New 7" driver tests

[Mark K]

Something to digest.
New 7" tests.
some interesting tests on the peerless hds exclusive and more.
...

http://www.markk.claub.net/

[cotdt]

boy is the XG18 a distortion generator! back when i used them i knew that it had a ton of distortion due to that grainy sound and it was really, really audible, but there were no distortion graphs at the time. now i realize just how horrible they are!

so how 'bout them XG18 5-tone tests? i'd like to see just how terrible they are.

It would be nice to modify the Ushers a bit to remove that 1-2kHz dip.

[cotdt]

Also did you guys know that recent Klippel tests show that the Usher actually has 10mm of linear xmax? And the Seas Excels only have like 3mm... o_O

the ushers are on sale right now too, from PE.

[WillyD]

Excellent info, thanks a lot for the tests. I found the information to be very accessible for a novice like myself.

[mazurek]

Very interesting. I have been planning and slowly working towards for ages, a 3 way with: 2 x RS265 10"HF, 18W4531G, and Peerless HDS Tweeter, with crossover at 80-100Hz and 2000Hz, all crossed LR4 acoustic. The tweeter will be loaded in a very very shallow waveguide, roughly the shape of the 4531G cone.

These test results are very good for me. Now I am wondering if I should push things up, cross the RS265 higher, and move to a 15M4531G, given how well the SS are testing throughout the spectrum. Of course, I already tuned my box using 'The Edge' to correlate nicely with the 18W.

Its too bad, I wrote a matlab script to do Zaph type HD testing, but these results were very interesting, now I'll have to change it to incorporate this. I think I'm going to make it so it outputs movies with all the plots as frequency changes.

Anyways, thanks for the thorough testing.
Lee

[JoshK]

Wanna share your script? I am a matlab user.

[cjd]

So, have you run the RS150 through these tests?

I'm curious which is better suited for midrange duty, assuming crossover frequency is intelligently chosen either way. Because, of course, we used the RS180 as a midrange (crossing ~1.6k) in the big WWMTM project and I quite like it, but wonder if I would have been better off with the RS150 for a bit more extension out of the mid on the top (1800 maybe).

I'm pondering some experimentation with lower order slopes mid-tweet too.

C

[capslock]

Ok, some thoughts here:

1. Importance of saturation: to the best of my knowledge, the W18E and EX have exactly the same pole piece geometry and differ just in the size of the magnet, so the iron near the gap will be farther into saturation in the case of the EX. The EX comes out ahead slightly in most measurements, but I am not sure whether you can draw the conclusion that more into saturation is always better.

2. In swept HD tests done by HobbyHifi magazine, the Scan Revelator had lots of broadband second harmonic distortion from 500 Hz to 2 kHz (on the order of 1% at 90 dB/ 1 m). This is typical of flexing hysteresis in paper cones. The Usher (actually the non-plugged version, but it has the same cone) was not much better. I am surprised that this does not show up in these measurements.

3. The problem with the XG and XT supposedly is the surround, which generated lots of distortion when it moves more than 1 - 2 mm. The motor should be fine. While I have never taken one apart, my understanding is that it has a heavily undercut pole piece and copper rings. There are measurements of the XT18WH in Klang & Ton magazine, which do show high distortion below 200 Hz, but exceptionally low third order above (below 0.1%) and only moderate second order (below 0.3%). Based on those measurements, it would make an exceptional midrange.

More to follow on why I think motor design can explain most of Mark's findings.

[jkrutke]

Hi Mark, thanks for the tests.

As usual, your XG18 results don't look too good. So, why the difference? Though I still conur that the XG18 is far from the best in the group. I guess I can pull that XG18 MTM design off my website now, I don't think it's likely to be built anymore. :B

When I demo'ed that XG18 MTM in Chicago a couple years ago, people generally thought it sounded great, possibly one of the best at the show. But a few things began to sink in since then that lead me to the shocking conclusion that people are completely unable to pick the better performing speaker in a crowd of speakers... and in fact the ones that "stand out" are probably the ones with higher distortion, both linear and non-linear. I've seen enough examples of this to pretty much prove it. One was when Dave Littel brought a Seas CA15 MTM to the DC DIY meeting, it had a rollercoaster response curve as measured by Dennis Murphy and the CA15 is known to have comparitively high distortion. Now, when it was compared to the ZD5 that had a ruler flat response curve and low distortion, guess which one the crowd prefered. Yup, the CA15 MTM. Now, if I ever get my hands on a Visaton Ti100 and Hi-Vi planar tweeter that everyone seemed to love at one of the DIY meetings, I would not be surprised at all if these drivers turned out to be pieces of junk.

