Testing 1, 2, 3, testing...

[JoshK]

I must admit holding my breathe as I was reading the W22 results (have a bit invested in them).

[Dennis H]

Jon, youdaman! :T (no matter what Evil Twin says about you)

[cobbpa]

This is seriously interesting stuff...I'm interested to see what the 4 ohm Dayton's look like, too. Thanks for this info Jon...testing & facts are cool.

[cotdt]

Thanks Jon! This is really good stuff. I am very impressed by the performance of the Peerless. They should make it all-black.

[AJINFLA]

Thanks Jon.

Excellent info as always :T . Distortion & FR alone won't tell how a driver sounds, but both contribute critically to the sound and should be a known parameter.
Because I favor ultra low tweeter xo's myself, I tend to focus on the larger drivers (8+") rather than the smaller (6-7") - for dipole use of course. No funds at the moment for new driver spending, but hoping to turn the corner soon.
Some drivers I'd like to see tested vs the always used "home" units:
DLS 8" aluminum/neo: http://www.dls.se/english/speakers/Nobelium82.htm
Yes, a CAR unit. Why not? Has dipole written all over it. Availability as a seperate unit/high price are the question. As is xmax.
Possibly the ultimate dipole midbass, the JBL 218F, used in this monitor: http://www.jblpro.com/LSR6300/PDF/JBL.LSR6328P.pdf
specs: http://www.jblproservice.com/pdf/Thi...Parameters.pdf 15.7mm xmax :E , low inductance and mass. If it had an aluminum/ceramic cone rather than carbon fiber, I would already own it, even at $246 ea!!
B&C 8" paper/neo: http://www.bcspeakers.com/index.php?...izione_prod=38
Prefer hard cones, but this one looks interesting. 92+db sens and less money than a nextel Excel. Gets a little rough just past 1k however. Maybe a bit too questionable to try? Always wonder how some of those "pro" units stack up.
There's plenty others, but just some food for thought. My search for the ultimate driver goes way outside of just the "home" market. The W22 was that at the time I bought it, but its days are numbered, obviously.
Jon, two 830883's should be the better value - if they fit your design goals. Nice driver! Too bad DST has all but abandoned pistonic cones/domes :M.
Interested to see how the Accuton measures when you get it.
Keep up the good work.

Cheers,

AJ

[silvercans]

Wow, I must say for the money those Daytons measure quite nicely. What a bargain.

[capslock]

Interesting. I'll look up the HH test of the W22 to see if they have found a wiggle in their far field CSD at 1.7 kHz. There is really small wiggle in the W18 test I emailed you. To my understanding, the W18 and W22 use the same cone, but it is truncated in the case of the W18.

If the Peerless exclusive still use the motor used in the HDS PP versions, I would expect them to fall apart at higher SPL. They have plenty of shorting paths (4 mm alu outside shorting ring below the top plate and massive alu phase plug above + a roughly 0.4 mm thick copper sleeve over the pole piece that extends a few mm below the gap. However, the iron geometry is really simple, just a straight cylindrical pole piece and no extension to speak of (0.5 mm). The lower outside shorting ring is strong enough to deal with flux modulation (hence the excellent 2nd HD performance), but I suspect the asymmetry of the Bxl curve will begin to show at higher excursion.


The XG18 behavior is paralled by the XT18WHs (I believe I sent you the KT tests at the time), nice distortion at 80 dB, but pretty wide band second harmonic at 90 dB. I haven't taken the motor apart yet. It seems to have a turnded pole piece, so it would have a nicely symmetrical T-pole motor. I suspect that there is too little copper shielding to be effective against flux modulation at below 300 Hz.

[JonMarsh]

If the Peerless exclusive still use the motor used in the HDS PP versions, I would expect them to fall apart at higher SPL. They have plenty of shorting paths (4 mm alu outside shorting ring below the top plate and massive alu phase plug above + a roughly 0.4 mm thick copper sleeve over the pole piece that extends a few mm below the gap. However, the iron geometry is really simple, just a straight cylindrical pole piece and no extension to speak of (0.5 mm). The lower outside shorting ring is strong enough to deal with flux modulation (hence the excellent 2nd HD performance), but I suspect the asymmetry of the Bxl curve will begin to show at higher excursion.

