Vituixcad does have those tools in its suite. Does it integrate them or is it still like using separate tools?
Where I pay attention is that some tools are best as guidelines. They are very useful to us but for example, some kinds of room simulation can be a guess at best. We want to do the simulations but we want to be careful about where and when to do any integrating.
Where I pay attention is that some tools are best as guidelines. They are very useful to us but for example, some kinds of room simulation can be a guess at best. We want to do the simulations but we want to be careful about where and when to do any integrating.
Hi AllenB,
This is an excellent thread in that it lends itself to bridging the gaps between manufacturers test data, crossover design and why what you see is not always what you hear in your listening rooms. The why is we’ll explained in terms that can be understood.
I use Linear X Leap 5 and l collect my own data via LinearX LMS or REW. Leap 5 is a software simulator with a number of optimisation engines and a filter design tool with analogue active and passive crossovers.
However, as you mentioned on the previous page you still need to listen as you develop your crossover. A lot of what you end up with subjectively are the psycho acoustics of your particular room, baffle and driver layout that are not visible on a snap shot response measurement.
The size, shape and layout of the drivers on baffle can have a marked effect on what you hear. Your room also influences what you hear in reality.
This is because the loudspeaker sends sound waves off in 3D dimensions and these waves bounce around your room and then come back to your ears a short time later. Depending on the size, shape and surfaces in your listening room there might be slap echo over a range of frequencies and other effects leading to the subjective impression something is out of balance.
This is caused by the reflections and the reverberation time in a particular space or environment. RTA60 is measurement of the time taken for a particular sound to decay in level by 60 db.
https://www.cirrusresearch.co.uk/blog/2018/04/what-is-reverberation-time-and-how-it-is-calculated/
Why are these factors so relevant in practical real world crossover design?
JBL legend Greg Timbers recently explained in an interview on Erin’s Audio Corner that an early subjective assessment in a real listening room of your proposed crossover frequencies and slopes can be an early indicator of the success or failure of your crossover network, choice of drivers and baffle layout.
Greg Timbers explained he used a switch box of various capacitors and inductor values to be able to subjectively listen to different crossover frequencies and slopes using the actual drivers on the enclosure baffle in a real listening room. Not an acoustic test chamber. This approach helped fast track the more imperial design practice used by JBL back in the 70’s through to the last 10 or 15 years.
Only recently have highly advanced programs like Loudsoft Crossover enabled manufacturers to quickly assess different crossovers in a test and learn approach.
I thought this would help bring to life your excellent thread.
Making up such a crossover board would be a fun project and as outlined above a very useful tool in a diy loudspeaker builders workshop.
lan
This is an excellent thread in that it lends itself to bridging the gaps between manufacturers test data, crossover design and why what you see is not always what you hear in your listening rooms. The why is we’ll explained in terms that can be understood.
I use Linear X Leap 5 and l collect my own data via LinearX LMS or REW. Leap 5 is a software simulator with a number of optimisation engines and a filter design tool with analogue active and passive crossovers.
However, as you mentioned on the previous page you still need to listen as you develop your crossover. A lot of what you end up with subjectively are the psycho acoustics of your particular room, baffle and driver layout that are not visible on a snap shot response measurement.
The size, shape and layout of the drivers on baffle can have a marked effect on what you hear. Your room also influences what you hear in reality.
This is because the loudspeaker sends sound waves off in 3D dimensions and these waves bounce around your room and then come back to your ears a short time later. Depending on the size, shape and surfaces in your listening room there might be slap echo over a range of frequencies and other effects leading to the subjective impression something is out of balance.
This is caused by the reflections and the reverberation time in a particular space or environment. RTA60 is measurement of the time taken for a particular sound to decay in level by 60 db.
https://www.cirrusresearch.co.uk/blog/2018/04/what-is-reverberation-time-and-how-it-is-calculated/
Why are these factors so relevant in practical real world crossover design?
JBL legend Greg Timbers recently explained in an interview on Erin’s Audio Corner that an early subjective assessment in a real listening room of your proposed crossover frequencies and slopes can be an early indicator of the success or failure of your crossover network, choice of drivers and baffle layout.
Greg Timbers explained he used a switch box of various capacitors and inductor values to be able to subjectively listen to different crossover frequencies and slopes using the actual drivers on the enclosure baffle in a real listening room. Not an acoustic test chamber. This approach helped fast track the more imperial design practice used by JBL back in the 70’s through to the last 10 or 15 years.
Only recently have highly advanced programs like Loudsoft Crossover enabled manufacturers to quickly assess different crossovers in a test and learn approach.
I thought this would help bring to life your excellent thread.
Making up such a crossover board would be a fun project and as outlined above a very useful tool in a diy loudspeaker builders workshop.
lan
Do check out Tony Gee's Black Box design.
It uses almost the same drivers, there is a lot of info in the file, plus the complete xover with components.
I have to admit the xover is over complicated and not to my liking and if you simulate it in XSim it will probably produce a horrible result but it is still educational.
Here's the file (plus check all other projects when you go to Humble Home Hi-Fi).
https://www.humblehomemadehifi.com/archive.html
Cheers
Stan
Excellent thread - so thanks @AllenB . Should have read it before embarking on my active crossover project but have done so now. So better late than never as the saying goes. I really love the pragmatic combination of data and experience approach. Get things in the right ballpark with the theory and then fine-tune / tweak using listening experience. This seems just right since the theory can rapidly become a rabbit hole to disappear down and so much of the real-world implementation is idiosyncratic .... the room, how it is furnished, where you sit etc. etc. And DIY is just fun. I learn a lot by doing.
