Dayton WWMT results

[Rick Craig]

As promised in an old thread I'm reporting back on the Dayton WWMT (crossover designed by Dennis Murphy) that an owner invited me to test. It turns out everything was constructed per the plans but the mids were wired opposite phase from the schematic. After reversing the phase I tested the speakers on-axis and off-axis (vertical and horizontal measurements). I also ran nearfield tests of the mid and woofers.

We also measured the individual drivers without the crossover in place. I don't know if anyone has posted on-axis versus off-axis horizontal curves for the three? crossover designs with this combination. I would suggest looking at them before choosing one of the alternatives.

[Dennis H]

I don't see any graphs in your post.

[---k---]

Rick,
That is great news that you were able to get it fixed. I assume it sounded pretty good after that. It would be great if you could post testing results. To the best of my knowledge, no one has throughly measured them.

[Jim Holtz]

We also measured the individual drivers with out the crossover in place. I don't know if anyone has posted on-axis versus off-axis horizontal curves for the three? crossover designs with this combination. I would suggest looking at them before choosing one of the alternatives.

Hi Rick,

I have the on axis for the Seas 27TBFC/G version of the WWMT's that Dennis measured which I've attached. I don't have the RS28 version. I'm interested in hearing about what you found with the speakers you measured.

Dennis did do in cabinet development of the design in his normal fashion. I believe the other versions by Jon and Curt were sims.

Jim

[Rick Craig]

Attached are the measurements. The "HTNEAR" file is a nearfield measurement of the woofers without the crossover in place. This indicates a good box tuning. The "HTMNEAR" measurement is a nearfield of the midrange (again, no crossover). As you can see the response is ragged with a large dip centered about 95hz. I also measured an impedance peak in that area. When the mid was tested outside of the enclosure the impedance was normal. I checked the subenclosure and it was fine - joints sealed and the cabinet constructed per the drawing for the rectangular version. The only thing I can think of is that the subenclosure design is creating a resonance because we also altered the stuffing and there was no significant change. The dip at 160hz also shows up in the impedance plot and with the shallow slope on the crossover these two areas are only 10-15db down electrically so this could be an audible problem.

The "DMHTONV" file is the vertical on-axis response taken from on-axis with the midrange to up above the tweeter. The dip in the 3-4.5K area is at the upward angles with the flattest response actually on axis with the midrange. This indicates that the crossover lobe is pointed somewhat downward and seated listening will be better on this axis versus standing. I also observed quite a bit of overlap with the midrange and tweeter in that range which could be reduced and thus increase the vertical listening window. Other than the 3-4.5K area everything looks good.

The "DMHTOFF" curves are the on-axis response versus the horizontal response 30 degrees off-axis. The response looks good except there's a 2-3db peak off-axis in the 2.2K-3.2K area. Listening tests confirmed the problem and I would advise a crossover change to correct it. I tried adding some resistance on the tweeter to make the response smoother in the 2.5K-10K range but it didn't help the off-axis peak.


Regards,

Rick

[Rick Craig]

I don't see any graphs in your post.

I've added a new post containing the graphs.

[Jed]

The tweeter looks a bit bright, depending on how far away you took the measurements. Should be an easy fix with L-pad adjustments. That might also help reduce the broad overlap between the mid and the tweeter. I wonder how much the Dayton drivers have changed over the years. Maybe that explains the response variables of your measurements versus the original measurements used to design the crossover.

[Dennis H]

Thanks for the measurements, Rick.

[Rick Craig]

The tweeter looks a bit bright, depending on how far away you took the measurements. Should be an easy fix with L-pad adjustments. That might also help reduce the broad overlap between the mid and the tweeter. I wonder how much the Dayton drivers have changed over the years. Maybe that explains the response variables of your measurements versus the original measurements used to design the crossover.

I believe it was about 1.5M distance. An L-Pad won't fix the problem with the overlap - already modeled that. The tweeter (without the filter) also has a peak in the 2K area off-axis so both the BP and HP sections need to be revised.

I agree that the drivers could be to blame but there's no way to know for sure unless some of the other owners can measure their speakers. I also would be interested to see if the subenclosure problem is present in the other systems out there.

[augerpro]

post edited to prevent further embarrassment :B

[Dennis H]

https://www.htguide.com/forum/showthread.php?t=15323

I doubt Rick's using an uncalibrated mic with his CLIO measurement $ystem.

https://web.archive.org/web/20090328010410/http://estore.websitepros.com/1736754/Categories.bok?category=CLIO

[---k---]

Just FYI, Rick Craig is Rick from Selah Audio.

http://selahaudio.com/

Rick, thanks for the measurements and thoughts.