Well, as usual your insight is dead on to my next step, which is further testing of the 830883 at higher level. This driver is a candidate for a "smaller" speaker in a "family", and if it holds up to the upper mid 90s or beyond in that frequency range, then it will probably get the nod. Otherwise, I'm guessing I'll have to go with the W18.

The 830883 paired up may be the midrange in the upcoming "Arvo Tall" (which is based on the current form factor, but with some changes) (which isn't "tall", really, but is, sort of, but that's another silly story). The question will be whether two will hold up performance adequately down to ~ 200 Hz crossover in dipole (beyond just running the number's in SL's nice spreadsheet). Some changes in the bass configuration coming, all will be explained soon.

The XG18 behavior is paralled by the XT18WHs (I believe I sent you the KT tests at the time), nice distortion at 80 dB, but pretty wide band second harmonic at 90 dB. I haven't taken the motor apart yet. It seems to have a turnded pole piece, so it would have a nicely symmetrical T-pole motor. I suspect that there is too little copper shielding to be effective against flux modulation at below 300 Hz.

Well, those comments would go a long way to explaining the behavior we see. I didn't have too high a hopes for this once, but I thought it worth throwing into the mix. Of course, we're being pretty picky, this would be a "great driver" by DIY standards in the early to mid 90's.

[JonMarsh]

Thanks Jon.

Excellent info as always :T . Distortion & FR alone won't tell how a driver sounds, but both contribute critically to the sound and should be a known parameter.
Because I favor ultra low tweeter xo's myself, I tend to focus on the larger drivers (8+") rather than the smaller (6-7") - for dipole use of course. No funds at the moment for new driver spending, but hoping to turn the corner soon.
Some drivers I'd like to see tested vs the always used "home" units:
DLS 8" aluminum/neo: http://www.dls.se/english/speakers/Nobelium82.htm
Yes, a CAR unit. Why not? Has dipole written all over it. Availability as a seperate unit/high price are the question. As is xmax.
Possibly the ultimate dipole midbass, the JBL 218F, used in this monitor: http://www.jblpro.com/LSR6300/PDF/JBL.LSR6328P.pdf
specs: http://www.jblproservice.com/pdf/Thi...Parameters.pdf 15.7mm xmax :E , low inductance and mass. If it had an aluminum/ceramic cone rather than carbon fiber, I would already own it, even at $246 ea!!
B&C 8" paper/neo: http://www.bcspeakers.com/index.php?...izione_prod=38
Prefer hard cones, but this one looks interesting. 92+db sens and less money than a nextel Excel. Gets a little rough just past 1k however. Maybe a bit too questionable to try? Always wonder how some of those "pro" units stack up.
There's plenty others, but just some food for thought. My search for the ultimate driver goes way outside of just the "home" market. The W22 was that at the time I bought it, but its days are numbered, obviously.
Jon, two 830883's should be the better value - if they fit your design goals. Nice driver! Too bad DST has all but abandoned pistonic cones/domes :M.
Interested to see how the Accuton measures when you get it.
Keep up the good work.

Cheers,

AJ

Thanks for the kind words and useful suggestions, AJ. I'll check them out.

I may have to do a batch order on those 830883 soon. Will do some further testing next weekend, it's probably my best bet in a couple of applications, AND I promised someone I'd do a Modula MT for it. They sent driver files, but I probably feel more comfortable double checking.

One think I'm curious about is how the W22 would fare with a waveguide tweeter setup, crossed around 1100 to 1200. But for now, building the waveguide is an obstacle. I did get "Speak", so maybe designing it won't be once I get that installed on the Praxis system.

~Jon

[capslock]

How about using the W22 as a waveguide? I have some rounded N52 magnet segments on order, which together with a TBC replacement dome would make a killer tweeter.

