It's really cool, but you have to wear those funky glasses with different colored lenses!
. Seriously though, here's a link that where Bill explains the additional features: Xsim-3D development... I could use some math help
It's an unfinished major revision of Xsim, intended to handle off-axis responses and room boundary support (3D), as well as active and dsp filters. The 3d part ran into complications and I ran out of enthusiasm to keep developing that one. And Kimo had his VituixCad handling off-axis as well as baffle effect already. There didn't didn't seem to be much need to keep going on a 3D version of Xsim, then. And also XSim3D didn't come across as intuitive or easy to handle as was the original goal with XSim. So I shared the technique/architecture for free-form circuit design to Kimo and stopped work Xsim3D.
Is still use it sometimes for active filters or simulation of boundary effect, but the software has a number of bugs - so be careful to make sure what you are seeing makes sense if you use it!
Hmmm, yeah, ok.
But it was the first time I saw it wired like that.
And it doesn't have the same effect on FR as when using a cap in line with the tweeter and an inductor in line with the woofer.
So, I was wondering about the interactions when wired this way.
Bill,
Kimo isn't a fan of modelling crossovers in just the forward (or one) axis and pretty much dismisses Omnimic as being inappropriate for crossover design purposes due to its approach to timing/phase measurements, if I understand this correctly.
The work I've done with XSim and Omnimic has produced very pleasing results - to me.
What am I missing here?
Thanks.
Kimo isn't a fan of modelling crossovers in just the forward (or one) axis and pretty much dismisses Omnimic as being inappropriate for crossover design purposes due to its approach to timing/phase measurements, if I understand this correctly.
The work I've done with XSim and Omnimic has produced very pleasing results - to me.
What am I missing here?
Thanks.
You can use OmNiMic with VituixCAD the same way than you have used OmniMic and XSim. Also VituixCAD contains MP extraction and timing adjustment in microseconds or Z mm so it's fully capable to handle simple design method limited to axial response or max. within listening window and USB mics.
BUT that method does is not capable to work 3D to high off-axis angles without either measured phase with common timing reference + simple "X-Ray" geometry calculation from radiator's origin to mic OR quite accurate timing/phase calculation from radiator's throat around horn/waveguide/cone surface around baffle edges around box to mic + normalized or minimum phase responses. In addition, minimum phase extraction does not work at all with radiator's having polarity swapping(s) typically to 90...180 deg such as dipoles and resistance enclosures or monopole+dipole gradients. Normalized phase would work though.
Therefore it's very clearly recommended that VituixCAD user should play with the first option (measured phase with common timing ref + simple x-ray geometry calculation). But it's just recommendation you can flush from the toilet in case measurement gear does not comply. Cost is that you cannot get or optimize power response and DI and responses to high off-axis angles. Needs measurements after XO is already designed which is kinda too late especially with passive XO.
Finalized XSim 3D would face the same challenge with VituixCAD: program must handle radiator's acoustics and mechanics (speed of sound and sound path length as a function of frequency) if measured phase with common timing reference is not available.
Thanks you guys.
I've built 1st and 2nd order cross-overs and could also a 3rd order. Many hours spent listening to different capacitors and their values,,, even a .1 uf changes phase, making or breaking nirvana much of the time. What I was expecting (hoping) from this program was a graph-simulation of say a 2nd order with drivers I have in my cash. It looks like there are other programs needed to be added to do what I was hoping. Programming or even sorting through programs are not my high points.
I try and try to use passive instead of settling with my 3x2 DSP. I've no problem getting great sound with DSP but hate to think I'm not getting the most out of my TDA-1541 NOS DAC. I have a dozen DSP units still if anyone is interested. Just need a couple units for my self and thing didn't go as planed.
Hi 6NS, you can do what you want I think with Xsim, the only other program/hardware you need is something to measure your actual drivers as you'll use them. Published responses and impedance curves at best help you choose drivers that might work well together. But they vary over production runs and often are supplied without reference to baffle arrangement or other conditions; and some graphs, like some manufacturer supplied T/S parameters, can be little better than msrketing fiction!
I believe he is referring to Omnimic's inability to measure true distance (total phase delay including time of flight to the microphone) due to its feature of being able to measure without a physical connection between speaker/amplifier and the mic/laptop measurement gear. In other words, the computer and Omnimic V2 can't know exactly when the signal left the speaker. That makes for a remarkably easy and convenient way to measure speakers and drivers with almost zero setup time. But it does make it very tricky If you want to measure each driver over a set of angles and use that large set of data to design different speakers for response over off-axis angles. The alternative with Omnimic is to have the drivers already mounted in their final baffle arrangement and measure them all with delays of drivers determined relative to each other driver (in that situation) over a range of angles. But that way doesn't allow use of a single independent library of drivers portable over different mounting combinations.
In actuality, I've heard of few who go to the trouble of either of those two methods, as it's a LOT of measurements and files to make and configure and a lot of ways make errors. Usually designs are done for one or maybe two angles with drivers selected and arranged in ways best expected to behave well off-axis. I've done it the many-angled way with my synergy-horn styled designs, and the results were dead-on (after several re-do-s!) but it does take a lot of time and effort. (Even though Omnimic software does make it relatively convenient to collect and save sets of automatically named frd files in succession as the mic or speaker is rotated.)
