I used to diy speakers many years ago, but I'm getting back into it for a one-off project - pair of two way TL speakers.
I'm not very switched on with current crossover software.
I've been practising with Xsim and VCAD.
I like the power of VCAD, but I'm a bit worried that I might end up with a bad design if I inadvertently apply the wrong settings in say Preference Ratings.
I will be doing a range of horizontal measurements (for my FRDs) in an elevated outdoor area which should have no early reflections.
I've put a lot of work into the enclosure design (Hornresp), and a lot of work into the construction (I'm not a cabinetmaker's a***hole).
So I want the crossover to get the best out of my speakers.
Should I continue trying to learn VCAD, or just stick with Xsim?
BTW, I'm not an audiophile, and the speakers are only for casual listening in our untreated living room (maybe that answers my question. 🤔).
I'm not very switched on with current crossover software.
I've been practising with Xsim and VCAD.
I like the power of VCAD, but I'm a bit worried that I might end up with a bad design if I inadvertently apply the wrong settings in say Preference Ratings.
I will be doing a range of horizontal measurements (for my FRDs) in an elevated outdoor area which should have no early reflections.
I've put a lot of work into the enclosure design (Hornresp), and a lot of work into the construction (I'm not a cabinetmaker's a***hole).
So I want the crossover to get the best out of my speakers.
Should I continue trying to learn VCAD, or just stick with Xsim?
BTW, I'm not an audiophile, and the speakers are only for casual listening in our untreated living room (maybe that answers my question. 🤔).
When you process and handle the data the right way, you can get the same quality result from either of these.
Vcad offers to do more of the work for you.. but it does offer detailed results with very little input from the user, so you need to be sure it is giving you the information you need. With sufficient knowledge of it and speaker design you can cross check what it is doing.
Vcad offers to do more of the work for you.. but it does offer detailed results with very little input from the user, so you need to be sure it is giving you the information you need. With sufficient knowledge of it and speaker design you can cross check what it is doing.
Thanks Allen 👍
Kimmo has given me some Preferece Ratings which should suit my fairly basic design, so I guess I should be OK to use his VituixCAD.
As you said, once you have added all your information (drivers, FRDs, ZMAs, etc.), his optimizer should be able fine tune any setup you want to try.
It just feels like I'm using a Ferrari to go shopping.
I didn't mean to give that impression. There is more to it than using the right preferences. Not even the optimiser can get it right without the right data presented in the right way.
Your safety net will be your own mastery of the process.
BTW, what kinds of measurement are you planning to make? An frd and a zma for each driver maybe?
Last edited:
Hi Geoff
I've done a lot of reading and research over several months, and tried running example files through VituixCAD and Xsim.
And after all that, my understanding is only scratching the surface.
But I'm still willing to give it a go.
Both programs have support forums where experienced guys seem always willing to help.
My plan, once the enclosures are built and drivers installed, is to measure (using REW) the individual drivers FRDs at several horizontal angles (probably on one side only); then measure the FRDs of the two drivers together (to determine my Zoffset), then measure the ZMA of each driver.
I've been told I probably don't need to measure any off-axis angles in the vertical plane (thank goodness).
I'm fairly confident using VituixCAD, and entering all my data. But because it is so overwelling to me, I am a little worried I may be wasting my tme if I have a setting or imput wrong.
Ok, it's a good place to start.
If you do, you want to keep the angles the same for both planes, so duplicate your horizontals rather than mirroring.
You should not need to know the offset if you are using proper measurement techniques. This suggests to me your microphone doesn't do timing properly, is it a USB mic?
Maybe. Vertical can be different to horizontal. Also while you can sometimes do one side only with horizontal, you may need up and down with vertical.
If you do, you want to keep the angles the same for both planes, so duplicate your horizontals rather than mirroring.
+1 for using a non-usb microphone like the emm6. Not having to measure/guesstimate offsets is fantastic. When you can load your measurements in VituixCAD knowing that at least that part is taken care of, the rest becomes easier. It's unbelievable how correct VituixCAD can simulate a crossover.
I used to have an EMM6, complete with a small mixer (to provide the phantom power). It would then run through an old SB external sound card, which after much frustrating troubleshooting, I found out that the sound card was dodgy.
At that time I was only doing room measurements with REW, so I took the easy way out and bought a USB umik-1.