Forgive my asking (I don't want to sift through too many pages again), what is your current Avro baffle configuration? Flat? Width? No rear damping?

[ThomasW]

Current recommended baffle is flat, at least 15.5" wide, no damping.

[capslock]

Current recommended baffle is flat, at least 15.5" wide, no damping.

40 cm? That makes for a low SAF!!!

[AJINFLA]

One think I'm curious about is how the W22 would fare with a waveguide tweeter setup, crossed around 1100 to 1200. But for now, building the waveguide is an obstacle

Yeah, I'm really pushing the XT19 past where I'd like to be distortion wise. The solution there might be the H1212 and that Hifitalo WG made just for it.
No need to build . Plug' n play. Question is will they ship here? I bet there might be enough interest on this board where some sort of group buy would make it worthwhile. The H1212 is (rightfully) a popular unit. Definately one of the top choices for 1-1.3k xo-ing.

How about using the W22 as a waveguide? I have some rounded N52 magnet segments on order, which together with a TBC replacement dome would make a killer tweeter.

The cone definately has the right profile, but you would definately have to use a thin felt/foam ring at the cone edge/surround. Maybe even a small one right at the throat along with some trimming of the vc former. If anyone here can pull that off it might be you Eric. Or Mr. Idland :W .
Cross above 300hz and cone(WG) movement shouldn't be an issue.
That dia. cone should provide some nice loading - and you'll need it since 1.4k is about as high as you would want to go. Lower is better so the tweeter best be fairly robust. Using the pole vent (room) for a rear chamber might be advisable.

Jon, I sure hope that Accuton tests well. I checked out the price on Mad 8O .
Oh mylanta. That's almost 2 JBL's :E . I guess we'll see.

Cheers,

AJ

[cotdt]

How about using the W22 as a waveguide?

That will be one expensive waveguide! Aren't there cheaper drivers with the right profile?

[TacoD]

Great measurements! For midrange duties I prefer the W18 above the W22, but hey then you need more :-).

I am trying to get hold on a pair of H1212, but it looks like every reseller in the Netherlands is waiting for supplies from Norwegian....

Attached my try of the Peerless 830883, measurements taken at 1m. Unfortunate I have a small room so gating was quite aggresively (also I have the stock Praxis mic). Below 200Hz is nearfield. Scale is 5 db/div.
I almost forgot, measured in a small enclosure. Dip @ 1,3 kHz is a baffle diffraction effect, it dissapears when measuring on another angle.

[Jonasz]

The H1212 is (rightfully) a popular unit. Definately one of the top choices for 1-1.3k xo-ing.

So, would the H1212 be a better choice than the RS28 at 1.4 khz? I use the RS28 now and I love it, I just lack some level above 10khz. Will the H1212 be as good as the RS28 down low and at the same time have some more output up high?

[cotdt]

So, would the H1212 be a better choice than the RS28 at 1.4 khz? I use the RS28 now and I love it, I just lack some level above 10khz. Will the H1212 be as good as the RS28 down low and at the same time have some more output up high?

The H1212 has resonance peak a little past 20kHz, meaning they show up in the harmonics much earlier. This edges me toward using the RS28A.

[AJINFLA]

So, would the H1212 be a better choice than the RS28 at 1.4 khz?

"Better" would be a bit too generalized. The answer is maybe. Or lets put it this way. If "I" were to build "your" speakers, I would use the H1212. I really don't think you could go wrong either way. "My" answer would be yes.
FYI, coincidentally, Mark K is considering doing another version of his RS 8" 2 way. Using a H1212. Crossed @ 1.4k.
Try lighting a fire under his butt to get it done. I'm sure he'll come up with the usual excuses about having seven kids. But you know he's yearning to. You can't be in DIY and leave well enough alone. Ever. The unrelenting urge to change something should be ever present.... :spin:

cheers,

AJ

[JonMarsh]

Yes, the H1212 vs RS28a-4 vs 27TDFC vs Millenium Excel is an interesting topic.

Got them all on hand, as well as some others. I'm really pondering options on what way to go- including "finalizing" the M8ta design. It's sort of a legacy design, or an homage to my roots, but they're also supposed to be my bedroom speakers!!

