute/rooted. Move main directory with it's subdirectories if you want to move project with measurement files to another partition/drive or some cousin directory.
Edit: Just remember that Save as -command saves only vxp file to different location or name. Response files stay in their original places and paths in vxp file change to absolute/rooted. Move main directory with it's subdirectories if you want to move project with measurement files to another partition/drive or some cousin directory.
Last edited:
Bugger.
Conversion from 1 to 2 would be nice programming challenge, but I'm afraid that you all have converted projects manually before I have even started. Our summer holiday season delays everything.Now my 0 - 90deg measurements model correctly
It's quite educational to notice that design by axial response only is fundamentally very sh*t method which relies on luck or long experience how to avoid e.g. stuffy but harsh sound. One of the main reasons why VituixCAD exists is to offer these shocking experiences
It was quite mindblowing to model few earlier designs with angled measurements in VituixCAD. There are numerous designs from respected designers that have either luck or additional expertise (or actual polar simulations) but looks like they are designed with only axial response, but still show quite good polar response and continuous DI. Then there are those that do not
Mostly every 1" dome + 8" woofer and MTM designs with dome (i.e high XO point).I have designed ~8 speakers with VituixCAD since 2014 (yes, early adopter) and could not live without it. It packs all mandatory features that commercial equivalents, but more user friendly interface and dashboard that gives you full control of your design at a glance. I haven't tested these new features nor version 2.x but the development looks very promising.
Yes I have been struggling to work out what I should be aiming for. I read a bit on power response and from what I can gather a downward tilt is what is generally preferred, Toole found that sharp dips in the power response didn't seem to affect listener peference, but I'm guessing my first posted graph was definitely not good
This is what I have come up with at the moment. Not as smooth on axis, but I guess that is the thing with speakers, everything is a compromise. I got something that was looking quite good (optimizing for power response) and then I realised it had effectively made the tweeter acoustic response about 1600Hz crossover, which is lower than I think I can go with these speakers.
Also thinking I probably need to go with full 180 deg of measurements, and proper set of individual measurements for each driver in the MTM, rather than the M's combined on the tweeter axis.
The second attachment is how the current (8 yo) implementation of the crossover models. I probably should export the off axis simulated results and compare to my actual off actual measuements with the crossover in place as a sanity check. I'm not sure that the new one will be an improvement...
Tony.
This is what I have come up with at the moment. Not as smooth on axis, but I guess that is the thing with speakers, everything is a compromise. I got something that was looking quite good (optimizing for power response) and then I realised it had effectively made the tweeter acoustic response about 1600Hz crossover, which is lower than I think I can go with these speakers.
Also thinking I probably need to go with full 180 deg of measurements, and proper set of individual measurements for each driver in the MTM, rather than the M's combined on the tweeter axis.
The second attachment is how the current (8 yo) implementation of the crossover models. I probably should export the off axis simulated results and compare to my actual off actual measuements with the crossover in place as a sanity check. I'm not sure that the new one will be an improvement...
Tony.
Attachments
Last edited:
Hi Tony,
A couple of things to think about are the room signature and your hearing preferences. If you are measuring speakers that will ultimately be in one room, you would want to take into account the room signature. There may be reinforcement of certain frequencies or more absorption of others. If you are looking at speakers that can be moved from place to place, then targeting "ideal" curves makes more sense. I personally find that I prefer a flat frequency response on axis in a room that is well treated. Of course, I look for a smooth power response and it will shift downward with increasing frequency as beaming occurs at the higher frequencies. Smoother is generally better but as you say, there will always be compromises. There will be an area where you don't see much of this off axis effect at lower frequencies where no significant beaming is occurring. You may be able to smooth off axis effects with baffle treatments and rounded edges, hence, smoothing your power response. This article kind of addresses much of this: https://www.harman.com/sites/default/files/LoudspeakersandRoomsPt2_0.pdf. As you know, you are limited by driver limitations so like you said, you need to be cautious to not design a crossover that the tweeter cannot handle at power levels that it will see, realizing that distortion puts additional strain on the tweeter.
Jay
A couple of things to think about are the room signature and your hearing preferences. If you are measuring speakers that will ultimately be in one room, you would want to take into account the room signature. There may be reinforcement of certain frequencies or more absorption of others. If you are looking at speakers that can be moved from place to place, then targeting "ideal" curves makes more sense. I personally find that I prefer a flat frequency response on axis in a room that is well treated. Of course, I look for a smooth power response and it will shift downward with increasing frequency as beaming occurs at the higher frequencies. Smoother is generally better but as you say, there will always be compromises. There will be an area where you don't see much of this off axis effect at lower frequencies where no significant beaming is occurring. You may be able to smooth off axis effects with baffle treatments and rounded edges, hence, smoothing your power response. This article kind of addresses much of this: https://www.harman.com/sites/default/files/LoudspeakersandRoomsPt2_0.pdf. As you know, you are limited by driver limitations so like you said, you need to be cautious to not design a crossover that the tweeter cannot handle at power levels that it will see, realizing that distortion puts additional strain on the tweeter.
Jay
One more thing, Tony. Though the overall power response is a good measure to look at, also of importance, is that off axis measurements show no significant aberrant deviations from the previous measurement.
Regarding Dr. Toole and his team's work, remember that they primarily look at listener preference and in designing your own speakers, the most important listener is the audience who will be listening to the speaker.
Regarding Dr. Toole and his team's work, remember that they primarily look at listener preference and in designing your own speakers, the most important listener is the audience who will be listening to the speaker.
