I've got a 8" woofer and a CD + horn. I've used DSP with delay to time align drivers before, and liked the result. So for this passive build I thought I'd try to get the acoustic centers aligned physically.
So to investigate what that would require, I set up some boards between 2 tables so I could hang my woofer and then my tweeter at the same vertical location, facing up. I set up a mic above, pointed down and didn't move it between measurements. Then I measured with REW with a timing reference channel, and that told me the distance (acoustically) to each driver. Then I just gotta make my baffle so that I can line 'em up...
Looking at the woofer and CD+horn I thought it would be pretty close. The measurements suggest otherwise... =\ I've got the Tymph AULA02014 8" woofer and the Celestion CDX1-1446 on a JBL / Selenium HM17-25. I have a thread for those drivers here: https://www.diyaudio.com/community/threads/tymphany-aula02014-0006-pe-buyout.416365/ but this discussion is to focus on baffle structure ideas.
REW thinks the tweeter is 70 mm behind the woofer. To line them up I need to bring the tweeter forward rather than back, and by a fair 2.8 inches! Now, I've seen the dual chassis type of horn build, where there's a box with a big woofer in it, and then the big multi-cell horn leans over the front of the box, and that achieves a horn-forward type of build. I kind of had a more visually discreet aesthetic in mind. The HM15-25 isn't a huge monster, I had imagined mounting it in the box / on the baffle. XO is expected to be ~ 2 kHz I was hoping to keep C2C distance from becoming enormous, but also don't want to put a big diffraction object right near the woofer.
Has anyone made / seen an elegant horn mouth fwd of woofer type build? I'm looking for ideas.
- A
So to investigate what that would require, I set up some boards between 2 tables so I could hang my woofer and then my tweeter at the same vertical location, facing up. I set up a mic above, pointed down and didn't move it between measurements. Then I measured with REW with a timing reference channel, and that told me the distance (acoustically) to each driver. Then I just gotta make my baffle so that I can line 'em up...
Looking at the woofer and CD+horn I thought it would be pretty close. The measurements suggest otherwise... =\ I've got the Tymph AULA02014 8" woofer and the Celestion CDX1-1446 on a JBL / Selenium HM17-25. I have a thread for those drivers here: https://www.diyaudio.com/community/threads/tymphany-aula02014-0006-pe-buyout.416365/ but this discussion is to focus on baffle structure ideas.
REW thinks the tweeter is 70 mm behind the woofer. To line them up I need to bring the tweeter forward rather than back, and by a fair 2.8 inches! Now, I've seen the dual chassis type of horn build, where there's a box with a big woofer in it, and then the big multi-cell horn leans over the front of the box, and that achieves a horn-forward type of build. I kind of had a more visually discreet aesthetic in mind. The HM15-25 isn't a huge monster, I had imagined mounting it in the box / on the baffle. XO is expected to be ~ 2 kHz I was hoping to keep C2C distance from becoming enormous, but also don't want to put a big diffraction object right near the woofer.
Has anyone made / seen an elegant horn mouth fwd of woofer type build? I'm looking for ideas.
- A
Adam,
I've seen many horn mouths forward of the woofer, none that I would consider "elegant" in appearance.
As another idea, rather than using a physical offset, you could phase align the drivers at the crossover point as this design using the HM17-25 does ~2 kHz does:
https://www.mtg-designs.com/diy-speaker-plans/vbs-6-2
Smooth phase alignment and frequency response could be achieved with a variety of crossover and polarity combinations, the MTG Design above is just one example.
Art
I've seen many horn mouths forward of the woofer, none that I would consider "elegant" in appearance.
As another idea, rather than using a physical offset, you could phase align the drivers at the crossover point as this design using the HM17-25 does ~2 kHz does:
https://www.mtg-designs.com/diy-speaker-plans/vbs-6-2
Smooth phase alignment and frequency response could be achieved with a variety of crossover and polarity combinations, the MTG Design above is just one example.
Art
Hi Art, thanks for having a look!
Yes, I think with modern XO sim software I can probably jiggle the response curves to get good summing for the overlap region. I do have this impression that actual alignment (as opposed to asymmetric crossover phase effects) makes for better listening. Not sure if that's just my kooky thought or if it's generally accepted.
As an aside, I've seen that MTG designs website before, I think the 6" and 10" GRS woofers he uses (6PT-8 and 10PT-8) look like good fun for cheap. I hadn't been thinking about the horn he was using, so thanks for bringing that back to my attention.