The last tests I saw from MarkK show the 27TBFC is pretty robust; I'm curious to see what it looks like pushed, but also at more "normal" levels. I.E., 90, 96 dB, and 100 dB at one meter.

I'm also very curious about the H1283, and how it might react to a waveguide.

You know, what Mark needs is his 8" two way with a waveguide loaded H1212 crossed at 1 kHz. The flapper mode of the R225 is at 1500 Hz.

I think what I'm going to do is see how well the whole thing works out with the MCM guide and a ~1800 Hz or 2 kHz crossover. If it all makes sense, then maybe with the Gedlee software we can design for down to 800 Hz or so, and then just have to figure out how to build it! The SPEAK software is having problems on my laptop, maybe on my Praxis system with older XP and less updates it will play nice....


~Jon

[AJINFLA]

I'm really pondering options on what way to go- including "finalizing" the M8ta design. It's sort of a legacy design, or an homage to my roots, but they're also supposed to be my bedroom speakers!!

LOL. Like I said. DIY is about that irresistible urge to fiddle with stuff :lol:
My computer speakers are coaxial dipoles for crying out loud. I think my alarm clock might be the only thing I haven't modified. Ever heard a doorbell with a subwoofer :W ? Change is good :T

I'm also very curious about the H1283, and how it might react to a waveguide.

and I'm curious how it might react to a waveguide and a rear chamber.

You know, what Mark needs is his 8" two way with a waveguide loaded H1212 crossed at 1 kHz.

Yep. Screams for it IMHO .
I made my pitch to Ken Kantor about manufactuer optional WG faceplates, but I'm guessing its just not economically feasible. Maybe Krutke could make a pitch to MCM for something like the Hifitalo. They are selling it for cheap. Why not someone on this side of the pond? It doesn't have to be 100% perfect - to be better than none at all - as I've said on many a occasion.

Cheers,

AJ

[JoshK]

I spoke with Bob @ SPTech and he said he could do something for us DIY'ers, but alas I don't think it'll be cheap enough to warrant enough interest. Ed LaF said he'll spin us some if we decide on a form and help him/us test it. I think AB (oneolddude) is currently testing some of the prototype profiles.

Jon, why do you need Geddes software? What does it do precisely? I have been thinking about expanding the spreadsheet to include some parameters of interest, which is easy enough to calculate. I just need to locate the equations of interest in his papers (have them all printed out). Part of my masters was in numerical solutions to PDE's, so this shouldn't be a reach.

Maybe when I get back from sunning in the carribean 8) , I can write up some VBA.

[JonMarsh]

Geddes software is both for design and simulation of a design; hopefully it will make for something like a one pass process, as opposed to repetitive cut and try.

[oneoldude]

Here is some horn design software that at least will run. And it is free.

The on axis SPL function is described by the writer in private communications as "Power Response" and it models the driver as a flat piston with dropping HF response like that of a compression driver. It will allow depth and diameter changes to be made and an examination of lift at lower freqs and and examination of polar response is possible.

Here is the URL http://www.users.bigpond.com/dmcbean/

[Mark K]

Hi Jon,

It's good to see some more data on these drivers. They pretty much parallel mine and Zaph's.

If anyone wants to send me some horns, I'll be happy to take a look!

[JonMarsh]

Hi Jon,

It's good to see some more data on these drivers. They pretty much parallel mine and Zaph's.

Well, I guess that means we're getting the same drivers more or less! not something that's always asured, is it?

I'm planning on doing some of the torture testing like you've been doing, too, but I also wanted to look at them from another viewpoint- something I didn't expect when I did some sub driver testing last year was how some of them can be fairly mediocre performers even at relatively nominal levels. I guess the ideal thing would be to be able to put together a linearity map- BL plots should give us strong clues to that, along with motor inductance.

When the best drivers are 15 dB or better in distortion even at nominal levels like 90 dB output, I have to think that counts for something as regards sound quality. We aren't always kicking out the jambs, so to speak.

