Hi,
please critique my first attempt at a crossover for a Living Voice Auditorium clone I'm working on.
The speaker is a 2-way MTM with 2 x Vifa C17 6.5" midwoofers and initially a Scanspeak D2608/93000 tweeter. I got better measured response with Morel CAT 378 tweeters, so I used those in this simulation. I measured all drivers according to the Vituixcad procedure, however I only have on-axis response for the Morel since I just got it. So I have to leave out power response and directivity for now. The last image shows the response of the drivers without filter, for reference.
I'm a little wooried about the broad overlap area between 2khz-5khz, is this any reason for concern? I think the phase tracking in this area looks pretty good though.
Also would the large impedance hump centered around 3khz cause any issues?
please critique my first attempt at a crossover for a Living Voice Auditorium clone I'm working on.
The speaker is a 2-way MTM with 2 x Vifa C17 6.5" midwoofers and initially a Scanspeak D2608/93000 tweeter. I got better measured response with Morel CAT 378 tweeters, so I used those in this simulation. I measured all drivers according to the Vituixcad procedure, however I only have on-axis response for the Morel since I just got it. So I have to leave out power response and directivity for now. The last image shows the response of the drivers without filter, for reference.
I'm a little wooried about the broad overlap area between 2khz-5khz, is this any reason for concern? I think the phase tracking in this area looks pretty good though.
Also would the large impedance hump centered around 3khz cause any issues?
Attachments
I've heard these speakers a couple times; they sound really great, especially with valves!
I cannot help on your questions but these links may be interesting. I believe the Living Voices originally evolved from a clone of an old Dali Vifa MTM.
They now use custom Scanspeak drivers for woofers and tweeter. Did you model the 9500? It has a very even frequency response.
http://www.troelsgravesen.dk/C17mk4.htm
https://www.diyaudio.com/community/threads/troels-c17-modifications.355516/
I cannot help on your questions but these links may be interesting. I believe the Living Voices originally evolved from a clone of an old Dali Vifa MTM.
They now use custom Scanspeak drivers for woofers and tweeter. Did you model the 9500? It has a very even frequency response.
http://www.troelsgravesen.dk/C17mk4.htm
https://www.diyaudio.com/community/threads/troels-c17-modifications.355516/
Hi motokok, thank you for chiming in 🙂
Yes I have read that article and thread on diyaudio some time ago, interesting stuff!
And you are absolutely right, Kevin Scott based the LV speakers on Dali 104.
The latest version, the R25A uses the same Vifa C17, but manufactured by Scanspeak since the original was discontinued years ago.
But the tweeter is a standard Scanspeak D2608. I tried this as well, but it has a really poor measured response. I can only imagine this is caused by diffraction.
Don't know how Kevin Scott gets around this, if he does.. I don't know how audible it is.
Yes I have read that article and thread on diyaudio some time ago, interesting stuff!
And you are absolutely right, Kevin Scott based the LV speakers on Dali 104.
The latest version, the R25A uses the same Vifa C17, but manufactured by Scanspeak since the original was discontinued years ago.
But the tweeter is a standard Scanspeak D2608. I tried this as well, but it has a really poor measured response. I can only imagine this is caused by diffraction.
Don't know how Kevin Scott gets around this, if he does.. I don't know how audible it is.
Farfield means measurements at 1 meter distance in this case. Also made nearfield measurments of woofers and port that were merged with the farfield measurements in vituixcad. Tweeter is only farfield.
Yes please Allen, I would definitely appreciate some help with the filter. The results I have obtained so far can be seen on the attached images in my first post.
I think the spl and phase response looks okay? But I had a couple of questions about the overlap and impedance.
Phase is not necessarily supposed to be at 0 degrees between ways, as Joe D'Appolito once explained. However there's more going on and until you have all your measurements I can imagine you want to get some results with what you have.
What did you want to know about the overlap?
Did you try moving your gate slider and finding the reflections to see whether 5ms is reasonable? If the responses are not right then this can be more difficult to do. What did you want to know about the impedance?
In any case, if you'd like to share your files then we might look for a filter that suits.
What did you want to know about the overlap?
Did you try moving your gate slider and finding the reflections to see whether 5ms is reasonable? If the responses are not right then this can be more difficult to do. What did you want to know about the impedance?
In any case, if you'd like to share your files then we might look for a filter that suits.
Thank you Allen.
What do you mean by "Phase is not necessarily supposed to be at 0 degrees between ways"? This confuses me.
Regarding the broad overlap area between 2khz-5khz, I'm concerned this will compromise SQ because of interference between the drivers. It may not be an issue, but it certainly doesn't look like textbook rolloff to me.