Back to the subject at hand, "why the difference". I've got a few theories, all related to the difference of testing on a baffle and free air.

1) A driver with more motor noise will test subtantially worse free air than it does on a baffle. The rear motor noise is reflected rather than absorbed, and eventually makes it back to the mic to be registered as high order spectral content.

2) When tested free-air, a driver with a lower overall weight will test worse than a driver with higher weight unless the drivers are securely anchored by the magnet or frame.

3) When tested free-air, a woofer's excursion requirements for a given SPL go way up on the low end, and will reflected poorly on those with lower Xmax.

Some of this became clear to me a while back when I tested a Tangband neo subwoofer. It had some operating noise (airflow and clicking) and was very lightweight. It did in fact measure much better bolted on a baffle, and even better yet when mounted in an enclosure.

It's still obvious that the XG18 makes a better midrange than a woofer. I've seen this myself - when offloading the lower bass to a woofer, sound quality takes a large step up. If it had a yellow cone, we could all make B&W 801 clones.

The Ushers do look nice, I have to get myself some of those to play with. They might do nice in my waveguide TMM.

[jkrutke]

There are measurements of the XT18WH in Klang & Ton magazine, which do show high distortion below 200 Hz, but exceptionally low third order above (below 0.1%) and only moderate second order (below 0.3%). Based on those measurements, it would make an exceptional midrange.

That almost perfectly describes the results I get for both the XT and XG18. Any chance you can scan and post them?

[JonMarsh]

When I demo'ed that XG18 MTM in Chicago a couple years ago, people generally thought it sounded great, possibly one of the best at the show. But a few things began to sink in since then that lead me to the shocking conclusion that people are completely unable to pick the better performing speaker in a crowd of speakers... and in fact the ones that "stand out" are probably the ones with higher distortion, both linear and non-linear.

I agree with the psychology of this; it's why TV's arrive (in most cases) with the most gawd awful video settings that have to be undone to make them watchable; they're trying to make them stand out on a showroom floor, and catch the eye. Mostly they just give me a headache.


Your theories about why there are different measurement results probably have a lot of vailidity, which is one reason I don't measure free air, but on a good sized baffle.

Still, everyone complained when I couldn't replicate your low distortion results.

And since the other drivers DO measure better under those conditions as well as others, well, it's not hard to understand why the results are interpreted in a given way by most.

I agree with Eric and was concerned the first time I saw the series that the surround design was a problem- in more than one way.

Lastly, even with a simple "easy" test on a good size baffle, these were still among the poorest performing 7's I measured, so I would exclude them even for dipole midrange use.






Compare with HDS 830883...






I wanted to like them, and I wanted to use them in a dipole midrange based on the other tests I'd heard about.

~Jon

[TacoD]

Magazine is copyrighted, and the figures are not that large.

John, Mark do you burn in the woofers which you measure? I know that HobbyHifi/ Klang & Ton are saying they do.

[---k---]

But a few things began to sink in since then that lead me to the shocking conclusion that people are completely unable to pick the better performing speaker in a crowd of speakers...

I agree. I only wonder how long it takes a person to pick out the better (meaning measures flat) speaker in the quiet of his own home.

[jkrutke]

Magazine is copyrighted, and the figures are not that large.

John, Mark do you burn in the woofers which you measure? I know that HobbyHifi/ Klang & Ton are saying they do.

True about the copyright, but that hasn't stopped be from collecting as many scans of their tests as I can over the years. 8O

Klang & Ton are such cool magazines, along with Hobby Hifi, that I really wish we could get them here in english.

Does anyone know which month/year has the Vifa XT tests in it? I tried to buy a couple pdf magazine downloads before and failed, but I'll give it a shot again.

Woofers I run for 10 hours at full Xmax, face down on spacers. (it occasionally un-does some cone sag) I also give 1 hour to cool down. Mids and tweets I typically don't break in for much more than a few minutes of noise though them.

[JonMarsh]

Something to digest.
New 7" tests.
some interesting tests on the peerless hds exclusive and more.
...

http://www.markk.claub.net/

Geat work as usual, Mark! :T Should be required reading for midwoofer selection...