[augerpro]

ops:








:rofl:

[Jed]

Rick,

Are you planning on doing a crossover design for this driver combination? If not, any chance you could send the driver measurements to me so I could play with a crossover design? Or maybe we could compare notes/topologies for fun. Would be a neat exercise.

Jed

[cjd]

Ahh, this problem is familiar! I think I have a similar issue with the single RS150 in my HT in-walls, regarding some resonance. Curious. Curious indeed. I'll have to review the box designs to see what similarities we have in the mid enclosure, try to narrow this down.

C

[Rick Craig]

Rick,

Are you planning on doing a crossover design for this driver combination? If not, any chance you could send the driver measurements to me so I could play with a crossover design? Or maybe we could compare notes/topologies for fun. Would be a neat exercise.

Jed

Actually I'm configuring the woofers in series so that wouldn't help you create a solution for everyone here. Maybe someone else has measurements they can send you and we can compare the mid/tweeter crossover. After thinking about it maybe it's an issue of the mid's inductance but I can't say for sure without seeing some other measurements.

[Jed]

Actually I'm configuring the woofers in series so that wouldn't help you create a solution for everyone here. Maybe someone else has measurements they can send you and we can compare the mid/tweeter crossover. After thinking about it maybe it's an issue of the mid's inductance but I can't say for sure without seeing some other measurements.

Couldn't they just configure the woofers in series also? Not following you, unless you got the 4ohm woofers or something like that.

Jed

[dawaro]

Here are on & off axis measurements Dennis did on the RS28 version along with his comments:
"I think it's quite good now, though it's so much more sensitive than the other speakers I have on hand it's difficult to do A-B comparisons. Maybe it's still a tad edgy, but it certainly sounds integrated. The bass is quite astounding.
I did a tone sweep, and was causing major house trembles at
25 Hz. That's about the lower limit. The overall bass presentation is
a little muddy and accentuated, but the prototype cabinets are 1/2" mdf with, I believe, no bracing and very little sound treatment. I've attached a measurement I took on axis and a little to the left horizontally of on-axis. There's a diffraction dip at 4 k, which fills in to the left, and gets worse to the right (the tweeter is mounted
off-center). I think a larger radius on the baffle would help here.
I tried a design with 4th order slopes, but I didn't have much luck.
It's hard to shape the midrange correctly and still maintain a fairly
low mid-tweeter crossover. The current crossovers are around 380 Hz
and 2100 Hz. Cheers, dennis."

[dawaro]

I have the individual driver measurements Dennis made for the RS28 version but they exceed the attachment size here so if you are interested in them shoot me a message. Below are the screen shots & Dennis' comments on the measurements and driver placement settings in LspCad:
"Here are my measurements. I'm sending the frd and zma files,
plus screen captures of my lspcad settings for the woofer, mid, and tweeter. These will show the driver locations relative to the tweeter, and the dX setting which in effect shows the adjustment I had to make in the depth of the acoustic centers of the woofer and mid to get the predicted response of the three drivers running simultaneously without crossover components to match the actual measured output of all three
drivers running free at the same time. All of the driver measurements
were made on the tweeter axis at a distance of 1.1 meters. The design listening distance is 4.5 meters. My approach introduces some error in the design-distance response because the two woofers are treated as one woofer located midway between the two woofers with a diameter of zero.

lspCAD cannot adjust for changes in the response at longer listening distances that are due to changes in the off-axis response of 8" drivers as you move back. But this is a minor problem at the frequencies involoved, and the advantage is that the mic can stay in one position and you don't introduce measurement errors due to changes in the mic position as you measure each driver on its own axis. Anyhow, I was able to get very reliable predictions from the measurements at the measuring
distance of 1.1 meters. This is a complicated subject, so just tell
people to e-mail me with questions. Cheers, dennis."
Mid Imp/Freq

Images not available

Tweeter Imp/Freq

Images not available

Woofer Imp/Freq

Images not available

[dawaro]

Rick,

Are you planning on doing a crossover design for this driver combination? If not, any chance you could send the driver measurements to me so I could play with a crossover design? Or maybe we could compare notes/topologies for fun. Would be a neat exercise.

Jed

I have e-mailed you the files for the WWMT. If you would like the WMTW files and supporting e-mails let me know.

[dawaro]

Attached are the measurements. The "HTNEAR" file is a nearfield measurement of the woofers without the crossover in place. This indicates a good box tuning. The "HTMNEAR" measurement is a nearfield of the midrange (again, no crossover). As you can see the response is ragged with a large dip centered about 95hz. I also measured an impedance peak in that area. When the mid was tested outside of the enclosure the impedance was normal. I checked the subenclosure and it was fine - joints sealed and the cabinet constructed per the drawing for the rectangular version. The only thing I can think of is that the subenclosure design is creating a resonance because we also altered the stuffing and there was no significant change. The dip at 160hz also shows up in the impedance plot and with the shallow slope on the crossover these two areas are only 10-15db down electrically so this could be an audible problem.

