I bought these drivers back in end August I think it was, and I've finally got a box for 'em. I've measured up the in box response and played around with the results in XSIM. I still have to order XO parts, and do a final sand and finish on the boxes. But I thought I'd share where I'm at.
I kicked out a thread to discuss the drivers when I ordered them, it's https://www.diyaudio.com/community/threads/tymphany-aula02014-0006-pe-buyout.416365/
Woofer (buyout, NLA) is this 8" tymphany pro audio with a big honker magnet
https://www.parts-express.com/Tymph...-Paper-Cone-Woofer-8-Ohms-299-2265?quantity=1
I got 'em for $17 each!
Compression driver is Celestion CDX1-1446
https://www.parts-express.com/Celes...-Compression-Driver-8-Ohm-299-2257?quantity=1
This one is also a buyout but I think the same driver is available regular production.
I've got it on this JBL Selenium HM17-25 https://www.parts-express.com/Selen...al-Horn-60x40-1-3-8-18-TPI-264-308?quantity=1
I have modeled it up in 30 L and ported at 60 Hz. Software wanted to suggest smaller / higher, but I'm trying to get this thing to dig down far enough to cross to my sub. I have 3D printed a port according to geometry in this Harmon paper about optimizing ports. https://www.diyaudio.com/community/attachments/harman-port-study-pdf.893225/
I was trying to conceptualize how I'd put this speaker together, and I thought of trying to time-align the drivers, so I measured the acoustic offsets. The horn+tweeter came out 70mm behind the driver (mounting features in the same plane), which was more than I was expecting. I had a little thread to consider options about that https://www.diyaudio.com/community/...nique-for-horn-out-front.419357/#post-7851615 eventually I remembered that @xrk971 has a crossover topology that wants the tweeter a fair bit delayed WRT the woofer, and that's the Harsch XO, he's got a thread about it here - https://www.diyaudio.com/community/threads/s-harsch-xo.277691/. I was considering a 2kHz XO, and the delay according to the Harsch thing was 85mm, so another sheet of plywood for the woofer to sit on was just about right.
So that's about how I dreamed this one up! Here's how it looks...
Ah... every thread is better with pictures. I've got roundover I think it's 1" on the vertical corners. Boxes made of arauco ply 3/4" from Home Depot, I'm still trying to decide how I'm gonna finish them. I like the lines from the ply on front.
Ground plane, on-axis (spkr tilted down to point at the mic) 2m distance acoustic measurements look like this
THD is good on the woofer, and FR is fine to 3k, but I think that is a low baffle step from 100 to 500? Speaker is 12.5" wide. Anyway that shift ended up being around the upper of the two impedance peaks in low woofer response, that made it kinda tricky to correct passively. FR on the tweeter seems pretty nice, but measurement looks like kinda high 2nd order HD.
I've been fiddling with the XO sim to get something I like.... here's where I'm at.
Initial idea was to have the cross at 2k, but in the course if trying to make it work I have floated up from there. So I think my offset is actually a little too much, and my phase peak is a bit larger than it might have been.
If you go for this sort of XO, it is (quoted from XRK's thread)...
1. Set the low pass filter for the woofer as a 4th order Butterworth at central frequency, fc for the XO centerpoint.
2. Set the high pass filter for the tweeter as a 2nd order Bessel at fc.
3. Set the delay of the tweeter equal to 1/2 of the period of one cycle at fc.
4. Use all positive phase on woofer and tweeter.
Here's what I've learned: 2nd order -acoustic- roll off is pretty slow! 2nd order electrical is normally fine, but when you are looking at the acoustic rolloff that happens naturally that doesn't leave much space to put in electrical protection. So I ended up at higher F than I was initially targeting. It looks like 2nd order electrical on the CD but it's pretty low Q, that inductor is only helping out at lower frequencies. Impedance compensation on the CD can help you get the shape theory wants.
I think I've attached the XSIM file in case anyone wants to play along.
I'm looking forward to hearing these, but I haven't pulled the trigger on XO parts yet. Every time I buy parts a couple days later I'm like "you know what I SHOULDA done...?" so I'm still meditating on it.
What do you think?
- Adam
I kicked out a thread to discuss the drivers when I ordered them, it's https://www.diyaudio.com/community/threads/tymphany-aula02014-0006-pe-buyout.416365/
Woofer (buyout, NLA) is this 8" tymphany pro audio with a big honker magnet
https://www.parts-express.com/Tymph...-Paper-Cone-Woofer-8-Ohms-299-2265?quantity=1
I got 'em for $17 each!
Compression driver is Celestion CDX1-1446
https://www.parts-express.com/Celes...-Compression-Driver-8-Ohm-299-2257?quantity=1
This one is also a buyout but I think the same driver is available regular production.
