Simulation in VituixCAD - what did I miss?

To date I've built my speaker driver and circuitry design understanding to a level where, using VituixCAD, I can plug in some drivers, trace SPL/Z, make up an enclosure design, do some diffraction simulation, and finally start messing around with a crossover configuration that result in some nice looking graphs (in my opinion, anyway!).

Before I pull the trigger on some components I decided to validate what I've been doing by taking an existing design and simulate it and see if the graphs match what you'd think sounds normal, or even good!

The first thing I did was take Paul Carmody's Amiga MT and plug everything in as I've been doing. I traced the SPLs/Zs, did the baffle diffraction, then put Paul's exact XO design in to the XO tool. Curiously, the Woofer response looks ideal but the tweeter is a mess. I made an assumption on the 20mm Z position on the tweeter. No amount of playing with the tweeter circuit using the same layout results in a reasonable result.

Referring to the tweeter diffraction graph, it looks pretty tidy... The overall impedance is even a pretty close match. Is what I'm doing here an issue with me, VituixCAD, something else? :confused:
 

Attachments

  • diyaudio1.PNG
    diyaudio1.PNG
    947.7 KB · Views: 265
  • diyaudio2.PNG
    diyaudio2.PNG
    463 KB · Views: 252
Hi, yeah it is going to be a guestimate without measurements of the actual device. Working with shady data leads to doubt, if you mind about it. Some people do crossovers with pencil and paper and be happy with it :) Both cases don't represent reality accurately so micromanaged simulation is pointless other than studying purposes. I wouldn't buy expensive components based on this kind of sims because there is gonna be some tweaking after building and measuring. I get in doubt and I guess many beginners do.

It looks like a typical two way speaker. Check out some other two way designs, whats the difference to this one? Maybe the dip at xo is a feature, just magnified in the sim due to insufficient data (measurement)? Accidental or deliberate error in the published xo doc might also cause this? You have ~140deg phase difference at the great dip. The dip should be lot less if you invert polarity of either driver?
 
Last edited:
diyAudio Moderator
Joined 2008
Paid Member
It's not unusual to prefer the treble down that far. If you take into account the woofer narrowing above 1k, and the crossover narrowing, it may even be a part of the natural flow.

Yes, flip the polarity and let's see what's there.
 
Looking at the sim you have no Y or X offset on the speakers I.e. they are places on top of each other which, unless they are coaxial speakers needs to be addressed. I have also found that Vituix is not so good at simulating the distance from the speaker, Jeff Bagby’s PCD spreadsheet seems to do this better.
 
Looking at the sim you have no Y or X offset on the speakers I.e. they are places on top of each other which, unless they are coaxial speakers needs to be addressed. I have also found that Vituix is not so good at simulating the distance from the speaker, Jeff Bagby’s PCD spreadsheet seems to do this better.

This raises an interesting point which I wasn't completely sure about; both driver response graphs with diffraction simulated are located as per the baffle design. You can see in the second attached image the tweeter is placed above the mic - when I did the same for the woofer it was placed below where the mic is.

Are speaker X/Y also meant to be inputted in to the XO layout spec as well as included in the positioning in the baffle diffraction?
 
I'm not sure how the diffraction tool utilizes the supplied SPL graph, probably uses it as the driver source without assumptions? There probably is info in the VCAD manual for this.

Thinking of it, if you keep the mic at designed listening axis for both driver diffraction tool measurements then the Y coordinate should be 0 for both in the main program, like you have done. If you moved the mic to center of each speaker in the diffraction tool, then you would enter the Y coordinate in relation to the designed listening axis for each driver in the main program. I suggest you do the latter because then you can tweak the Y parameters in the main program and kind of get the information you need before building the real thing, what the distance between drivers should be for optimal xo.

Anyway I think the diffraction tool does "virtual measurements" so leave your mic distance in the diffraction tool somewhere in the farfield (> 1m) and you should be alright. Then you can use any (your real) listening distance in the main program.

How close this would be to reality I don't know. Tune up the xo after you have build and measured the real thing to be sure you are getting better results than with pen and paper.
 
Appreciate the input, all! :) Flipping the polarity on the tweeter does indeed remove the dip.

I'm going to do a little more playing with mic positions and also plug in another known design and see what the result is. Fun!

The more I screw around with this stuff, the more tempted I am to get some measuring equipment...
 

Attachments

  • diyaudio3.PNG
    diyaudio3.PNG
    940.4 KB · Views: 177
Unfortunately the polarity of the Amiga tweeter is not reversed, so the cause of the phase misalignment in your sims lies elsewhere and that's the reason for the the big dip in your response at ~3kHZ.

I'm not that well versed in VituixCAD but I'm reasonable certain that you don't have your Y and Z coordinates right and that will affect your phase response which again will affect your FR. Try changing the tweeter to -20mm or try it as a +20mm on the woofer. And if you are not going to include the Y coordinates, I suspect the Z value should be a little larger.

Also, you should be extracting minimum phase for your drivers after you have included box and baffle effects. I'm not quite sure how VCAD does that but XSim will do it for you as well as some of Jeff B's Excel programs.

If you think about it, what the simulations need to do to try to be as accurate as possible is to imitate the actual measurement conditions as closely as you can. Typically for a 2-way, the mic will be placed on the tweeter axis anywhere from about 500-1000mm away and left there for both the tweeter and woofer measurements. So setting the mic (axis) distance in your baffle diffraction sims at 3000m is not going to give the best results. Unfortunately, VCAD baffle sims don't like it when you reduce the mic distance to such small distances either (look in particular at the amount of baffle step loss at 500mm vs 3000mm - it should be -6dB but is not even close) and this is 1 of the reasons I don't like to use VCAD just for the purposes of simulations, even though you can import the driver FR and see the immediate effects of baffle size and driver placement instantly in real time which I have been wishing other baffle diffraction programs could do (better) for years.

Lastly, the response discrepancy you are seeing in the tweeter level comes down the accuracy of the manufacturer's measurements. In the case of the RS180-8, it's actually pretty good, but in the case of the tweeter, something is definitely off. If you fix your phase problem and change the Amiga tweeter parallel resistor back to 4.7ohms, you will end up with a sim very much closer to Paul C's original design.
 
I'm a novice at passive crossover design.

What is the purpose of the capacitor in parallel with the first lowpass inductor?

Also, what is the purpose of the resistor in series with the shunt capacitor? I thought that component configuration was used to reduce an impedance peak at resonant frequency, yet without the resistor I can see the 3rd order lowpass filter for the woofer.
 
diyAudio Moderator
Joined 2008
Paid Member
@ Pida, yes it looks like that.

@ Shaz. The capacitor turns it into a notch, in this case in the stopband. You can see the response of the woofer cuts out quickly but rises up again near 10k. The resistor is needed or else there will be a capacitor short circuit across the amp.