When I get some actual projects done, I want to go back and try the 8 tone tests one some other drivers, too. It's a bit like Dean Jensens Spectral Contamination test, which would be another good test signal- even if it doesn't tell the whole story- (the Extremis looks good with that test, but has other issues. Of course, it's another light driver example, but then the energy storage issues in cone and surround and limited ability of surround to support the Xmax are other matters).

~Jon

[JonW]

Mark-
Super job. :T Thanks for the very helpful data. I'm trying to pick drivers for an MT project. I've been going back and forth over Zaph's excellent testing. I'm really drawn to the very low distortion of the Usher 8945P but a little concerned about the response dip at ~1400 Hz. Maybe next best in distortion is the SS 18W. A wee bit more distortion, but no response dips. And now your testing also brings those 2 drivers to the top o' the heap. Very nice to see a second set of data. I'm a newbie to all this but I really appreciate your, Zaph's, and JonM's testing. I'd be nowhere without it.

By the way, there seems to be a slight uptick in the impedance curve of the SS 18W at ~900 Hz (Zaph's plot). Is this anything to be concerned about?

[JonMarsh]

That's the classic SS break in cone behavior, helps deliver the rising response, also delivers a blip in energy storage phenomena.

[Jim Holtz]

[QUOTE=jkrutke]

When I demo'ed that XG18 MTM in Chicago a couple years ago, people generally thought it sounded great, possibly one of the best at the show. But a few things began to sink in since then that lead me to the shocking conclusion that people are completely unable to pick the better performing speaker in a crowd of speakers... and in fact the ones that "stand out" are probably the ones with higher distortion, both linear and non-linear. I've seen enough examples of this to pretty much prove it. One was when Dave Littel brought a Seas CA15 MTM to the DC DIY meeting, it had a rollercoaster response curve as measured by Dennis Murphy and the CA15 is known to have comparitively high distortion. Now, when it was compared to the ZD5 that had a ruler flat response curve and low distortion, guess which one the crowd prefered. Yup, the CA15 MTM.

Hi John,

I think I've finally figured out why we can't get you to attend DIY events. A suggestion. Why not share the knowledge at one of the DIY events of listening you've developed from all of your driver testing? The majority of folks listen to what they like regardless of the measurements. Ultimately audio is a subjective thing. Measurements and testing are where designers start but they even listen in the end during the voicing process. Anyway, my point is, critical listening is a learned skill for most. It would go hand in hand with all of the hard work you've put forth testing the drivers.

Now, if I ever get my hands on a Visaton Ti100 and Hi-Vi planar tweeter that everyone seemed to love at one of the DIY meetings, I would not be surprised at all if these drivers turned out to be pieces of junk.

That would be the Iowa DIY event that I co-hosted. If you search for Jed's posts he linked to K&T distortion measurements for the Ti100 and they were pretty darn good. There were a number of experienced folks in attendance that also spoke very highly of the Ti100. Dennis Murphy has since used it in Jim Salk's new V3 speaker that listeners have raved over. It's a pretty sweet mid, IMHO. However, I'm more interested in the TB W4-1337S that seems to be an improvement on the Ti100. I'm eagerly awaiting Jon's testing of this driver and some others he's lined up.

Come and see us next year John. We'd welcome your presence and I think you'd enjoy our event.

Best regards,

Jim

[JonW]

That's the classic SS break in cone behavior, helps deliver the rising response, also delivers a blip in energy storage phenomena.

Got it- thanks. So how bad of a thing is that? Otherwise, the SS 18W is looking like a might appealing driver to me.


Edit: I should also say that picking drivers for a project is not as easy as I thought it might be. Every driver that looks really good has some problem or another. I like the low distortion of the 8945P but am hesitant to use it because of the response dip at ~1400 Hz. There's always something.

[Mazeroth]

Also did you guys know that recent Klippel tests show that the Usher actually has 10mm of linear xmax? And the Seas Excels only have like 3mm... o_O

the ushers are on sale right now too, from PE.

Any links to the Klippel tests?

Also, THANKS MARK! :T

[JonMarsh]

Got it- thanks. So how bad of a thing is that? Otherwise, the SS 18W is looking like a might appealing driver to me.


Edit: I should also say that picking drivers for a project is not as easy as I thought it might be. Every driver that looks really good has some problem or another. I like the low distortion of the 8945P but am hesitant to use it because of the response dip at ~1400 Hz. There's always something.