However, John Mulcahy has now updated REW so that you can take timed measurements with a USB mic. You need to have another speaker pointing at the mic, using the other channel (usually the Left), and REW takes it timing from this speaker (at least that's my understanding).
Are you saying that you don't need to measure both drivers together if your measurements are time aligned?
As this is a one-off project, I really don't want the added expense of buying another mic (including some sort of phantom power supply).
I am dreading doing vertical plane measurements. It's easy to rotate the enclosure for horizontal measurements, even without a turntable.
But rotating the enclosure (it's 1240mm high) around the X axis to get vertical plane measurements sounds very difficult to me.
Someone suggested laying the enclosure on it's side, but that would still entail a completely new setup with the mic as well as the enclosure.
I'm just not sure if it's worth the effort.
I feel your pain but its the dodgy sound card that is the root cause. I new/better one would have two channels and provide phantom power for the mic. Laying a tower speaker on its side at the top of a tower is indeed a challenge. laying it flat on the ground not so much if you have a large flat spot. then you can move the mic along a radius a calculated chord length and do ground plane measurement...
Has anyone got any tips for getting the optimiser to work in VituixCAD?
I've tried clicking on everything, but it just seems to go anywhere but follow the target curve.
I've tried clicking on everything, but it just seems to go anywhere but follow the target curve.
A circuit optimiser doesn't change the circuit, only the values. It can't do well unless you make that decision for it. Here is an example of a circuit not being able to fit a curve - https://www.diyaudio.com/community/...overs-without-measurement.189847/post-6910468
Actually once you start to think about moving the mic (instead of the speaker), for the vertical plane measurements you could make a template from card or ply with the appropriate angles and mic locations. Then, maybe with someone holding the template for you, you move the mic to the next location. They just have to hold the template at right angles to the baffle (vertically and horizontally).
At this stage, for the vertical plane, I'm really only thinking about measuring 5 and 10 deg. up and down.
It's possible. I move the mic when measuring corner speakers as the room corners are a part of the speaker.
However when you use an acoustic timing reference it's necessary to move that extra speaker the same amount as the mic, which can be a challenge. Also, as you go close to the floor or ceiling you get a shorter reflection free time.
However when you use an acoustic timing reference it's necessary to move that extra speaker the same amount as the mic, which can be a challenge. Also, as you go close to the floor or ceiling you get a shorter reflection free time.
Actually you are probably better off asking that question here https://www.diyaudio.com/community/threads/vituixcad.307910/page-203#post-7134943
But before posting your question, do a search.
In the results, look for comments by kimmosto (the VituixCAD author) and Dcibel.
It is a real challenge to do this. The speaker must be positioned on its side, elevated, and then rotated about a fixed vertical axis. For a tall speaker, the only real option is to build (or buy) a rotating table or stand. Attached is a photo of what I built.
j.
Attachments
Where I propose to take my outdoor measurements, the ground (driveway) drops away from the test location, and any other objects are around 30 metres away.
So the idea of being able to keep the speaker fixed during vertical plane measurements is very appealing.
However I take you point regarding the importance of keeping the timing reference uniform throughout the measurement process.
Some tips that work for my rudimentary level of expertise:
-constrain the frequency range that you want the optimizer to work on; run it over and over in little chunks
-do one driver at a time (disconnect the earth on the other drivers)
-get the main curve done manually (by tweaking the values yourself) to get somewhat close to the target curve, then run the optimizer after
-you can micro manage the optimization in steps by turning components in one part of the filter optimization on or off (select it then press CTRL + Q or CTRL + E), e.g. leave a cap and coil on and run the optimizer, but disable it from running on the notch circuit, then turn the main filter optimization off and turn the notch back on, run the optimizer again
-in the bottom left you can set min and max values for parts so they don't waste time trying to evaluate silly values, e.g. set minimum on a resistor to 0.5ohm, set the min-max on a cap from (whatever) to 300uf, unless you want to pay loads of money for massive caps
-you can check "min impedance" higher than you'd be OK with, e.g. 8 ohms, just so it doesn't allow results below 1ohm, for example- it will still dip below 8 to meet the curve, but it won't allow nutso values
-checking "min phase" seems to do some phase alignment as part of the optimization
-you can set the number of optimizations to 200 or so, just so it doesn't spend time redundantly trying to optimize something to 0.1 of a value
-you can set rounding to E12 so it sticks to parts you can buy
hope this helps