I'm lucky that you, Zaph, and Monte have tested so much stuff already, because it gives me some opportunity for correlation, as well as weeding out the chaff a little earlier. Also, to some extent I want to see how much some of the simpler, faster tests can point to the areas deserving of more detailed evaluation. As well as narrow down the playing field in a meaningful manner.

Next weekend will probably be a quick look at some tweeters, maybe even with waveguides, though I may not have a chance to set one up by then. I'm hoping I can, though, at least to start with the H1283. Then, if that is promising and I'm real brave, the C13-6, depending on how it looks without the waveguide.

~Jon

[capslock]

Had another look at the W22 test in HH. There is a tiny bit of grunge in the far field (1 m) CSD between 1.5 and 2 kHz, but really benign, like -25 go 30 dB for about half a ms. I originally thought it was at the hump in the FR, which occurs closer to 1.2 kHz, but closer inspection showed that was in fact slightly higher in frequency than the hump. I was once told this hump is the resonance of the air volume trapped in front of the cone. It is true that you only see this in really hard-coned drivers, and that its frequency decreases as the cone size grows.

[cotdt]

I figured out how to make every RS28A work consistent and sound really great:

Image not available

Image not available

So I opened mine and guess what I found? Off-center felt button, and really poorly applied hot glue. There was also some junk on the button but I doubt that affects performance at all. The 1.0 mm off center is probably the biggest concern. And look at how badly the hot glue is applied, though it probably doesn't matter much. I'll bet I can do a much better job myself. The magnet is also 1.0 mm off center but that can also be corrected.

Image not available

So I redid the hot glue. Maybe there is better glue but I wanted to stick to the original design, no need for an "RS28B". I put it back to my Modulas and wow! Now it sounds even smoother and cleaner than my other RS28A! Sorry I don't have pics or measurements of the cleaned up RS28A, but I can say that it played the low end much cleaner than before, even better than the RS28A in my left channel speaker. What an easy fix.

I would also prefer the RS28A over the Seas H1212 because the resonance peak at 22kHz shows up at 8.8kHz in the harmonics. RS28A doesn't have this problem.

[dlr]

Hales voice coil dimensions

Well, perhaps I shouldn't tell you how to change the sound too much, (if you like it a lot already), but my suggestion would be to remove the acoustic diffuser/lens, and stuff the center pole piece with lambs wool. This should smooth out the extreme top a fair amount, and refine the rest of the range a bit, too. This is an underhung design; the logical extreme of develpment for this is the Hales tweeter I have a few pairs of; no phase diffuser; larger magnet system and gap, slightly lower Fs, and a Faraday cap for the pole piece. (this tweeter was made by Vifa as a custom)

I've been looking at the replacement dome for the Hales tweeter at Madisound, considering it for a hybrid experiment. Can you tell me if the former/voice coil dimensions are the same as for the DX25 and D27 series tweeters? I think that it will drop into other motors, but have hesitated as the replacement is around $35.

Also, have you had the opportunity to measure those? It looks like you need to build a baffle with a rear-mount setup.

Thanks.

[Mark K]

I figured out how to make every RS28A work consistent and sound really great:

So I opened mine and guess what I found? Off-center felt button, and really poorly applied hot glue. There was also some junk on the button but I doubt that affects performance at all. The 1.0 mm off center is probably the biggest concern. And look at how badly the hot glue is applied, though it probably doesn't matter much. I'll bet I can do a much better job myself. The magnet is also 1.0 mm off center but that can also be corrected.

So I redid the hot glue. Maybe there is better glue but I wanted to stick to the original design, no need for an "RS28B". I put it back to my Modulas and wow! Now it sounds even smoother and cleaner than my other RS28A! Sorry I don't have pics or measurements of the cleaned up RS28A, but I can say that it played the low end much cleaner than before, even better than the RS28A in my left channel speaker. What an easy fix.

I would also prefer the RS28A over the Seas H1212 because the resonance peak at 22kHz shows up at 8.8kHz in the harmonics. RS28A doesn't have this problem.