Well, Jon, you've identified the crux of the problem here- nothing's perfect. Manufacturers have different targets to what they believe is the right resopnse profile, and take different routes to get there. The classic SS break in the response and upwards tilt is rationalized as easing the accomplishment of baffle step compensation, since a single pole will by the cone have a zero inserted in the response by this cone behavior at the point where for an average size baffle for that kind of driver the response will have changed from 4 pi to 2 pi radiation as you go upwards in frequency.

A lot (but not all) paper cone drivers have this characteristic- one reason I like the original 830883 is becuase it's free of it. In it's more extreme forms it imparts a certain "eh" characteristic due to an energy smear, audible on pink noise, and to a good listener on music.

As Mark notes, metal cone drivers have other issues, and should be crossed low and crossed often to get the best results (there's a reason I've been doing these Modula and Arvo and M8ta crossovers in the 1200-1400 Hz region).

A two way system is the right kind of project to take on for your first, but picking the right drivers for the right reason is half the baffle. The SS seem to get worse in this regard as the cones get bigger, but that's to be expected, and the current revleator series are much better than the older Kevlar and Carbon fiber I worked with in the 90s.

In a similar price range I'd look at the C173-T6-95, but it doesn't have quite as great an Xmax, and you've still got a ceramic cone peak to avoid in the 4-5 kHz area, better damped that the Seas or Dayton resonances, for sure, but still of some concern. Not having tested it, I can say the published distortion data looks quite nice up to 1500 Hz, then has a rise in 3rd harmonic from the cone resonsnce, but it's still generally under 1%.

Do you have a budget for your first from scratch project? You could certainly do a lot worse than the Scanspeak. But considering the cost, if ou can find a way to manage a lower crossover frequency, such as 1200-1400 Hz, the Usher could be the way to go.

[augerpro]

It would have been very interesting if you'd tested the 8945A also. If the HD kept it in the top tier with the SS and 8945P then the smoother FR would make it a better choice.

The 8948A would be interesting also. Usher uses it in their top 3way models.

Thanks again for the testing Mark!

[JonW]

Oh my, more complexities are arising:
Mark tested the SS 18W4531 and Zaph tested the SS 18W8531.
Mark tested the Peerless 830889 and Zaph tested the 830883.
Are these drivers pretty much the same thing or should they be treated independently? The specs from the SS web site, for instance, are slightly different.





Jon-

Thanks for the perspectives.

A two way system is the right kind of project to take on for your first, but picking the right drivers for the right reason is half the baffle.

As I’m finding out. It’s interesting, though.

In a similar price range I'd look at the C173-T6-95,

For better or worse, I’m hesitant to pick a driver that does not have any independent test data available. If this is silly, feel free to call me on it.

Do you have a budget for your first from scratch project?

No. Like you, time is my most precious commodity, by far. I’m guessing this project will take me from now until it’s warm enough outside to wear shorts and a t-shirt. And that really seems like a long time away, given that it’s 13 degrees out right now, and dropping. So I’m not going to let, say, an extra $500 in drivers get in the way of trying to come up with something that sounds good. Spread out over 6 months... Heck, if it keeps me away from driving at the race track for one weekend, I'm practically making a profit.

You could certainly do a lot worse than the Scanspeak. But considering the cost, if ou can find a way to manage a lower crossover frequency, such as 1200-1400 Hz, the Usher could be the way to go.

So far, I’ve been mostly reading Zaph’s test data on drivers. Just haven’t had a chance to read through all of yours and Marks in the detail they warrant. But the 3 tweeters that look the nicest so far are the Scanspeak 6600, the Seas Millennium, and the Peerless 810921. I don’t think the Peerless could handle such a low crossover point, the other 2 look like they could. But I also have never tried this before. Nor have I had a chance to read all the available test data carefully.

Hey, as long as I continue with my illusion that I can actually come up with a decent sounding speaker, it’s all good fun.

[cotdt]

I vote for the Usher with 1.2-1.4kHz crossover point. 1.4kHz is a walk in the park for a lot of tweeters, including Seas and Dayton Reference. I bet the Scanspeak can do it too, and would be my top choice since it has more "air". Heck, 1.2kHz won't be a problem either with Jon's elliptical crossovers.

