Making a heatmap is very easy. It's basically a spinorama, but made pretty by the software through the magic of interpolation. That's right, VituixCAD makes it for you automatically when you import measurement files. See chapter 6 from the link. It's great if you want to impress/scare people.
There are ways to either approximate it from a simulation or measure the real thing. The latter is more accurate, but the first one is fantastic for learning!
http://www.htguide.com/forum/showthread.php?44717-Simplified-polar-data-generating-in-VituixCAD
You can even try it with some online project and see how close you can get.
When the paint dries on your fresh new speakers, you can read this:
https://www.sausalitoaudio.com/wp-content/uploads/2018/07/Interpreting-Spinorama-Charts.pdf
The easiest way of achieving a smooth directivity plot is to get all the drivers, measure their frequency responses, make pretty spinoramas, open a beer and just stare at the lines. You find a place where the tweeter has the same spread between the lines as the mid. Voila, somewhere around there your crossover point will have the smoothest response on all angles.
Then you remember that you can't cross the tweeter that low and you settle on a compromise. 😉
There are ways to either approximate it from a simulation or measure the real thing. The latter is more accurate, but the first one is fantastic for learning!
http://www.htguide.com/forum/showthread.php?44717-Simplified-polar-data-generating-in-VituixCAD
You can even try it with some online project and see how close you can get.
When the paint dries on your fresh new speakers, you can read this:
https://www.sausalitoaudio.com/wp-content/uploads/2018/07/Interpreting-Spinorama-Charts.pdf
The easiest way of achieving a smooth directivity plot is to get all the drivers, measure their frequency responses, make pretty spinoramas, open a beer and just stare at the lines. You find a place where the tweeter has the same spread between the lines as the mid. Voila, somewhere around there your crossover point will have the smoothest response on all angles.
Then you remember that you can't cross the tweeter that low and you settle on a compromise. 😉
Last edited:
Hahaha I love it. I'm below the legal drinking age, but I mean.. if you say it's a necessary part of the process...
Thanks for the info!
Before we had advanced simulation capability (such as VituixCad), we used some generalized rules to help us select crossover frequencies. Of course, with simulation, these rules are not needed. But in the absence of simulation, these rules can help deliver an acceptable result... as long as they are combined with all the other "good design practices". Simulation, on the other hand, allows our designs to venture off the beaten path.
One rule was that the maximum crossover frequency for a woofer or midrange was the point where the 45 degree frequency response was -3 dB below the on-axis response.
Another rule was to cross no higher than the point where Ka=2 for the woofer or midrange. Ka is the frequency associated with the wavelength equal to the circumference of the radiating surface. It is calculated from the radiating surface area (Sd) of the driver.
Another good guideline that was used in the past was to examine the on-axis frequency response and the 30, 45, 60 degree off axis responses. At some frequency, the response will usually become ragged. This is the upper end of the usable range. Often times the off axis response will become ragged before the on axis response. In other words, the off-axis and on-axis responses will follow each other nicely from low frequency to some point, but then the off-axis starts to diverge. If possible, we would have crossed at least an octave below that point.
I hope this helps... but the best path forward is Juhazi's advice
One rule was that the maximum crossover frequency for a woofer or midrange was the point where the 45 degree frequency response was -3 dB below the on-axis response.
Another rule was to cross no higher than the point where Ka=2 for the woofer or midrange. Ka is the frequency associated with the wavelength equal to the circumference of the radiating surface. It is calculated from the radiating surface area (Sd) of the driver.
An example of calculating Ka = 2 for a 5" driver.
The 5" driver has a radiating surface area (Sd) of 103 cm^2.
We calculate the diameter as 11.4 cm, and the circumference as 35.9 cm.
The frequency of a wavelength of 35.9 cm is 956 Hz.
So, for this driver Ka=1 is 956 Hz, and Ka=2 is twice that, 1911 Hz.
Another good guideline that was used in the past was to examine the on-axis frequency response and the 30, 45, 60 degree off axis responses. At some frequency, the response will usually become ragged. This is the upper end of the usable range. Often times the off axis response will become ragged before the on axis response. In other words, the off-axis and on-axis responses will follow each other nicely from low frequency to some point, but then the off-axis starts to diverge. If possible, we would have crossed at least an octave below that point.
I hope this helps... but the best path forward is Juhazi's advice
I think these rules of thumb more closely align with what my understanding of things were, but I still learned something new from your comment. I think part of my issue has been watching videos from the speaker designers on YouTube.. They seem to do a good job, but there seem to be a lot of folks around here that have good arguments for why they're doing a better job 🙂
With so many contradictory schools of thought, you can hardly blame a newbie for getting lost!
Care to share the tweeter and the waveguide you plan to use? WG aren't all the same and should be matched to the tweeter.
Ralf
By the way, I am too dumb an lazy to start learning VCAD... I made the normalized directivity heatmap with free test version of Omnimic (from Parts Express), because my favourite REW can't make those. ARTA can too, but is a bit difficult to learn.