Yes, I did look at the impulse response in REW to determine the amount of gating.
About the impedance. I am planning on using a tube amp with this speaker, so I was a little worried about the large hump at 3khz. Would this be an issue?
I'd be happy to share my files later today when I get home. Which files do you need?
Do you have a reason for feeling that phase should line up? In Joe's case this is about vertical dispersion.
In any case I'm just suggesting you focus more on the response and less on phase at this early point.
Difficult to say but a better filter is worth looking at.
Unless you can get vituixcad to bundle up your files then let's start with the impedance and response files. I can implement the filter myself and move from there.
Most tube amps have low output impedance and aren't affected by a varying impedance. Is yours one that does? In that case you can use a conjugate. We can look at that after the crossover.
Being new to speaker design, it's just what I have picked up from reading articles and posts on this forum.
That was my next question actually, the importance of frequency response vs phase response 🙂
That sounds great Allen, I will do this later today.
The amp is an Air Tight ATM-300 SET with 300B output tubes, I can't find anything on their website about output impedance.
Hi Allen, I attached frd and zma files. Also included the measurements for Scanspeak D2608 tweeter if you want to have a go at that as well.
Unfortunately, I only have on-axis measurement for Morel CAT378.
I'm just trying to get mine the same as yours before I move. It's not the same and I'm not sure why and I'll need to check that with you. I may post screenshots later. Now I notice you have a woofer impedance measurement that looks like it's for two combined yet you are using two woofers in the sim and this creates a problem. We can still find a filter just for now, but will need to convert the values.
Later on, if it were me I'd measure the two woofers together as one to get proper measured vertical interference and not use the vituixcad 2 drivers with +/-140mm separation because that is meant as a simulation for when people don't have measurements.
Later on, if it were me I'd measure the two woofers together as one to get proper measured vertical interference and not use the vituixcad 2 drivers with +/-140mm separation because that is meant as a simulation for when people don't have measurements.
What do you see different?
I've used the woofer *.zma on both, 0 degrees on all while ignoring the other angles for the time being.
I've used the woofer *.zma on both, 0 degrees on all while ignoring the other angles for the time being.
That is correct Allen, the impedance measurement is for both woofers connected in parallell.
To correct for this, all the way down on the drivers tab, scaling is set to 2 for both woofers.
This was something I asked about in the Vituixcad thread, and was told to use both woofers separately.
I mean, even when I have the measurements, I still have to tell the software where they are located relative to eachother? At least that is my understanding.
The tweeter level is higher in your simulation, as well as the lower part of the woofer response.
I guess this is because of the scaling option I mentioned above?
Yes, that was it. Ok, this doesn't have the right feel and I suspect the response is not what it should be. Does 1ms look different to 5ms? Does your measurement look anything like the factory plot?
You see, vituixcad can simulate the interaction of the two drivers by considering their spacing. On the other hand you can measure that very effect by measuring the two at the same time. It's one or the other.
Many people think that.. but it should only be that way under certain circumstances or you'll double deal the effect.
You see, vituixcad can simulate the interaction of the two drivers by considering their spacing. On the other hand you can measure that very effect by measuring the two at the same time. It's one or the other.
Hi Allen, thank you very much for your work so far!
It is very obvious where the first reflection is in the impulse window, if that is what you mean 1ms vs 5ms. Do you think there may be errors in the measurements? This is my first time following this procedure, so I may very well have done something wrong here.
There are no published factory plots to compare with, and I haven't found any reviews with measurements either.
Anyway, the LV uses the Scanspeak D2608, not the Morel CAT378. Maybe you could try to do something with that as well?
I agree, the simulation below does not look very good. Do you feel this is an improvement over my simulation? The crossover region sure looks better on your simulation though, not the large overlap I got on mine.
But to my understanding, the scaling in Vituixcad is a measure taken to avoid double dealing this effect?
I had to read the last part a couple of times, but I think I got it 🙂 So basically you are saying, as long as I use my measurement protocol with Vituixcad I will be okay, and a less powerful software without this feature would require measuring both drivers at the same time?
What about your measurement setup? I might have a look for factory plots.
Now I see you're talking about the electrical effects. I'm talking about three dimensional effects. The scaling and the driver positioning in vituixcad are not connected in any way.
I do things the way I do because I know how to get more accurate results than the vituixcad simulated positioning.. however this doesn't mean you musn't use it. It's important that you see the option is there so you have versatility in how you do things. If you see it as a more powerful thing then you may be missing the point.