I've got it on this JBL Selenium HM17-25 https://www.parts-express.com/Selen...al-Horn-60x40-1-3-8-18-TPI-264-308?quantity=1
I have modeled it up in 30 L and ported at 60 Hz. Software wanted to suggest smaller / higher, but I'm trying to get this thing to dig down far enough to cross to my sub. I have 3D printed a port according to geometry in this Harmon paper about optimizing ports. https://www.diyaudio.com/community/attachments/harman-port-study-pdf.893225/
I was trying to conceptualize how I'd put this speaker together, and I thought of trying to time-align the drivers, so I measured the acoustic offsets. The horn+tweeter came out 70mm behind the driver (mounting features in the same plane), which was more than I was expecting. I had a little thread to consider options about that https://www.diyaudio.com/community/...nique-for-horn-out-front.419357/#post-7851615 eventually I remembered that @xrk971 has a crossover topology that wants the tweeter a fair bit delayed WRT the woofer, and that's the Harsch XO, he's got a thread about it here - https://www.diyaudio.com/community/threads/s-harsch-xo.277691/. I was considering a 2kHz XO, and the delay according to the Harsch thing was 85mm, so another sheet of plywood for the woofer to sit on was just about right.
So that's about how I dreamed this one up! Here's how it looks...
Ah... every thread is better with pictures. I've got roundover I think it's 1" on the vertical corners. Boxes made of arauco ply 3/4" from Home Depot, I'm still trying to decide how I'm gonna finish them. I like the lines from the ply on front.
Ground plane, on-axis (spkr tilted down to point at the mic) 2m distance acoustic measurements look like this
THD is good on the woofer, and FR is fine to 3k, but I think that is a low baffle step from 100 to 500? Speaker is 12.5" wide. Anyway that shift ended up being around the upper of the two impedance peaks in low woofer response, that made it kinda tricky to correct passively. FR on the tweeter seems pretty nice, but measurement looks like kinda high 2nd order HD.
I've been fiddling with the XO sim to get something I like.... here's where I'm at.
Initial idea was to have the cross at 2k, but in the course if trying to make it work I have floated up from there. So I think my offset is actually a little too much, and my phase peak is a bit larger than it might have been.
If you go for this sort of XO, it is (quoted from XRK's thread)...
1. Set the low pass filter for the woofer as a 4th order Butterworth at central frequency, fc for the XO centerpoint.
2. Set the high pass filter for the tweeter as a 2nd order Bessel at fc.
3. Set the delay of the tweeter equal to 1/2 of the period of one cycle at fc.
4. Use all positive phase on woofer and tweeter.
Here's what I've learned: 2nd order -acoustic- roll off is pretty slow! 2nd order electrical is normally fine, but when you are looking at the acoustic rolloff that happens naturally that doesn't leave much space to put in electrical protection. So I ended up at higher F than I was initially targeting. It looks like 2nd order electrical on the CD but it's pretty low Q, that inductor is only helping out at lower frequencies. Impedance compensation on the CD can help you get the shape theory wants.
I think I've attached the XSIM file in case anyone wants to play along.
I'm looking forward to hearing these, but I haven't pulled the trigger on XO parts yet. Every time I buy parts a couple days later I'm like "you know what I SHOULDA done...?" so I'm still meditating on it.
What do you think?
- Adam
Hehe. At this stage expect to want to change things around and get handfuls of basic components.
Finished, I just went with clear poly. I tried a bunch of different tinted options and didn't like any of them. Pine just doesn't like stain, I guess. XO built temp / external with wagos.
Here's the XO as ordered, with available part values.
The green FR trace above is the validation measurement. Looks like I'm getting a little rise at 10k. Are my tweeter LPAD resistors a little out of spec? Or am I just poor at getting my tweeter axis pointed at my mic from 2m away? It was a bit windy as I was measuring, maybe just wind noise?
In REW with phase...
I did a horizontal off axis set of measurements. They came up a little odd...
These are done as ground planes, with the speaker tilted down to point at the mic. This makes it a little tricky to rotate them for off-axis measurements. The vertical axis of the speaker about which you'd need to rotate is no longer just vertical, it has the tilt to point the tweeter at the mic. So you can't just twist it on the ground. So I propped up a ramp to put the speaker on, so I could point it at the mic and then twist it on that axis. But you can't just have the speaker at the end of the ramp. When you twist it, one side moves back and the other side moves forward. So it has to be back from the end of the ramp so that you have some space for the corner to move forward into.
So these measurements have a little unusual shape around 1k Hz. I think this is due to the discontinuity in front of the speaker at the end of the ramp.