Just my opinion, and it may go against what others think, but personally I would not sacrifice and distort the on axis response to meet a power response target that the drivers are not capable of, especially at higher frequencies above say 1Khz.Yes I have been struggling to work out what I should be aiming for. I read a bit on power response and from what I can gather a downward tilt is what is generally preferred, Toole found that sharp dips in the power response didn't seem to affect listener peference, but I'm guessing my first posted graph was definitely not good
This is what I have come up with at the moment. Not as smooth on axis, but I guess that is the thing with speakers, everything is a compromise. I got something that was looking quite good (optimizing for power response) and then I realised it had effectively made the tweeter acoustic response about 1600Hz crossover, which is lower than I think I can go with these speakers.
Tools findings are clear - listener axis response is the primary ranking variable in what sounds best, power response and off axis response is a distant secondary factor with much smaller effect, especially at high frequencies.
So it is very debatable whether you will hear a moderate change in the power response towards a more "ideal" power response, but you WILL hear the sacrifice in on axis response made to do that.
By making the on axis response worse to make the power response better, the overall perceived response will almost certainly be worse since you're making a primary variable worse to improve a secondary variable.
The only things you can really do in a crossover to improve the power response without simultaneously sacrificing the on axis response is to shift the crossover frequencies, increase the crossover slopes, and/or change to a different crossover slope with more favourable phase relationships, for example even instead of odd, which can improve the vertical polar response.
In my speakers crossover redesign I moved the crossover frequency down from 4Khz to 3Khz, changed the slopes from 3rd order to 4th order and (with an all-pass filter for physical driver offset) corrected the phase response for accurate on axis phase tracking, and the improvement in off axis response and overall room response was obvious even to the ear when sitting to the sides of the room without even having to do any off axis measurements, (I will do them one day) but at the same time I managed to improve the on axis response considerably as well. End result - they sound MUCH better.
Of course you can only shift crossover frequencies if the drivers allow it - in my case 3Khz is as low as I can go with a ribbon tweeter even with 4th order filter. Below that would be going out of the comfortable range of the tweeter.
However any change to improve the power response that would make the on axis response worse is a sign that the power response change needs to be achieved some other way, by changing crossover frequencies, drivers, baffles overall design etc...
So its great that Vituixcad displays the off axis response and power response, but I wouldn't necessarily use that information to optimise the crossover design beyond crossover frequency and slope selection - at that point its often too late and different design choices or driver selections should have been made earlier in the design. I definitely wouldn't let the optimiser loose making random changes to try to optimise the power response.
Last edited:
Or just opposite i.e. you could be wrong. We can debate for ever what is primary and what are secondary variables by adapting reality to fit arguments we believe or our limited experience. Everybody should be aware that power response also could be primary variable, and axial response something we never hear. Everything depends on speaker concept, environment and setup.
Could we just agree that both should be very good because "no compromises" works also in reflecting environment, listening from far or totally off-axis.
If we have to make some compromise, we just adjust to direction what works best in owner's setup. That person could be designer - or not.
To my experience, smooth polar response, continuous DI and carefully chosen power response tilt is far better than perfect axial response. Based on those eight speakers I have designed with 180 degree measurements both hor and ver, I am happy to sacrifice axial response to get proper polar response, because that's how speaker actually works i.e radiates in 3D space.
Or we could look at the decades of research by people like Floyd Tool and Sean Olive who have extensively studied this using rigorously set up listening tests to evaluate what factors do and do not correlate with listener preferences and perception of sound quality.
Their conclusion is that while the power response has some effect it is a relatively minor secondary factor compared to the listener axis response. (Notice I said listener axis not on axis)
In fact listener axis response was far and away the single most important factor.
This is their conclusions not mine. (Although my own experiences agree with their findings)
If you have done your own decades of independent carefully controlled listener research that would rival what they have done then I'm happy to see your data.
Otherwise I'll defer to their expertise in this matter.
I'm sure you must have heard of the sliding window interpretation of how our hearing works ?
At low frequencies below about 300Hz the integration period is long relative to path lengths of reflections so we perceive basically the steady state response which is the sum of direct and all reflections. This is particularly so in the bass but it goes a bit higher than the bass.
At treble frequencies above about 3Khz our perception of frequency response is almost entirely the direct signal from the speaker (I'll call it listener axis, since it may be a little off axis on purpose) with no real contribution from room reflections, this is because the integration time is much shorter than even the delay of the first sidewall reflection, which is processed by the brain as an echo and does not contribute to the perceived frequency response of the signal.
At these high frequencies, for a person sitting on the primary listening axis power response and off axis response only serve to alter secondary effects like the overall room ambience and liveliness.
Between about 300Hz to 3Khz there is a "grey area" where the sliding window is transitioning where at the bottom end power response has significant effect and at the top end it does not. There is no sharp cutoff and this is one reason why midrange is so hard to get right.
Of course speakers are not always listened to from one point in the room so it is beneficial to have a good polar response as well, but this can only be achieved with the physical design of the speakers like driver selection and selection of crossover frequencies - IMO it is wrong to try to improve the power response of a speaker by making crossover adjustments that simultaneously make the listener axis response less flat.
It's the same reason why automatic room equalisers that measure the room response using a steady state measurement (usually pink noise) and then attempt to EQ this to flat (or even to a "house curve" which slopes down) fail miserably, because they don't measure the response in the same way a human perceives it.