Yes, this is part of the problem. That felt is glued in rather hapazardly in some of the units. It is possible to improve things by recentering that. Still, it's hard to do right without measuring distortion before and after.

And a point of correction, there is no 8.8k distortion peak, so to speak. That distortion is just the ~23k product, amplified by the dome resonance. The driving force may be the 8.8k fundamental, but the distortion occurs at 23k, not 8.8k. Although visually unappealing, it's not at all clear that tones at 23k are signifiantly audible, in the grand scheme of things.

[capslock]

Well, there are quite a few people who maintain that distortion products will actually downmix from the ultrasonic resonance and become audible.

For this to happen physically in the driver, the mechanical resonance of the dome would somehow have to influence the movement of the VC as a whole. However, these resonances seem to be localized for I cannot recall ever having seen an ultrasonic impedance blip (however, most of the impedance curves stop at 20 kHz).

Then there are those that say that our hearing is able to generate difference tones. This is most certainly true, but to the best of my knowledge only for frequencies that actually manage to excite nerve cells.

[JonMarsh]

I've been looking at the replacement dome for the Hales tweeter at Madisound, considering it for a hybrid experiment. Can you tell me if the former/voice coil dimensions are the same as for the DX25 and D27 series tweeters? I think that it will drop into other motors, but have hesitated as the replacement is around $35.

Also, have you had the opportunity to measure those? It looks like you need to build a baffle with a rear-mount setup.

Thanks.

I plan to measure it and some other tweeters this coming weekend. I'll be using a heavy felt pad to manage the frame discontinuity. They have the dome assemblies at Madisound? Whooo hooo, I need one! (one tweeter that got munched).

Well, here's a Hales Transcendence dis-assembled,







and here's the part from Madisound....





"Houston, it looks like we have a match on the bogey.... "


Are the dimensions for the standard Vifa parts online, in their data sheet? If so, I can check against one of the tweeters at home this evening.

Thanks for this tip, Dave. I'll probably order a few of these, depending on the test results. I'm "hoping" for something close to an RS28a. The frequency response is quite extended, just don't know about distortion.

[Mark K]

Hi Eric,

Well, there is evidence that our ears can perceive difference tones. There is a famous (well, sort of famous) psychoacoustic experiment, named after someone whose name I can't recall. Basically, you play two tones, just above the threshold of hearing through two entirely separate amp/signal generator/loudspeakers, say at 21k and 22k. When you do this, most folks hear a 1k difference tone because of their ear's nonlinearity.

My point is not so much whether you can devise scenarios where this might be audible, but how important is it? Most people hear some brightness or edginess that they don't like, then they see a visually unattractive peak in the FR, and they just connect the dots with a straight line.

[JonMarsh]

Whether it's justified or not by empirical evidence. :B

I've done that very experiment you describe back in the 80's, Mark. Works like a charm. Need a good tweeter to do it; most things in the 80's were "fizzy" soft domes that would generate a lot of stuff themselves.

OTOH, I'm looking at tweeters with big ultrasonic peaks with a skeptical eye these days, and trying to find some good candidates with minimal Q in that peak, that will be usable otherwise. This is what is leading me to try to devise system configurations workable with the H1283 and with the Accuton C13-6. We'll see if their ~35 kHz resonances are as well damped as advertised, too.

Some days I don't think of myself so much as a designer, but as a man wandering in the woods, reading the signposts, trying to find the way to system nirvanna- at this level of DIY, there's so little we can actually design, but the choices of what we use are so critical. I'm hoping we're posed to take things to the next level on several fronts.

[AJINFLA]

Most people hear some brightness or edginess that they don't like, then they see a visually unattractive peak in the FR, and they just connect the dots with a straight line.