So I think that's not real. But you can see the drop as you move off axis, there doesn't seem to be anything crazy going on. (this one, not so much extra 10k)
I kind of wonder what the vertical off-axis looks like, but haven't taken it.
Crossover is a big pile of components! I hope I can fit it in the box. Wire it up permanent like is the next thing. Then some lower stands for these, since the speakers are taller than the pair they'll stand in for.
Here's the XO as ordered, with available part values.
The green FR trace above is the validation measurement. Looks like I'm getting a little rise at 10k. Are my tweeter LPAD resistors a little out of spec? Or am I just poor at getting my tweeter axis pointed at my mic from 2m away? It was a bit windy as I was measuring, maybe just wind noise?
In REW with phase...
I did a horizontal off axis set of measurements. They came up a little odd...
These are done as ground planes, with the speaker tilted down to point at the mic. This makes it a little tricky to rotate them for off-axis measurements. The vertical axis of the speaker about which you'd need to rotate is no longer just vertical, it has the tilt to point the tweeter at the mic. So you can't just twist it on the ground. So I propped up a ramp to put the speaker on, so I could point it at the mic and then twist it on that axis. But you can't just have the speaker at the end of the ramp. When you twist it, one side moves back and the other side moves forward. So it has to be back from the end of the ramp so that you have some space for the corner to move forward into.
So these measurements have a little unusual shape around 1k Hz. I think this is due to the discontinuity in front of the speaker at the end of the ramp.
So I think that's not real. But you can see the drop as you move off axis, there doesn't seem to be anything crazy going on. (this one, not so much extra 10k)
I kind of wonder what the vertical off-axis looks like, but haven't taken it.
Crossover is a big pile of components! I hope I can fit it in the box. Wire it up permanent like is the next thing. Then some lower stands for these, since the speakers are taller than the pair they'll stand in for.
It looks like you have a Speakon connector on the back of your cab. If it's a 4 conductor why not connect the CD and woofer to it and make your crossover external? That way you can make adjustments easily. AllenB is more knowledgeable than I, so I'd trust his advice, but I'm thinkin' that's a lot of components ...
It is a 4 conductor speakon, and the XO is external this moment. I'd like to make it internal just to keep things neat, but if it puts up too much of a fight external may be the way.
I also think that it is a lot of components! The Harsch concept has target shapes I had to beat the response into. Conceptually it' not super crazy... woofer is 3rd order with a notch (baffle step functions are spread across the LP and the notch), tweeter is low Q 2nd order, with an L pad and a notch, and some impedance compensation to keep the HP shape matched to theory. A few components double up to get the values I wanted. Perhaps not the most simple / economical, but I wanted to try the Harsch thing out, it sounded like a cool idea.
It's kind of a lot of inductor$. Lookin at it all on a piece of cardboard with wagos holding it together it looks a mess. Maybe I'll start a thread dedicated to gnarly crossovers...
The sound is pretty good. They're in line to displace my current mains, which I've got set as active XO / DSP'd so they have an easy time sounding flat and being time-aligned. These do about as good a job at that passively, so that's pretty cool. These drivers are pretty high output, so they do have a sense of powerful output capacity as well. More so I expect, when I get them into the main setup high passed and integrated with a sub.
I also think that it is a lot of components! The Harsch concept has target shapes I had to beat the response into. Conceptually it' not super crazy... woofer is 3rd order with a notch (baffle step functions are spread across the LP and the notch), tweeter is low Q 2nd order, with an L pad and a notch, and some impedance compensation to keep the HP shape matched to theory. A few components double up to get the values I wanted. Perhaps not the most simple / economical, but I wanted to try the Harsch thing out, it sounded like a cool idea.
It's kind of a lot of inductor$. Lookin at it all on a piece of cardboard with wagos holding it together it looks a mess. Maybe I'll start a thread dedicated to gnarly crossovers...
The sound is pretty good. They're in line to displace my current mains, which I've got set as active XO / DSP'd so they have an easy time sounding flat and being time-aligned. These do about as good a job at that passively, so that's pretty cool. These drivers are pretty high output, so they do have a sense of powerful output capacity as well. More so I expect, when I get them into the main setup high passed and integrated with a sub.
I wouldn't say this is a lot of components.. it takes whatever it takes.
Harsch is known for having untidy vertical response. This scheme is all about one axis.
You can see it doesn't add perfectly below the cross on axis, so you know it will add at some other angle...
The temporary / verification build of the XO is not robust... maybe I'll try to take verticals after the XO implementation has been fortified.
A good crossover might deliberately involve a partial sum on the listening axis.. but what you show here is on the other side of the phase rotation. It's normal for the Harsch scheme and it varies across the crossover band.