[JonMarsh]

Go for that 66000, even I have the hots for that one... ;^) May retrofit it to the top end of my M8ta's! The Millenium get's buy just fine with a 1400 Hz crossover in SL's Orions. I think the 66000 wouldn't break a sweat either in that mode.

After seeing Mark's write up on the Usher, I may have to use that as my other development target (besides the RS180) for the compat 3 way, though I think the HiVi D6.8 would have less issues with causing diffraction from the tweeter. Decisions, decisions. Always tradeoffs there are, as Yoda would say.

[AJINFLA]

When I demo'ed that XG18 MTM in Chicago a couple years ago, people generally thought it sounded great, possibly one of the best at the show. But a few things began to sink in since then that lead me to the shocking conclusion that people are completely unable to pick the better performing speaker in a crowd of speakers... and in fact the ones that "stand out" are probably the ones with higher distortion, both linear and non-linear. I've seen enough examples of this to pretty much prove it. One was when Dave Littel brought a Seas CA15 MTM to the DC DIY meeting, it had a rollercoaster response curve as measured by Dennis Murphy and the CA15 is known to have comparitively high distortion. Now, when it was compared to the ZD5 that had a ruler flat response curve and low distortion, guess which one the crowd prefered. Yup, the CA15 MTM. Now, if I ever get my hands on a Visaton Ti100 and Hi-Vi planar tweeter that everyone seemed to love at one of the DIY meetings, I would not be surprised at all if these drivers turned out to be pieces of junk.

I think this is exactly what Geddes was referring to in his article http://www.gedlee.com/downloads/Obse...20Thoughts.pdf
Smooth response at varying angles + low distortion = "unexciting", "sterile", "uninvolving" and of course "unmusical" to many.
Chaotic power radiation and generous doses of distortion seems to restore "naturalness" and "musicality" for the numbers-don't-matter-listen-with-your-ears-not-machines crowd. At least in the short term.

cheers,

AJ

[mazurek]

Wanna share your script? I am a matlab user.

Here is my script. I have some projects that won't die, and then I will develop this even further, but it is commented and useable.

[jkrutke]

I think this is exactly what Geddes was referring to in his article http://www.gedlee.com/downloads/Obse...20Thoughts.pdf
Smooth response at varying angles + low distortion = "unexciting", "sterile", "uninvolving" and of course "unmusical" to many.

My main point with that little blurb was that it can take longer than 10 minutes with a few tracks before fatigue sets in and the initial "excitement" of distortion fades. (Assuming of course that distortion isn't so bad that you cringe right away) The rapid fire listening sessions one after another just aren't appropriate for comparison purposes. If I only had 10 minutes, I'd sooner take some measurements and trust them instead.

BTW, I wasn't saying the XG18 was junk and the crowd liked them anyway. It's still one of my favorite drivers regardless of Mark and Jon's measurements. I've had roughly 10 of them come through my house for various projects and I'm well aware of what they sound like and what their limitations are. They definitely sound better in the midrange than most of the other sub $100 drivers that come through here. Basically, to me they sound like my measurements look unless of course my mind is playing tricks on me and making me hear what I see. (it's possible, I don't rule anything out) I'm still a little concerned about Jon's measurement differences since he tests on a baffle also. It could just be level differences.

Still looking for the Klang & Ton XT18 tests. Got the PDF download to work but didn't find issue it was in. It wasn't in #205 with the load of 17cm woofer tests, though I did find the Hi-Vi round planar tweeter in there. It had horrible distortion below 5kHz. It's slightly comforting to know that most of the drivers tested both by myself and K&T show similar results. I have to say that even with the text in German, K&T is still a fun magazine to read. It's 5 times better than speakerbuilder ever was, and AudioXpress mostly a garbage collection of tube amps and related crappy speakers.

John

[JonW]

Go for that 66000, even I have the hots for that one... ;^) May retrofit it to the top end of my M8ta's! The Millenium get's buy just fine with a 1400 Hz crossover in SL's Orions. I think the 66000 wouldn't break a sweat either in that mode.

Yeah, an MT with a Scanspeak 6600 tweeter and a Scanspeak 18W woofer could be interesting. They're both high on my list at the moment. But I won't make any decisions until I've read all of the Jon/Mark/Zaph testings in real detail.

I've found the Scanspeak referred to as the 6600, the 66000, and the 660000. All the same thing? And there is that response dip at ~12,000 Hz.