I disagree. I think they see the unattractive peak in the raw FR, THEN hear brightness or edginess. This has been raised ad infinitum on forums before. Remember the guy who didn't like the Orions because he could "see" the FR peak of the W22 and smell the hysteresis of the XL12,etc,etc :gossip: -forcing John Stone (SEAS NA) to respond. I'm with Dennis Murphy on where my money would be if there was a wager to find out exactly who could hear what when blind testing of metal drivers were involved.
Of course, when the golden earred fail to "hear" when they can't "see", the testing quickly becomes flawed... :roll: oke:

Oh, BTW, the H1212 is as smooth as silk. Proving once and for all that I'm probably deaf, but *maybe* not dumb. :B

Cheers,

AJ

[AJINFLA]

BTW, Jon have you heard any Revels?

Cheers,

AJ

[JonMarsh]

Yes. Another reason I like inverted dome drivers. Tweeters, mids, woofers, you name it. I think it's probably the best possible cone design for dynamic voice coil drivers. I'd love to build speakers with Revel drivers.

Still wish I could get a driver that mated the M8a cone and the W22 motor. Or better yet, an inverted dome cone using the Seas metallurgy. Sighhh...

[cjd]

If anyone does a test on the metal domes, I sure would be interested in giving it a shot.

I know that I can not always hear ultrasonic issues, but I also know that I can sometimes (for example, I can usually hear the ultrasonic cleaner in a jewelry shop.)

The H1212 sounds great but wears on my ears over time - in fact, very similarly to the stupid stock earbuds that come with the iPod. The RS28A does not - its peak is simply higher up (I believe). I have heard the H1212 in good designs, well implemented, so I know it's not simply an issue of poor work. I'm not entirely sure why this is the case, because that peak should still not be excited continuously, regardless. And ribbons, which can hit 40kHz don't usually give me issue.

As much as anything, I would be curious to know what it is my ears don't like.

C

[JonMarsh]

They're all on the list and here at the "workshop"....

Seas H1212
Dayton RS28a-4 (unshielded)
Seas H1283
Focal TC120dx2
Hales/Vifa Transcendence
Accuton C13-6
MB Quart as used by Avalon
Tweaked LPG26T
North Creek D28 (another Usher product, like the RS28a)
SS 60000
Millenium Excel (tested previously at MarkK's)

There's probably something else I'm forgetting. Have some Founteks, but wasn't going to setup for those. Not yet, anyway.

Oh yeah, Seas H1189.

[capslock]

Hi Eric,

Well, there is evidence that our ears can perceive difference tones. There is a famous (well, sort of famous) psychoacoustic experiment, named after someone whose name I can't recall. Basically, you play two tones, just above the threshold of hearing through two entirely separate amp/signal generator/loudspeakers, say at 21k and 22k. When you do this, most folks hear a 1k difference tone because of their ear's nonlinearity.

My point is not so much whether you can devise scenarios where this might be audible, but how important is it? Most people hear some brightness or edginess that they don't like, then they see a visually unattractive peak in the FR, and they just connect the dots with a straight line.

Not sure what that experiment tells us. Could be motor nonlinearity. Was it verified that the tweeter wasn't generating the 1 kHz by itself? I have a hard time believing that the eardrum, malleus & incus and the cochlea have a nonlinear transfer function when it comes to transmitting sound. And finally, there are no receptors for 21 kHz and 22 kHz.

The experiment will work nicely with 14 and 15 kHz, but then that is the nonlinearity of the receptors and the nerves.


An experiment any of us could probably do is to drive a tweeter at 1 kHz below its dome resonance and look for radiated 1 kHz with a microphone...

[capslock]

I know that I can not always hear ultrasonic issues, but I also know that I can sometimes (for example, I can usually hear the ultrasonic cleaner in a jewelry shop.)

Might be a subharmonic you are hearing. I can hear TV sets (15.6 kHz in Europe), and I can sometimes hear the stereo carrier from a radio transmission 1 kHz), but I know I should not be able to hear a VGA CRT. Sometimes I do, though, on an old unit we have at the office. I figured it must be 1/3 or 1/4 the scan frequency. If I give it a whack, the sound will disappear.