Mount it flat on the baffle for a better look, and push it into shape in the XO? Or make some kind of spacer for time alignment, and deal with an ugly face and maybe some sort of diffraction biz? I'll have to meditate on it. I wonder if I can cad + print a spacer that doesn't look hideous? What if instead of a spacer to lift the horn out, I did a small horn / angled recess to push the speaker back?
Anyone else out there seen a clever means to handle this?
mmm-hmm. yeah.
Yes, I think with modern XO sim software I can probably jiggle the response curves to get good summing for the overlap region. I do have this impression that actual alignment (as opposed to asymmetric crossover phase effects) makes for better listening. Not sure if that's just my kooky thought or if it's generally accepted.
As an aside, I've seen that MTG designs website before, I think the 6" and 10" GRS woofers he uses (6PT-8 and 10PT-8) look like good fun for cheap. I hadn't been thinking about the horn he was using, so thanks for bringing that back to my attention.
Mount it flat on the baffle for a better look, and push it into shape in the XO? Or make some kind of spacer for time alignment, and deal with an ugly face and maybe some sort of diffraction biz? I'll have to meditate on it. I wonder if I can cad + print a spacer that doesn't look hideous? What if instead of a spacer to lift the horn out, I did a small horn / angled recess to push the speaker back?
Anyone else out there seen a clever means to handle this?
I would suggest you look for yourself, because even when aligned you'll see excess group delay as a result of the filters.
I made two crossovers in vituixcad to compare (it's the same active or passive). The first was a time aligned LR 2nd order at 1kHz.
The second crossover is shown on this attachment. I've adjusted one of them to 4th order to compensate for the woofer being brought forward by 1/4 wavelength relative to a waveguided tweeter.
The two crossovers are overlayed on the response plot. The Q factors and frequencies have been juggled a little as shown to get close to the same as the standard LR crossover.
Phase is shown on the right (red and green). They are similar to the LR over a sufficient band before they deviate. What is of interest is the group delay. The blue trace on the right is the standard crossover group delay and the grey trace is the electrically compensated one.
Which would you prefer?
Just curious if you were expecting to align the mouth of the horn and the front plate/voice coil of the woofer. It's often the throat of the horn that's considered the acoustic origin.
TAD does a decent job of it - or at least gets close enough that the basics are adaptable.
https://www.ebay.com/itm/145809519911
Similar approach to the problem here
pinterest.com/pin/393431717426170010/
Last edited:
Hmmm, interesting plots, AllenB, and a good idea to rummage around in some simulation. I'll do something similar... what is going on with the green trace for phase, why does it stop at 3k?
The numbers I discussed (tweeter is 70mm behind woofer) are based on acoustic measurement with the drivers mounting features at the same plane. So whatever counts as the actual acoustic source, I let REW decide what that time / distance was. As I talk about bringing the CD + horn forward 70mm that is the relationship between the two drivers mount / screw holes.
Yes, those look nice. They push the woofer back with some horn action. I wonder how much of those woofer horns are about acoustic gain, and how much they are about acoustic alignment.
We shall see if I can discern anything useful from simulation with theory-perfect drivers, per AllenB. It would make life easier if I could convince myself I don't need an offset...
Thanks both, for your comments!
- A
For what it's worth, lining up the mouths to maintain clean radiation, if that's what it takes, is what I find important. The fact that group delay is similar to or even better this way (ie, the time alignment issue) may not be a significant factor, although it could become one in extreme cases. If faced with such a case I still wouldn't accept damaging the radiation, if possible.
The response for that filter has fallen so low by that frequency that it has become irrelevant, and vituixcad chooses to not show phase. In fact, the response is low enough for practical purposes even before that frequency.
My guess is it's more about getting the acoustic origins aligned without doing too much damage to the rest of the acoustics, especially in the second example. Though even the impact of more aggressive geometry is debated. There are plenty of horns out there with large steps in geometry, so some don't seem to care too much.
The TAD is large enough that it's likely working into the midbass. Whether that makes much difference in a typical hi-fi scenario seems debatable. Joseph Crowe has a page about "small" front horns.
https://josephcrowe.com/blogs/news/design-study-front-horn-loaded-reflex-cabinet
I've mentioned this dozens of times, but I personally believe that one of the reasons that the Danley speakers image the way they do, is because the woofers, midranges and tweeters are all located at different distances from the listener, but they're in phase.
The net effect is that the depth of the soundstage is nebulous; it's hard to tell if the back wall of the room is ten feet away or fifteen feet away, especially with the lights out.
If you have DSP and understand how to use filters to achieve this trick, it might be worth a look. Depending on how things are configured, it doesn't necessarily have to be a Unity horn. LeCleach used a similar idea with his speakers, note the tweeter is significantly behind the midrange:
The net effect is that the depth of the soundstage is nebulous; it's hard to tell if the back wall of the room is ten feet away or fifteen feet away, especially with the lights out.