After seeing Mark's write up on the Usher, I may have to use that as my other development target (besides the RS180) for the compat 3 way, though I think the HiVi D6.8 would have less issues with causing diffraction from the tweeter. Decisions, decisions. Always tradeoffs there are, as Yoda would say.

If you're looking to save some money, the Usher 8945P woofer looks good. As does the Peerless 810921 tweeter. Not the cheapest drivers, not super expensive, but with test results that seem well above average.

I quite like the sound of that HiVi D6.8 in my commerical Totem speakers.

[JonW]

A couple general questions about driver testing and picking drivers for a project:

-I know about Mark’s, Zaph’s, and JonMarsh’s testing. Are there any other testing resources out there I should be looking at?

-Some of the tested drivers seem to have various close relatives out there. For example the Scanspeak 18W4531 and 18W8531. Is it safe to consider the strengths and weaknesses of closely related drivers to be the same?

-And if trying to pick one, would it be wise to generally go with the one of higher impedance. The 18W8531 is 8 ohms and the 18W4531 is 4 ohms. So pick the 8 ohm 18W8531?

[Evil Twin]

Monte Kay, the brother of SL's partner in AudioArtistry, has a site (still?) with some driver tests. Very similar to MarkK's approach and methods, also uses Praxis.

In the case of the Scanspeak, they are just different impedance versions. They will usually be very similar, but it's wise to check T/S parameters and the impedance curves and SPL plots before making a choice. Also, while a 4 ohm driver will give you greater sensitivity in a single driver system, it may contribute to a more difficult load to drive once the interaction with the crossover is accounted for, especially in three way speakers. The Dayton 4 ohm drivers don't have sheilding, and are specifically designed for Automotive.

~Jon

[Evil Twin]

I quite like the sound of that HiVi D6.8 in my commerical Totem speakers.

I've used the HiVi before in a system for my daughter, and may use it in my next set of HT three ways.

[Feyz]

I would expect a little more nonlinear distortion in the upper midrange from the 8 ohm version vs the 4 ohm version. The reasoning behind this is 8 ohm version has a little higher inductance, and since the rest of the motor is same, the same ration of inductance modulation will result in higher distortion in the version having higher inductance. This is an oversimplification because lower impedance version will drive more current, though I didn't check the full specs, if 8 ohm version has lower sensitivity, the current issue may become a wash. In short, the first sentence I said , untill someone tests them and prove me wrong in my guess there.

There was a similar thing with SS 8535 vs 8545 in Mark K's previous tests. 8535 has somewhat better nonlinear distrotion, and 8535 has 38mm voice coil, 8545 has 42mm voice coil, 8535 resulting in lower inductance, resulting in lower nonlinear distortion. 18W Revelator also has 38mm voice coil, Usher 8945P has 42mm voice coil, so the 8945P has really a nice motor. Bigger coil means higher power handling due to better heat dissipiation from surface area.

[cotdt]

I think you are probably right, Feyz. In Zaph's distortion tests of the RS150S versus the RS150-4, above 1.5kHz the 4 ohm version had lower distortion, especially 3rd, 4th, and 5th order. but it was kind of close so it could just be driver variation.

[Dennis H]

Feyz,

The reasoning behind this is 8 ohm version has a little higher inductance

Shouldn't you look at Le/Re rather than just looking at Le? I know you implied that in the disclaimers but, given the same magnet, that's seems the easiest 'simple' spec.

[Feyz]

Shouldn't you look at Le/Re rather than just looking at Le? I know you implied that in the disclaimers but, given the same magnet, that's seems the easiest 'simple' spec.

You are right.

[noah katz]

“For some reason sandwiching the damping layer between two surfaces works much better than just affixing it to a single layer.”

When you bend a phone book, the pages slide against each other. If the same material was one piece, it would be a stiff block.

So bending forces create shear forces in planes perpendicular to the applied force.

It takes adjacent layers for the shearing action to take place; there is no shear stress on the free external surfaces.

“From the standpoint of weight added, it looks like using aluminum is basically about the same as using another 3/4" layer of MDF, and steel is a definite loser.”

Depends on what you’re after. If weight is the concern, just use more wood; if volume is the concern, use steel.

Wood gives the highest bending stiffness/weight, steel is best stiffness/volume. Aluminum is in the middle.

“My hope is that it'll diminsh energy transmitted into the enclosure by mechanical vibrations, and possibly even energy transmitted back into the front baffle by resonances.”