// I probably shouldn't mention this yet, but watch for a future Omnimic revision (V3) to provide a timing "loopback" feature allowing true time of flight/delay determination (while still letting the mic be free of wires to the speaker system!). As well as a number of other useful improvements. //
In actuality, I've heard of few who go to the trouble of either of those two methods, as it's a LOT of measurements and files to make and configure and a lot of ways make errors. Usually designs are done for one or maybe two angles with drivers selected and arranged in ways best expected to behave well off-axis. I've done it the many-angled way with my synergy-horn styled designs, and the results were dead-on (after several re-do-s!) but it does take a lot of time and effort. (Even though Omnimic software does make it relatively convenient to collect and save sets of automatically named frd files in succession as the mic or speaker is rotated.)
// I probably shouldn't mention this yet, but watch for a future Omnimic revision (V3) to provide a timing "loopback" feature allowing true time of flight/delay determination (while still letting the mic be free of wires to the speaker system!). As well as a number of other useful improvements. //
How are you dealing with the latency of the conversion from Analog to digital?
I use a system that has a true time of flight capability. (Smith&Larson Woofer Tester Pro) It is useful for different things. And it does have a loop back feature that is in use all the time if enabled.
You can do some interesting things with a true time of flight setup.
Just so: the cap and inductor values are usually dissimilar between series and parallel filters for a given transfer function. And yes, there are interactions; with series filters everything is in circuit, so with certain exceptions, if you change one component, you'll affect everything. There was a fad for series filters a few years ago, and a few years before that, and no doubt it will rise again at some point, like the analogue undead.
1st order series (and only 1st order) has one unique property of maintaining flat electrical summing even when the impedance load changes (the frequency shifts) but that doesn't mean it will do so acoustically as that depends on the FR of the drivers. Series filters can also sometimes allow you to hit a given set of transfer functions with fewer components than an equivalent parallel circuit, since some components may be pulling double-duty. Otherwise, by and large anything you can do with parallel you can do with series -only it's usually harder to achieve given the innate interactions, so while it would be excessive to call them a solution in search of a problem, since they do have uses and are functional enough, they are also often a good way of making life harder than it really needs to be.
wow, thank you! and thank you very much for the original xsim, i like it very much and i play with it almost every day
Trade secret for now. Latency isn't the issue, the problem was synchronization.
Just so: the cap and inductor values are usually dissimilar between series and parallel filters for a given transfer function. And yes, there are interactions; with series filters everything is in circuit, so with certain exceptions, if you change one component, you'll affect everything. There was a fad for series filters a few years ago, and a few years before that, and no doubt it will rise again at some point, like the analogue undead.
1st order series (and only 1st order) has one unique property of maintaining flat electrical summing even when the impedance load changes (the frequency shifts) but that doesn't mean it will do so acoustically as that depends on the FR of the drivers. Series filters can also sometimes allow you to hit a given set of transfer functions with fewer components than an equivalent parallel circuit, since some components may be pulling double-duty. Otherwise, by and large anything you can do with parallel you can do with series -only it's usually harder to achieve given the innate interactions, so while it would be excessive to call them a solution in search of a problem, since they do have uses and are functional enough, they are also often a good way of making life harder than it really needs to be.
Thank you for this. Appreciated.
Yes, indeed. I needed 7 components to achieve what this 1st order series did in 4.
One more thing to investigate. Ain't life full of little things to learn?
Hi everyone. New member here. I was recently told about XSim software and suggested to use it for optimizing my attempt to convert Norman Lab Model 10's to a single 12" 2-way system using a compression driver for high frequencies. My inspiration, of course, was the econowave thread over on AudioKarma. I chose a Celestion 12" driver - TF1225 - for the low frequency work as the specs seemed to best match my 2.6 cu.ft. cabs. My compression driver and horn are JBL Selenium D220Ti and JBL 338800-001.
I am curious if there is a library of XSim files for various drivers?
Thanks for any suggestions on getting started and thank you bwaslo for sharing your extremely useful tool!
I am curious if there is a library of XSim files for various drivers?
Thanks for any suggestions on getting started and thank you bwaslo for sharing your extremely useful tool!
I spent some time learning REW and XSim and I made a lot of progress on this! I used REW to measure each driver individually in the cabinet with no crossover. I then let REW export the data and created the .frd file for each. I then plugged that information into XSim and was able to model crossovers. I started with the existing crossover and the FR was nearly identical to what I actually measured last week. I was able to tweak the existing xover and improve the FR a lot and improve the impedance curve a lot. Finally, I tried creating a few new xovers from scratch - 2nd, 3rd, and 4th orders. The one with the best curve does have a few impedance swings, but nothing as drastic as the original. I'd appreciate input on how this looks and whether the impedance looks ok for a tube amp.
Raw data on each driver in cabinet with no xover captured with UMIK-1 and REW:
Original Xover and predicted FR and impedance curve:
Improved Xover and predicted FR and impedance curve:
Tom
Raw data on each driver in cabinet with no xover captured with UMIK-1 and REW:
Original Xover and predicted FR and impedance curve:
Improved Xover and predicted FR and impedance curve:
Tom