[Mark K]

Not sure what that experiment tells us. Could be motor nonlinearity. Was it verified that the tweeter wasn't generating the 1 kHz by itself? I have a hard time believing that the eardrum, malleus & incus and the cochlea have a nonlinear transfer function when it comes to transmitting sound. And finally, there are no receptors for 21 kHz and 22 kHz.

The experiment will work nicely with 14 and 15 kHz, but then that is the nonlinearity of the receptors and the nerves.


An experiment any of us could probably do is to drive a tweeter at 1 kHz below its dome resonance and look for radiated 1 kHz with a microphone...

I'm not sure you're getting the key point of this experminet. Perhaps I didn't explain it well. There are two separate driver/amp/signal generator combinations, each with a single tone. No IM distortion can be formed here, and it is not some sort of subharmonic. You can do and measure this. Or, rather, you can't. If you set up a measurement mic, it will not measure a 1k difference tone, because the 1k tone is IM inside your ear. Who said the ear was linear anyway?

I'm not saying this stuff 20-40k isn't audible. I'm just saying there is only scanty evidence that it is. So one should look for more plausible explanations of a particular phenomena.

[joecarrow]

Isn't the 1khz tone you hear a "beat frequency"?

http://en.wikipedia.org/wiki/Beat_frequency

[kappus]

I'm not sure you're getting the key point of this experminet. Perhaps I didn't explain it well. There are two separate driver/amp/signal generator combinations, each with a single tone. No IM distortion can be formed here, and it is not some sort of subharmonic. You can do and measure this. Or, rather, you can't. If you set up a measurement mic, it will not measure a 1k difference tone, because the 1k tone is IM inside your ear. Who said the ear was linear anyway?

Hey all. First post--I have little speaker building experience, so I rarely have anything to contribute. But, non-linear acoustics I do have experience with!

Mark --

The ear does have non-linearaties, but so does the air. If driven at moderate levels the sum and difference frequencies are measurable. There's an entire company built around this effect called American Technology Corporation http://www.atcsd.com/
interesting white paper -- http://www.atcsd.com/pdf/HSSWHTPAPERRevE.pdf

I had a chance to hear one of their transducers when the inventor gave a guest lecture at the university. It was extremely eerie as the sound seemed to be originating from a cylinder extending from the device. Upon striking a wall it would scatter and seem like a point source.

This effect is also used to make ELF sonar pulses in the ocean where extremely large transducers are impractical. Instead they tow two reasonably sized transducers and use the same non-linear down-mixing to create whatever 10Hz or lower pulses they wish.

As to whether ultrasonic breakup is audible, I'm still unsure. First the breakup node(s) would have to be activated by a harmonics, then it has to couple back down into the audible (inherently inefficient). I vote someone try to measure it--take 2 27TBFCG's with a big breakup around 26kHz, drive them at 13kHz and 13.5kHz hard and see if you get something at 1kHz. Might work.

-Brian

[capslock]

I'm not sure you're getting the key point of this experminet. Perhaps I didn't explain it well. There are two separate driver/amp/signal generator combinations, each with a single tone. No IM distortion can be formed here, and it is not some sort of subharmonic. You can do and measure this. Or, rather, you can't. If you set up a measurement mic, it will not measure a 1k difference tone, because the 1k tone is IM inside your ear. Who said the ear was linear anyway?

I'm not saying this stuff 20-40k isn't audible. I'm just saying there is only scanty evidence that it is. So one should look for more plausible explanations of a particular phenomena.

OK, that rules out amp and motor nonlinearity. I am still not convinced that the nonlinearity of the air and the ear's sound transmission path are big enough at reasonable levels for this downmixing to take place. I have no doubt the experiment will work fine below 15 kHz, but then the nonlinearity of our receptors overrides any other nonlinearity.

But it really is an easy experiment to set up, and maybe one of these days I will give it a try (Versuch macht klug!). Just have to figure out how to get hold of a second ultrasonic signal generator.

By the way, there are non-linear limiters in the ear (those muscles attached to the little bones that transmit the sound from the eardrum to the cochlea), but I guess their attack and release times would be >> 50 µs.

[cotdt]

Jon, so is tweeter power handling an issue for the project?