If you have DSP and understand how to use filters to achieve this trick, it might be worth a look. Depending on how things are configured, it doesn't necessarily have to be a Unity horn. LeCleach used a similar idea with his speakers, note the tweeter is significantly behind the midrange:
Right! I think there is something to this. But I'm struggling with how to show / find this property in in simulation. (I'm using XSIM)
You can look at a normal (theory perfect) 2LR with a driver inverted - and the XO region is in-phase. So (if we accept that in-phase has an advantage) my thought is that it's not just the overlap region, it's wider bandwidth where drivers-in-phase is useful. But what result do you look at to see this phenomenon? (I mean, apart from how you've wired the drivers, which is not an acoustic measurement)
I've been thinking that you'd look at phase. But every time I look at phase in a sim, you can only keep it flatish if your drivers are aligned - the extra delay from one driver makes phase start to wrap. Which made me think that time alignment was important. But realistically, there's always delay, you're always some distance away from the speaker - which would look in a phase result like wrapping (const delay, shrinking period). So maybe seeing this sort of wrapping in the phase result isn't the indicator I should be looking for.
I saw this same sort of quandry when I was looking at XRK's Harsch XO, which is asymmetric and includes a physical offset. Ok, it makes good phase through the XO but then the driver delay shows up as wrap in the phase result. When you look at a measurement and subtract a const delay to get a flattish phase, you only get 1 distance (time) to subtract...
group delay? impulse / step?
Short answer: You figure out the length of a wavelength at the crossover point, and then you move the midrange CLOSER to the listener by a distance that's equal to the delay of the filter. For instance, if the crossover point is 1350Hz (which is ten inches long) and the high pass is third order, move the midrange forward by about five inches (2nd order is 180 degrees, or one half a wavelength of 1350Hz.)
Long answer: You're likely wondering why it's moved by half a WL, not 3/4 of a WL. Your answer is here, note the PDF from LeCleach in particular. https://www.diymobileaudio.com/threads/crossovers-a-step-further.163839/page-6
Wow I am getting old, it was Jeff Markwart and it was just spread frequency not asymmetric. Sorry for the confusion
Any of these
Which is why there's normally a feature to unwrap phase.
You have to decide whether time alignment is something you care about. Some do, some don't.
Step response can certainly show some of the final effects of lack of time alignment in a multi-way horn.
https://www.stereophile.com/content/klipsch-la-scala-al5-loudspeaker-measurements
Sorry, instead of unwrapping phase, that should have been removing delay/excess phase. This is often done automatically by measurement systems.
Why not mount the woofer inside the box, on the back of the baffle and create a nice smooth exitHi Art, thanks for having a look!
mmm-hmm. yeah.
Yes, I think with modern XO sim software I can probably jiggle the response curves to get good summing for the overlap region. I do have this impression that actual alignment (as opposed to asymmetric crossover phase effects) makes for better listening. Not sure if that's just my kooky thought or if it's generally accepted.
As an aside, I've seen that MTG designs website before, I think the 6" and 10" GRS woofers he uses (6PT-8 and 10PT-8) look like good fun for cheap. I hadn't been thinking about the horn he was using, so thanks for bringing that back to my attention.
Mount it flat on the baffle for a better look, and push it into shape in the XO? Or make some kind of spacer for time alignment, and deal with an ugly face and maybe some sort of diffraction biz? I'll have to meditate on it. I wonder if I can cad + print a spacer that doesn't look hideous? What if instead of a spacer to lift the horn out, I did a small horn / angled recess to push the speaker back?
Anyone else out there seen a clever means to handle this?
Ya, move the woofer back with a little horn action. That's an option, it'll have an FR impact that could be fixed with a filter.
I played around in XSIM looking at different things, then I remembered the Harsch XO, XRK made a thread about it here... https://www.diyaudio.com/community/threads/s-harsch-xo.277691/
He says:
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.
So tweeter delay is what I got, and I'm looking for good phase / step response so this seems like a good thing to try. Half of 2khz delay is 85 mm, so I'm gonna set the woofer forward by an additional sheet of ply with a little woofer inset. We'll see if it works!
I played around in XSIM looking at different things, then I remembered the Harsch XO, XRK made a thread about it here... https://www.diyaudio.com/community/threads/s-harsch-xo.277691/
He says:
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.
So tweeter delay is what I got, and I'm looking for good phase / step response so this seems like a good thing to try. Half of 2khz delay is 85 mm, so I'm gonna set the woofer forward by an additional sheet of ply with a little woofer inset. We'll see if it works!