This is quite subjective process. Better to stay above 1/time_window to get some frequency resolution to time windowed far field response. For example if right side of time window was 4.2 ms while IR to FR conversion, transition Hz should be above 1/4.2ms=240Hz.
Near field measurement is valid i.e. possible to convert to far field with baffle effect response when frequency is below 344/pi/Dd (see user manual page 37). For example 12" cone with Sd=531cm^2, maximum transition frequency is 344/pi/0.26m=421Hz.
Initial transition frequency could be logaritmic average of those two = 10^((log(240)+log(421))/2)=318Hz. I have usually merged 12" at ~350 Hz.
Thanks Kimmosto, nice explanation with examples, I can make myself an excel with that
One more question if you don't mind (That hide sub-questions) :
In VituixCAD Merger what would be the best option for low frequency measurements that have been taken with the driver on an IEC baffle using near field? I don't even know if that is considered 2pi
-No Baffle Loss would not be accurate I suppose because it's not an infinite baffle. (but it's near-field so...)
- Unfortunately I do not understand the default option of sphere baffle step even after reading the tooltip, is it for 4pi measurements?
I could simulate IEC Baffle Diffraction and include that, would that be the best to use, but do we need that when you do near-field ?
I'm confused on the best option there, if it was a test I would answer none of the above
Last edited:
Why would you measure anything in IEC baffle? That is for comparing drivers at mid...high frequencies in some standard manner. If you have already selected drivers by experience, recommendation or datasheet, IEC baffle is not needed (unless final design of your speaker is dipole in IEC baffle).
Drivers should be measured in project cabinet or mechanically and acoustically compatible prototype to get valid result for crossover design.
Meaning of near field measurement(s) is to produce quasi far field responses - typically to free full space if anechoic or large free space is not available i.e. near field meas. for LF and time windowing for HF is the best option.
Cabinet's "external effect" is simulated with Diffraction tool and diffraction response (without directivity, 'Half space response' field empty) is exported. That response is loaded to Merger tool with near field responses, and merged with far field HF responses to 0-180 deg. So 'Diffraction response' is correct option in Merger for typical boxed speakers.
P.S. Excel is not needed in this process. Merger tool shows maximum transition frequency with red background. Your playground is usually about 1/2 octaves below that point.
Example, BMS 12S305 merged at 367 Hz
An externally hosted image should be here but it was not working when we last tested it.
Thank you Kimmosto, I see better how it work now
The main reason I planned to test the drivers on a IEC baffle is because I was told it would allow me have cleaner frequencies responses to use in projecting cabinet shape & position of the driver, maybe so I don't have as many prototypes to make, but as I have no experience I tend to not have a very clear picture, the quote below from ScottG will explain in it better.
^I would not purchase driver with unknown half space response. IEC baffle is not needed if reliable datasheet with at least axial response (to 2pi) and possible few off-axis responses is available.
In addition, also IEC baffle is non-ideal diffracting component which requires anechoic or time windowing and calculated compensation to give "half space response in infinite baffle" down to 20 Hz. I wouldn't perform such measurement and calculation procedure to get few responses which are just preliminary information - not directly useful for final design.
As already mentioned in some other thread on diyaudio, diffraction feature of speaker enclosure is directivity component. Possiblities to shape axial response without ruining something in directivity are very limited. Usually you can compensate some part of minor cone edge resonance with diffraction hump but that's it. Focusing to axial response and shaping it with diffraction to on-axis would sound ancient design culture (before mid 80s').
I prefer strategy that enclosure is quite universal directivity component suitable for any decent driver with the same size, and crossover adapts response and directivity features of drivers and enclosure to reach design target.
In addition, also IEC baffle is non-ideal diffracting component which requires anechoic or time windowing and calculated compensation to give "half space response in infinite baffle" down to 20 Hz. I wouldn't perform such measurement and calculation procedure to get few responses which are just preliminary information - not directly useful for final design.
As already mentioned in some other thread on diyaudio, diffraction feature of speaker enclosure is directivity component. Possiblities to shape axial response without ruining something in directivity are very limited. Usually you can compensate some part of minor cone edge resonance with diffraction hump but that's it. Focusing to axial response and shaping it with diffraction to on-axis would sound ancient design culture (before mid 80s').
I prefer strategy that enclosure is quite universal directivity component suitable for any decent driver with the same size, and crossover adapts response and directivity features of drivers and enclosure to reach design target.
Last edited:
I think I understand why IEC is not that helpful to the cabinet design,
After doing DATS measurements some drivers where not very accurate to the announced specs and I was just under the assumption that I could not trust very much in the specs and measurements in the datasheets because of that results, but doing some makeshift measurements with cardboard I can see that I am close enough.