Since the driver is mounted on the baffle, the only way to do that is to soft-mount it.

“your MLV or steel sheet isn't going to carry any load - the glue will shear instead.”

As pointed out, that’s exactly the intention of CSD.

“I seem to recall reading somewhere that even 1 screw through the panel will greatly defeat the effect your are after. If you must use any screws, I'd put them near the center of the panel it will have the least effect on the horizontal shearing action that way.”

Good point; too many screws will mechanically “short” the panels together.

The center would indeed be the best place to put them, or ideally it. By symmetry if a panel was flexing there would be no shearing at the center and the ends of panels will have max relative motion.

Wait a sec though – this is for a panel with free ends, whereas in a cabinet the ends are fixed to perpendicular panels (a good thing, because it greatly increases stiffness).

So putting screws in both the center and ends of the panels is OK.

[Irwin AR]

Still looking for the Klang & Ton XT18 tests. Got the PDF download to work but didn't find issue it was in. It wasn't in #205 with the load of 17cm woofer tests, though I did find the Hi-Vi round planar tweeter in there. It had horrible distortion below 5kHz. It's slightly comforting to know that most of the drivers tested both by myself and K&T show similar results. I have to say that even with the text in German, K&T is still a fun magazine to read. It's 5 times better than speakerbuilder ever was, and AudioXpress mostly a garbage collection of tube amps and related crappy speakers.

John

From the following links it shows nowhere that the XT18 duo ever got tested
by both magazine.
As i myself also have almost all of the available K&T pdf copy, i've never came
across those drivers being tested either in the testing section nor the project
highlights section.
But i can't comment on hobby hifi.Perhaps someone else can

http://www.lautsprecher-shop.de/hifi/verl3.htm (Klang+Ton)
http://www.lautsprecher-shop.de/hifi/verl3hh.htm (Hobby Hifi)

[Mark K]

:huh:

Well, don't expect me to explain my findings...

Random thoughts.

The XG18 isn't a "bad" driver. It's part of the outcome of a group evaluation. Someone is on the bottom. But it's the bottom of the top group. I guess I should have thrown in some cheap dog drivers. Plus, I wouln't mind looking at another pair down the road. I bought mine second hand a while ago.

Having said that, there are better choices.

I really would like to like the xg18. It has a nice motor, basket and cone. Monte Kay and Zaph both got pretty good results, so maybe I'm the odd one out...

[Evil Twin]

There is another who didn't get very promising results, who ALSO wanted to like the driver, and use it as a midrange...

[jkrutke]

Well, there's nothing like putting a big question mark over a driver. :B It's pretty strange that our results are so different. On the other hand, the XG18 is a pretty strange driver. There has got to be something about it that really brings out the differences in our test methods. At least we can all agree that the Usher P and the SS 7" rev are great drivers, no arguments there.

I noticed that the 7" usher that was on sale at PE for $75 sold out. Bummer. Maybe we'll luck out and they will make the phase plug version the deal of the day soon.

[TacoD]

I noticed that the 7" usher that was on sale at PE for $75 sold out. Bummer. Maybe we'll luck out and they will make the phase plug version the deal of the day soon.

Next time, first buy your drivers, then publish your results .

[capslock]

My main point with that little blurb was that it can take longer than 10 minutes with a few tracks before fatigue sets in and the initial "excitement" of distortion fades. (Assuming of course that distortion isn't so bad that you cringe right away) The rapid fire listening sessions one after another just aren't appropriate for comparison purposes. If I only had 10 minutes, I'd sooner take some measurements and trust them instead.

BTW, I wasn't saying the XG18 was junk and the crowd liked them anyway. It's still one of my favorite drivers regardless of Mark and Jon's measurements. I've had roughly 10 of them come through my house for various projects and I'm well aware of what they sound like and what their limitations are. They definitely sound better in the midrange than most of the other sub $100 drivers that come through here. Basically, to me they sound like my measurements look unless of course my mind is playing tricks on me and making me hear what I see. (it's possible, I don't rule anything out) I'm still a little concerned about Jon's measurement differences since he tests on a baffle also. It could just be level differences.

Still looking for the Klang & Ton XT18 tests. Got the PDF download to work but didn't find issue it was in. It wasn't in #205 with the load of 17cm woofer tests, though I did find the Hi-Vi round planar tweeter in there. It had horrible distortion below 5kHz. It's slightly comforting to know that most of the drivers tested both by myself and K&T show similar results. I have to say that even with the text in German, K&T is still a fun magazine to read. It's 5 times better than speakerbuilder ever was, and AudioXpress mostly a garbage collection of tube amps and related crappy speakers.