This is the enclosure that I built Dennis to use for the measurements. I am not sure is the different sub-enclosures being used for the mids could cause a different loading of the driver. Is this the box your customer had?

Image not available

[dawaro]

Sometimes it really pays to be a pack rat and save all those old e-mails I guess!

[Jed]

I'll take a look at the data... thanks for emailing it to me.

Jed

[Rick Craig]

Couldn't they just configure the woofers in series also? Not following you, unless you got the 4ohm woofers or something like that.

Jed

They are the 8-ohm drivers. I found that wiring them in series and manipulating the impedance with a quasi-zobel gave me better balance with the mid than running them in parallel. It also resulted in a more moderate impedance around 6 ohms.

[Rick Craig]

This is the enclosure that I built Dennis to use for the measurements. I am not sure is the different sub-enclosures being used for the mids could cause a different loading of the driver. Is this the box your customer had?

Image not available

Sealed version but I believe the same layout for the sub enclosure. If you look at the impedance Dennis took for the mid it shows two impedance peaks in the low end and should only be one.

[Rick Craig]

Today I went back and found the impedance plot I had for the midrange in the earlier design I had tested. The results were almost exactly the same so it seems the 2-3K off-axis peak isn't a driver problem unless the tweeter is to blame. I can't compare the tweeter information because the earlier design used a Seas and the one I'm working on has the Dayton RS28.

[Jed]

Attached are the measurements. The "HTNEAR" file is a nearfield measurement of the woofers without the crossover in place. This indicates a good box tuning. The "HTMNEAR" measurement is a nearfield of the midrange (again, no crossover). As you can see the response is ragged with a large dip centered about 95hz. I also measured an impedance peak in that area. When the mid was tested outside of the enclosure the impedance was normal. I checked the subenclosure and it was fine - joints sealed and the cabinet constructed per the drawing for the rectangular version. The only thing I can think of is that the subenclosure design is creating a resonance because we also altered the stuffing and there was no significant change. The dip at 160hz also shows up in the impedance plot and with the shallow slope on the crossover these two areas are only 10-15db down electrically so this could be an audible problem.

The "DMHTONV" file is the vertical on-axis response taken from on-axis with the midrange to up above the tweeter. The dip in the 3-4.5K area is at the upward angles with the flattest response actually on axis with the midrange. This indicates that the crossover lobe is pointed somewhat downward and seated listening will be better on this axis versus standing. I also observed quite a bit of overlap with the midrange and tweeter in that range which could be reduced and thus increase the vertical listening window. Other than the 3-4.5K area everything looks good.

The "DMHTOFF" curves are the on-axis response versus the horizontal response 30 degrees off-axis. The response looks good except there's a 2-3db peak off-axis in the 2.2K-3.2K area. Listening tests confirmed the problem and I would advise a crossover change to correct it. I tried adding some resistance on the tweeter to make the response smoother in the 2.5K-10K range but it didn't help the off-axis peak.


Regards,

Rick

Dawaro sent me Dennis's files to play around with in LSPcad and I did just that tonight. I set all the driver coordinates just as Dennis specified. I didn't notice too much that would confirm the peakiness of the RS28a that you were seeing, in his measurements- but the off axis response dip was there, like in your measurements and the midrange/treble looks a bit hot. Other than that nothing would indicate that his RS28a model is similar to your RS28a. My conclusion is that the RS28a samples differ and that is what is causing the response variables you noticed. It has been pretty well documented that there has been a consistency problem with the RS28a. Looks like this current info may extend that trend- unfortunately.

[dawaro]

In case you did not see it Dennis posted a reply on the PE Board about this:
"That looks like a lot to take on. But one thing is pretty clear after looking at Rick's measurements. All of my designs are optimized for a listening distance of 10-12 feet, not one meter. In a multidriver 3-way like this one, that can make a substantial difference. I'll send you two plots--one at 1 meter, which resembles Rick's. and one at 4 meters, which is much smoother."

[cjd]

It is also worth nothing here that the RS28A and RS28AS are not drop-in replacements for each-other. At least, not always, but maybe sometimes. Definitely differences between these, though.

[jkrutke]

In case you did not see it Dennis posted a reply on the PE Board about this:
"That looks like a lot to take on. But one thing is pretty clear after looking at Rick's measurements. All of my designs are optimized for a listening distance of 10-12 feet, not one meter. In a multidriver 3-way like this one, that can make a substantial difference. I'll send you two plots--one at 1 meter, which resembles Rick's. and one at 4 meters, which is much smoother."