What it could do is render individual reflections to have a different tone to others so they become more distinct and difficult to ignore.. as well as potentially putting a kink into the power curve.
John Kreskovsky is one of our local experts on the characteristics of the transient perfect crossovers.
What it could do is render individual reflections to have a different tone to others so they become more distinct and difficult to ignore.. as well as potentially putting a kink into the power curve.
John Kreskovsky is one of our local experts on the characteristics of the transient perfect crossovers.
I am interested in the transient aspect, and went with the Harsch when I found out the horn for the CD put it almost the correct distance behind the woofer. But I'd enjoy a chance to learn of other means to accomplish similar goals... I'll look around for some of his posts!
It was an interesting segue around the turn of the century which led to us learning that what looks right and what sounds right aren't always the same thing.
There are two parts to this. Firstly, the audibility of phase variations comes down to group delay thresholds. It doesn't need to be 'flat' but it should be within a certain limit. This gives us plenty of room to work within for our conventional crossovers.
Second is that power and off-axis response are things that can be audible and probably shouldn't be the factors left to chance in any good compromise.
There are two parts to this. Firstly, the audibility of phase variations comes down to group delay thresholds. It doesn't need to be 'flat' but it should be within a certain limit. This gives us plenty of room to work within for our conventional crossovers.
Second is that power and off-axis response are things that can be audible and probably shouldn't be the factors left to chance in any good compromise.
I’d leave well enough alone as the response above 10k is fairly consistent at 0,15, and 20…..nothing a simple twist of a treble control couldn’t tame…..better to have it to take away than not and have to boost it back.
crumbles to dust
I have gone back and forth between thinking about things as phase or group delay or polarity. Do you have a link to the discussion? Or perhaps it was carved onto rocks or clay tablets...
This is my thinking also, though I may do more measurements to try to get to the bottom of it.
Consider group delay as thinking of phase in the context of the frequencies around it. In other words, the slope of the phase curve.
John K and I differ in whether we accept lobe tilting, but I don't see that as relevant to this topic. It's the part of this post about the subjective qualities of a transient perfect crossover that I wanted to show - https://www.diyaudio.com/community/...er-to-time-phase-coherent.345669/post-5996714
Well, I've been listening to this speaker on and off. And it's been giving me a hard time. It sounds fine, but I can't listen for long. After 10 or 15 min it makes my ears start to ring. I've hooked it up to eq to try to fix, I've pulled the XO and gone to active. But it doesn't really sound like a FR issue to me, that part seems fine-ish. It just hasn't wanted to stop assaulting my ears. I haven't seen a smoking gun in the measurements (tho I see 2nd harmonic THD higher than I'd like with these CDs).
But here's one thing that I've noted: the horn and the CD don't screw all the way together. Here's a pic..
This looks like between 1 and 2 mm of gap to me, the space is too narrow to get my ruler down into. L and R do the same thing, the horn screws on to the CD, and then starts getting tight and I can't get it to screw to a nice flush point. I've got a Dayton waveguide on a different CD (with a metal threaded insert, this horn is molded plastic all the way), and the two screw together reasonably loose until they come to a positive stop, but this combo gets progressively tighter and comes to a stop. I've swapped the parts around, the HM17-25 is the issue.
I understand the CD to horn transition to be pretty important in horn performance. Could this be causing the unpleasant effect I've noticed?
I saw this thread: https://www.diyaudio.com/community/threads/selenium-hm17-25.156439/
And it looks like @Pano maybe had the same issue screwing it down all the way, but folks in that thread seemed to think things were basically alright with the thing. Perhaps I'll have to think about filling the gap (how will I do that...) but I do sort of wonder if I'm even on the right track.
But here's one thing that I've noted: the horn and the CD don't screw all the way together. Here's a pic..
This looks like between 1 and 2 mm of gap to me, the space is too narrow to get my ruler down into. L and R do the same thing, the horn screws on to the CD, and then starts getting tight and I can't get it to screw to a nice flush point. I've got a Dayton waveguide on a different CD (with a metal threaded insert, this horn is molded plastic all the way), and the two screw together reasonably loose until they come to a positive stop, but this combo gets progressively tighter and comes to a stop. I've swapped the parts around, the HM17-25 is the issue.
I understand the CD to horn transition to be pretty important in horn performance. Could this be causing the unpleasant effect I've noticed?
I saw this thread: https://www.diyaudio.com/community/threads/selenium-hm17-25.156439/
And it looks like @Pano maybe had the same issue screwing it down all the way, but folks in that thread seemed to think things were basically alright with the thing. Perhaps I'll have to think about filling the gap (how will I do that...) but I do sort of wonder if I'm even on the right track.
I did not find any difference either measured or audible by smoothing the gap. But it's easy enough to try with some putty.