Maybe it isn't that big of a deal and I'll try to make them work
Skipping that IEC baffle would be a relief considering I had to make one that I could disassemble between use (apartment)
I think I'll burn some incense, light a candle and proceed with a cabinet prototype and rotating base
Thanks!
I wouldn't do that either because I wouldn't make front baffle with diffractive sharp edges and large area compared to surface of direct radiators. That kind of classic/vintage box is random off-axis response generator rather than well-behaving "universal" directivity component.
Sorry that I haven't read many messages here so impossible to know what is going on
I wouldn't do that either because I wouldn't make front baffle with diffractive sharp edges and large area compared to surface of direct radiators. That kind of classic/vintage box is random off-axis response generator rather than well-behaving "universal" directivity component.
Sorry that I haven't read many messages here so impossible to know what is going on
I understand that you are not going to read it all.
What is going on is I bought myself the Gradient AXP-08 (Woofer) and Fostex FF85WK (Mid-High)
I worked myself over to learn the procedure of measurement with REW, I initially tested the principles with cardboard and corrected the mistakes I was making with the use of that software.
The result (considering cardboard) is that the Fostex measurement isn't very far from the manufacturer datasheet at least for the mid upper range, the Gradient look similar also on that range but it isn't where it will be used.
But the upper treble of the Fostex don't satify me so a new Radian LT2 ribbon is on the way to satisfy my need, so it would be a 3 way cabinet
I did make that simulation in VituixCAD with the enclosure tool for the Gradient, but not yet the diffraction
The plan was to test all drivers on that IEC baffle, to see all the frequency responses in relation together, to see if it was possible to use them together for that 3 way cabinet, and then I would have made that cabinet.
But you told me it was no a good idea and I should use the actual cabinet to do the measurements instead.
Now should I understand that actually making that cabinet isn't also a good thing to do, when I tried to use traces from the datasheets for drivers you also told me that was not the right way, Then what is?
Besides that IEC baffle, I believed I was following Checklist in the Help of VituixCAD and the Measurement preparation that you wrote, I don't get what else I should do to progress anymore with designing this.
This question was fully expected and earned after my twists.One of the safest strategies is to design enclosure with minimum off-axis response variations due to diffraction. Off-axis responses look quite the same as axial response, except directivity due to surface of direct radiators (if horns or wave guides are not used).
Maximum rounding or multiple bevels to all possible edges produce minimum variation due to diffraction i.e. minimum random responses to different off-axis directions. For example if you have 150mm mid and 104mm tweeter in width=300mm cabinet, you have 150mm capacity for side edge rounding/bevels for mid, and 196mm for tweeter. Same for top corner. That approach does not (significantly) produce narrow peaks/dips to axial response with diffraction, so you don't have to make compromises to get logical off-axis responses, smooth DI response and smooth power response. Assuming that raw responses of drivers can be controlled with crossover and directivity features matched.
No cardboard or wooden prototypes needed, no IEC baffle needed. Just build final cabinet, install drivers, measure drivers, process measurement data (IR->FR, diffraction, merger), simulate crossover, purchase components, glue, solder and start listening music
Of course some knowledge about proper target for simulation is needed, but after that measurement, data processing and simulation takes 1/2..1 work days. Listening and tuning XO prototype are voluntary though that could be the funniest part of diy projects. Most of local VCAD users make some tuning, but I've skipped that part few times with common 2..3-way boxed projects. Target is so well known that listening & tuning could be just waste of time
Last edited:
This has been asked quite a few times so I just link to documents:
How To Start...
VituixCAD Measurement Preparations.pdf
VituixCAD_help_20.pdf
I think that document could help you with that "simple" question
:https://kimmosaunisto.net/Software/VituixCAD/VituixCAD Measurement Preparations.pdf
Yeah it's like you baited me there
Yes I understand that we should smooth the transition from front baffle to the sides as close to the sphere as possible, but my skill or cost involved will limit what I can do to evade the diffraction's, If I understood well the narrower width of that baffle (close to the driver) help limiting the diffraction to the shortest wavelength, so as long this is the case the axial placement isn't that much an issueOne of the safest strategies is to design enclosure with minimum off-axis response variations due to diffraction. Off-axis responses look quite the same as axial response, except directivity due to surface of direct radiators (if horns or wave guides are not used).
Maximum rounding or multiple bevels to all possible edges produce minimum variation due to diffraction i.e. minimum random responses to different off-axis directions. For example if you have 150mm mid and 104mm tweeter in width=300mm cabinet, you have 150mm capacity for side edge rounding/bevels for mid, and 196mm for tweeter. Same for top corner. That approach does not (significantly) produce narrow peaks/dips to axial response with diffraction, so you don't have to make compromises to get logical off-axis responses, smooth DI response and smooth power response. Assuming that raw responses of drivers can be controlled with crossover and directivity features matched.