To answer your question the answer is No unfortunately.
Some preliminary objectives need to be established so you have compatible drivers.
I had a quick look at your driver selection and proposed crossover schematic.
1. You need to know what the combined sensitivity of both your woofers is going to be relative to your mid range and tweeter.
2. individual low pass crossover filters on each woofer are in appropriate unless you intend to use a first order low pass filter on the lower woofer to compensate for a known baffle step requirement.
3. Your woofer low pass crossover frequency look too high. Aim for 500- 600 hertz and 5,000 hertz for the midrange and tweeter.
If you wish l can assist you in your own thread or send me a pm. I can also run through how to use REW to get real measurements on your project. Non of this is difficult with some guidance.
lan
Thanks for your response and everyone elses response here. Really appreciate it.
I attached the data sheets of each individual driver for everyone to have look. Getting the sensitivity on par with other drivers isn't an issue here as you probably seen in my first attempt that came out pretty flat.
Crossing the woofers at 700Hz or even 1000Hz shouldn't be an issue as it starts getting directional above 1k and the cone starts breaking up at 3k. These would work perfectly well in a 2-way. However crossing them lower might present an issue with power distribution and overloading the mid.
Crossing the mid at 5k where it obviously gets directional would be detrimental compared to around 3k and the tweeter can go much lower than that.
However I've been further playing with this theoretical design and came up with this
0° response
30° response
Phase and impedance
Now the only issue I have is the impedance dropping below 2 Ohms in the 1.3k region and would as for a hint how the address this issue as I don't know how happy my amp would be with this.
This is a completely different design from my point of view. I'm planning to use different drivers and build floorstanders with double RS225P-8. Something similar to Troels Gravesen's Ekta Grande or at least in that direction.
Bit I will try do disect his crossover design and try to learn something from it. Especially the impedance part of it looks fabulous.
Last edited:
It’s a bit hard to know without looking at the voltage drives.
l would try removing R3 and C5. I am not sure the purpose of these parts?
l would try a simple RC zobel simulation to flatten the combined parallel woofer impedance to four ohms from 700-3000. Do not force the impedance below four ohms.
Then test and fit an ideal 2nd order filter for the actual impedance. Ie 4.2 ohms. The response curve should then be close without modifying the L or C which can upset your system impedance. Try a Linkwitz 2nd order filter as they sum flat with electrical phase reversal.
See what this does to the system impedance?
Your idea seems to be almost a fill in driver between the woofers and the tweeter.
Can you outline your driver layout and your baffle dimensions?
lf you intend to use a tall narrow baffle say 1200mm x 240mm you might simulate the possibility of baffle step in Jeff Bagleys Excel spread sheet. If you need baffle step “Compensation” l suggest using one of the woofers as a low pass helper woofer below the baffle step frequency ie 300 hertz. Use a simple 1st order filter.
You will get much lower THD at low frequencies, more manageable system impedance of around eight ohms in the crossover region and still good system sensitivity.
You could use an LCR series trap to flatten the mid drive resonance for a better high pass crossover function and possibly a RC zobel
to flatten the impedance for the low pass filter.
If your software can do it try a mid bandpass filter because your crossover points are relatively close together. It will possibly give some passive filter gain which can be useful to trim out the response level matching.
See how it goes.
l would try removing R3 and C5. I am not sure the purpose of these parts?
l would try a simple RC zobel simulation to flatten the combined parallel woofer impedance to four ohms from 700-3000. Do not force the impedance below four ohms.
Then test and fit an ideal 2nd order filter for the actual impedance. Ie 4.2 ohms. The response curve should then be close without modifying the L or C which can upset your system impedance. Try a Linkwitz 2nd order filter as they sum flat with electrical phase reversal.
See what this does to the system impedance?
Your idea seems to be almost a fill in driver between the woofers and the tweeter.
Can you outline your driver layout and your baffle dimensions?
lf you intend to use a tall narrow baffle say 1200mm x 240mm you might simulate the possibility of baffle step in Jeff Bagleys Excel spread sheet. If you need baffle step “Compensation” l suggest using one of the woofers as a low pass helper woofer below the baffle step frequency ie 300 hertz. Use a simple 1st order filter.
You will get much lower THD at low frequencies, more manageable system impedance of around eight ohms in the crossover region and still good system sensitivity.
You could use an LCR series trap to flatten the mid drive resonance for a better high pass crossover function and possibly a RC zobel
to flatten the impedance for the low pass filter.
If your software can do it try a mid bandpass filter because your crossover points are relatively close together. It will possibly give some passive filter gain which can be useful to trim out the response level matching.
See how it goes.
This is worth looking at, and it's fortunate because I was meaning to ask whether Toaster79 had considered the baffle step. Obviously if it isn't then when it is it would change the crossover.
These components are sometimes used to create baffle step compensation, but they may just be for shaping. There may be other ways to do the same thing. The downside is that if not done carefully they can introduce problems with low impedance, it's important that there is a resistor in there.
With a couple of quick simplifications I've been able to bring them a little closer. Sometimes it is just a matter of the values you use and sometimes you need a more complex circuit, but I usually look toward the former first.
I think improvements could be made by improving the measurements, ensuring they are free of disturbances.
I think improvements could be made by improving the measurements, ensuring they are free of disturbances.