I don't think Rick mentioned what his measurement distance was, but it's pretty obvious it wasn't 4 meters. I noted the same tendencies in most of the tall projects I've done, including the ZDT3. The design distance is 2.5 meters. Measured or modeled from 1 meter on the tweeter axis, the response goes to hell. The effect could be worse in Dennis' design though because of the larger mid and tweeter diameters. When the ZDT3 is reposted with the new tweeter, I will actually be posting some near and far models and measurements to show the difference.

Getting useful measurements at 2.5 meters or more in a typical home environment is troublesome. Too much of the room shows up, and windowing that out turns the whole response into mush. The far field model and ears become more important for these taller 3 way designs.

Some unique facts about my eight RS28 measurements: under 10kHz, the response curves have all been extremely consistent. Somewhere around here I have an image with 8 RS28's overlaid, and all are within 1dB under 10kHz, and that's including the defective ones. 10 to 20kHz was another story, with as much 15 dB variance. Some had no breakup and no dip.

[Rick Craig]

I don't think Rick mentioned what his measurement distance was, but it's pretty obvious it wasn't 4 meters. I noted the same tendencies in most of the tall projects I've done, including the ZDT3. The design distance is 2.5 meters. Measured or modeled from 1 meter on the tweeter axis, the response goes to hell. The effect could be worse in Dennis' design though because of the larger mid and tweeter diameters. When the ZDT3 is reposted with the new tweeter, I will actually be posting some near and far models and measurements to show the difference.

Getting useful measurements at 2.5 meters or more in a typical home environment is troublesome. Too much of the room shows up, and windowing that out turns the whole response into mush. The far field model and ears become more important for these taller 3 way designs.

Some unique facts about my eight RS28 measurements: under 10kHz, the response curves have all been extremely consistent. Somewhere around here I have an image with 8 RS28's overlaid, and all are within 1dB under 10kHz, and that's including the defective ones. 10 to 20kHz was another story, with as much 15 dB variance. Some had no breakup and no dip.

About 1.5M was the measurement distance. These are in a loft and the listening distance isn't that far away (maybe 7-8 ft.?). I also performed some other measurements to better capture the woofer/mid integration and everything looked fine.

I didn't find anything that would make me think that the tweeter is defective or out-of-spec; in fact, it measured pretty much like I had expected. The measurement distance is far enough way to accurately capture the crossover area and the off-axis peak is very audible. The overall balance was acceptable, maybe a little bright due to the side walls being fairly close to the speakers in this particular setup. That could easily be changed but the 2-3K peak would need more extensive modification to the crossover.

[Dennis Murphy]

As my lspcad simulatons show, it does make quite a difference whether the listening position is 1 M or 4 meters--and moving out to 1.5 meters gives you pretty much the same result asw 1 M. As for the off-axis peak, I always check that carefully, and I didn't see it in my measurements or hear it in my ears. Unfortunately, I had a computer crash and lost all my old screen captures of this design (and all the other ones I've done).

[Rick Craig]

As my lspcad simulatons show, it does make quite a difference whether the listening position is 1 M or 4 meters--and moving out to 1.5 meters gives you pretty much the same result asw 1 M. As for the off-axis peak, I always check that carefully, and I didn't see it in my measurements or hear it in my ears. Unfortunately, I had a computer crash and lost all my old screen captures of this design (and all the other ones I've done).

I lost a drive once so I know how much that can hurt. Thanks for weighing in.

[Dennis Murphy]

I lost a drive once so I know how much that can hurt. Thanks for weighing in.

Thanks for the condolences. I believe this design was done over 3 years ago, so I can't be absolutely sure what the off-axis response looked like. I just know I do check for those kinds of peaks. I'm not doubting you saw and heard one. I may have that crossover lying around, and maybe I'll be able to take some more measurements. I am pretty sure my mic is calibrated correctly, because I always seem to replicate raw driver measurements and system responses taken by manufacturers and other DIY'ers. But, then, there's always Murphy's law......

[Rick Craig]

Thanks for the condolences. I believe this design was done over 3 years ago, so I can't be absolutely sure what the off-axis response looked like. I just know I do check for those kinds of peaks. I'm not doubting you saw and heard one. I may have that crossover lying around, and maybe I'll be able to take some more measurements. I am pretty sure my mic is calibrated correctly, because I always seem to replicate raw driver measurements and system responses taken by manufacturers and other DIY'ers. But, then, there's always Murphy's law......

I failed to mention that I told the owner that if I didn't do my own design work that you would be one of the few people I would trust to design a crossover for me. I don't want to imply that you did something wrong.