No cardboard or wooden prototypes needed, no IEC baffle needed. Just build final cabinet, install drivers, measure drivers, process measurement data (IR->FR, diffraction, merger), simulate crossover, purchase components, glue, solder and start listening music
But there is also that Z axis between drivers phase that I could not predict without simulation (not smart enough probably)
Of course some knowledge about proper target for simulation is needed, but after that measurement, data processing and simulation takes 1/2..1 work days. Listening and tuning XO prototype are voluntary though that could be the funniest part of diy projects. Most of local VCAD users make some tuning, but I've skipped that part few times with common 2..3-way boxed projects. Target is so well known that listening & tuning could be just waste of time
You got me there also on "the knowledge about proper Target" I'm afraid I am missing that, I have expectations but certainly not in a controlled fashion, I just want to learn.
I believe you that you take only half or a full day for simulation, you built that (excellent) software
For me it's still a trial & error methodology, I feel there is so much I can do wrong.
Cutting of thick MDF or fillers are not not expensive or difficult, but if you are forced to compromise clear and logical sound stage on wide listening area, make as long edge radius as possible, locate drivers to golden ratio (by diffraction simulation) etc. and finally make good compromise between power&DI and axial responses to maintain at least sound balance.
Quite many commercial speakers are also compromises; easy manufacturing with standard tools vs. "perfect" shape. Cost could be that the best listening spot is smallish but sound stage still a bit fragmented. I've done it too to save sweat in manufacturing.
I think that document could help you with that "simple" question
https://kimmosaunisto.net/Software/VituixCAD/VituixCAD Measurement Preparations.pdf
Wouldn't you want to measure and optimise the crossover for the drivers response and phase at the actual listening height?
Seems strange to me to design a crossover based on on-axis measurements for each driver when that wont reflect reality...
Last edited:
I have seen that on this linked part of the Forum that others have asked and received answer on that specific question, have a look :
https://www.diyaudio.com/forums/software-tools/307910-vituixcad-129.html#post5822198
Measurement data could and should contain responses which reflect reality well enough at any simulated listening point in 3D space. Simulator tries to handle all practical listening points, distances and elevations as well as off-axis responses producing reflections.
Users typically measure with 5 or 10 deg angle steps including also vertical plane if driver, horn or enclosure is unsymmetrical. Axial response measurement of some individual driver is not necessarily selected for total axial response simulation if elevation of driver is higher or lower than speaker's origin i.e. expected listening elevation. For example response to vertical +5 or +10 deg is selected by simulator for the woofer, and horizontal +0 deg response for tweeter. 718 off-axis responses per driver is maximum which is quite adequate imo.
So there's nothing strange and wrong really. Exactly how it should be for simulation in two planes; 2 x 360 deg.
The initial diffraction attenuation I was going for was rounding over angle with a radius of between 18mm & 22mm (with a router) maybe in the future I could be doing bigger curves like with the "Kerfing" method but I have to practice that. Trapezoidal front face do not seem easy to do as are smooth transition curve around drivers, well at least for me, do you have an example of something better that could be done ?
Yeah I see that most speaker cabinet have sharp edges, my Cabasse are not exception to that, I should measure them to see how that response look
No cardboard or wooden prototypes needed, no IEC baffle needed. Just build final cabinet, install drivers, measure drivers, process measurement data (IR->FR, diffraction, merger), simulate crossover, purchase components, glue, solder and start listening music
Of course some knowledge about proper target for simulation is needed, but after that measurement, data processing and simulation takes 1/2..1 work days. Listening and tuning XO prototype are voluntary though that could be the funniest part of diy projects. Most of local VCAD users make some tuning, but I've skipped that part few times with common 2..3-way boxed projects. Target is so well known that listening & tuning could be just waste of time
That method, while "quick" often results in a "fixed" design. Crossover variation is of course available, but it's certainly not as flexible (at least without DSP correction), and that lack of flexibility can result in very similar objective results and uniformly mediocre subjective results (..indeed, a "waste of time").
Ex. what if you want to off-set (physically) the tweeter?Been there, done that, won't make that mistake again.
Of course if you already know the "exact" sort of design you want (with perhaps minimal crossover changes), then sure - it's the faster route.
As for the large test baffle: some drivers do change off-axis lower in freq.. but it's NOT picked-up with a limited bandwidth near-field "merge" (in addition to the typical "horn" situation). You can also "pick-up" some added diffraction and resulting freq. irregularity with increasingly smaller baffles (..though of course there are other methods to decrease this), particularly with respect to different measurement positions/axial-changes.
Last edited:
..my Cabasse are not exception to that, I should measure them to see how that response look
Ideally up on a stand outside near no reflection points.. barring that: lots and lots of pillows!
Honestly, measuring a loudspeaker's full bandwidth (in-room) well is far more difficult than the typical test-baffle measurement.
-of course it's less-so as you move higher in freq. looking for certain diffraction effects.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.