As well I have saved myself from learning to design passive xo's, I work with multi-ch dsp, minidsp and Hypex.
https://www.diyaudio.com/community/threads/aino-gradient-a-collaborative-speaker-project.231353/
As well I have saved myself from learning to design passive xo's, I work with multi-ch dsp, minidsp and Hypex.
https://www.diyaudio.com/community/threads/aino-gradient-a-collaborative-speaker-project.231353/
Ok, so I realize that I will be taking a gamble here but I cannot resist the temptation! If it does not work out I will probably use a tried and true tweeter waveguide combo, but this will be cheap enough that I want to give it a try.
https://www.thingiverse.com/thing:4748246
https://www.parts-express.com/LaVoce-TN131.00-1.3-Soft-Dome-Neodymium-Tweeter-8-Ohm-293-739
It is scary that the creator of the waveguide model didn't even bother to add measurements, but if his does not work out then I guess I will also get to learn how to design waveguides!
It's a wild (and probably not very good) idea, I know, but here is my thinking for the speaker.
Peerless SLS 830668 on lower bass duty, Usher 8836AC2 on mid duty, waveguided LaVoce with a fourth order slope high pass, AND a first order slope low pass in the super high treble region. Then a Dayton Audio ND13FA-4 1/2" tweeter upward firing as a super tweeter. Presumably because it is so small, it has an insanely gradual and even off axis drop off.
https://www.parts-express.com/pedocs/specs/275-104--dayton-audio-nd13fa-4-spec-sheet.pdf
I want to see what happens when I separate the treble response into two drivers, one with narrowish dispersion and one that tries to be omnidirectional. I understand that even in a competent designer's hands, this probably would not be very easy, but I just can't help myself from trying!
I have had many speakers in the past, Ohm Walsh 2's, Magnepan .7's, Linn Isobarik's, Klipsch Chorus', Vandersteen Model 3's, Selah Tanzanite's, and I am curious to see what happens when you try to cram completely opposite design practices together into one speaker.
I like super focused speakers, and I like wall of sound speakers. I want to experiment with what happens when you try to have both? Probably impossible haha but I'm having fun!
I have eventual intentions of reigning myself in a little bit, but DIY is so fun and new to me right now and I'm a dumb *** kid and right now I just want to do something crazy 🤣
Last edited:
Woah those look awesome! I will soon be building XRK971's open baffle speaker he calls the "XSD", but it looks like you opted for a..um.. no baffle? lol
I guess the hypothesis is that having lower dispersion at the start of the HF range and higher dispersion at the top will net a speaker that sounds airy and wide and simultaneously focused. I have absolutely no reason to think this mind you, but even if my efforts do not yield the sound I imagine, I'm sure it will at the very least be interesting!
I was reluctant as well, but I found it was a short learning curve. Well worth it.
It seems you are the kind of guy who thinks that all is possible. Good luck.
One of worst ideas in a speaker.
Ralf
"One of worst ideas in a speaker."
Not necessarily, but also the benefit is questionable. Typically the "supertweeter" is supposed to widen directivity or to project treble waves to the ceiling or frontwall, to improve balance of reflected sound. More often it is marketed as adding sparkle and clarity etc. and a horn loaded unit is used. Actually the second unit creates more comb filtering effect unless it is on the backside.
Anyway, of course you can do that as an excersise and learning project! It will be fun to take measurements 0-180deg!
With a 9" bass, I would use a smaller midrange 4-5". But a waveguide helps to match mid and tweeter, at the expence of narrowing radiation.
Nowdays the designer has to choose if (s)he aims for wide or narrow directivity, which affects how a stereo pair will work in the room. Smooth directivity change is essential in both cases.
Not necessarily, but also the benefit is questionable. Typically the "supertweeter" is supposed to widen directivity or to project treble waves to the ceiling or frontwall, to improve balance of reflected sound. More often it is marketed as adding sparkle and clarity etc. and a horn loaded unit is used. Actually the second unit creates more comb filtering effect unless it is on the backside.
Anyway, of course you can do that as an excersise and learning project! It will be fun to take measurements 0-180deg!
With a 9" bass, I would use a smaller midrange 4-5". But a waveguide helps to match mid and tweeter, at the expence of narrowing radiation.
Nowdays the designer has to choose if (s)he aims for wide or narrow directivity, which affects how a stereo pair will work in the room. Smooth directivity change is essential in both cases.
Last edited:
If one uses the pioneer's way to choose XO points, it works fine and IME the only way to do multi-ways no matter how many XO points are used, so guess we'll have to 'agree to disagree'.
I have found it to be really interesting and fun to have a knob that I can turn to smoothly vary the crossover frequency while listening. I use a dbx line level crossover to do it. I first use a miniDSP to make corrections to the individual driver responses so they are flat through the desired crossover range. Then patching through the dbx 4th order analog crossover lets me have that knob. Each driver imparts some character to the sound if only due to the dispersion and radiating area. The miniDSP lets you save four configurations, but the 2x4HD that I use requires a few seconds of silence to swap between configurations. That is a bit less than optimal.
I agree with olsond3, trying different xo slopes and frequencies (with dsp) and listening to them, is fun and educative. Eg. "perfect" LR2 vs. LR4 at same freq do sound different! I use minidsp 4x10HD or Hypex FA123, both with everal optional configs.
When one is making diy speakers for himself, the choice must be done by personal preference. Measurements are very important to get xos working like they should.
When one is making diy speakers for himself, the choice must be done by personal preference. Measurements are very important to get xos working like they should.