John

It's in issue 6/03 which is probably too old to be availabe for download. I think I can send you a scan in a personal email without infringing copyrights. Was wrong about the level, though, 2nd is 0.5% from 500 to 3 kHz (paper cone flex?), 3rd is a tad below 0.3%.

[capslock]

Some, but not all results, can be explained by motor design.

The Peerless has a cylindrical pole piece with almost no undercut or extension. It probably sports the worst Bxl linearity of any of the drivers tested, and this might be the reason for the poor LF 2nd harmonic performance. However, this does not seem to hurt in the midrange, where excursion is no longer significant.
It sports a massive (3 - 4 mm wall thickness) lower aluminum shorting ring that sits outside of the VC on the inner diameter of the ferrite magnet. The upper shorting ring is formed by the aluminum phase plug that sits on the pole piece. Also, there is a thin (0.3 mm) copper cap that extends about 6 mm down from the top of the pole piece. Even if this shielding is too thin to be effective at very LF, it may be responsible for the good midrange and multiple tone performance of this driver. The Peerless might be the only driver in the test field that actually has a shorting ring inside the gap!


I have never taken apart a Dayton, but there were pics of failed debucking attempts over at diymobileaudio. The basic design seems to be very similar to the Peerless. The pole piece is essentially cylindrical, but the undercut is more pronounced (maybe 1 mm). I could not tell from the pictures whether there is any extension, but the motor parts looked like cold rolled steel, so it is almost certain that there was no intricate shape hiding below the phase plug.
The Dayton also uses a lower outside ring. It is made of copper and roughly 2 mm strong. The upper shorting ring is again formed by the aluminum phase plug. There is no copper in the gap.

It might be both the improved Bxl symmetry (undercut pole piece) as well as a more effective lower shorting ring that the Dayton owes it better LF 2nd harmonic performance to (one would have to know the thickness of the Peerless and Dayton shorting rings and alloy used to know for sure). The Dayton's poorer 3rd harmonic and midrange / multitone performance might be caused by the lack of copper in the gap, or by less perfect matching of lower and upper shorting paths.


The Excels are an interesting case. They use beautifully machined pole pieces with a significant undercut as well as a seizable extension. They actually have the most symmetrical Bxl curve of any midwoofer I have taken apart and simulated (actually, I disassembled an L18, but that's the same motor).
They use inner copper rings both below and above the pole piece. The copper is probably too thin to prevent flux modulation at LF, and this might be the origin of the LF 2nd harmonic distortion. I must admit though that the surround looks like it hates high excursion, so this could be contributory, too. In the midrange or with the multitone signals, the performance is respectable because of the symmetrical Bxl curve and symmetrical short circuit paths. However, the gap itself is unshielded, and will produce eddy current distortion. This might be the reason why the Excel fails to distinguish itself from the Peerless, which has a fully shielded gap, in the midrange.


I have never taken apart a Scan Speak Revelator, so I can only speculate. Scan Speak used to build full length copper sleeves, but the newer midwoofers such as the 18W8545 reportedly used a design very similar to the Excel. Supposing the Revelator uses the same motor, maybe the rings are heavier, killing flux modulation (2nd order distortion) more effectively at LF. Also, the surround might handle more excursion than the Excel.

I do not have any idea what motor design was used in the Usher, only that Usher claimed at the time that they had not copied the Scan Speak design. As I noted before, I am surprised by the absence of mirdange second order cone flex distortion in both the Scan and the Usher, which was very evident in the HobbyHifi tests.


The Vifa is also for speculation only. Rumor has it the part was originally intended to be part of the Scan Speak lineup. From the design of the back plate, it is evident that the pole piece is turned, so we can at least assume an undercut design. Also, the inductance curve indicates that some copper was used. I have heard that the surround is a distortion generator, but even so, I am surprised at the poor midrange and multitone performance.

[Mark K]

Thanks Eric,

Very interesting, to be sure. I'll point PE in your and Feyz's direction as consultants next time they ask for motor/design advice...

It would be very interesting to design, test, make specific design changes, and see how they test out.

I guess I'll have to take apart a typical motor, make some mods, and retest it...