Some random driver tests

[Reet]

Biggest problem with Omnimic is the noise floor is quite poor, with just a basic USB audio codec. OM1 mic has low self noise, and can be connected to low noise ADC. Here's some basic comparison I made some years back when I got it, and used Steinberg UR22mkii at the time, I am using Motu M4 now which is even better.

https://diy.midwestaudio.club/discussion/1825/distortion-comparison-dayton-omnimic-vs-line-audio-om1-ur22mkii#latest

I haven't bothered to do any more comparison than that. The OM1 mic has proven to be perfectly adequate. Omnimic only collects dust now, until I decide I need absolute SPL reading.

[Reet]

Coming soon, Satori MW16P. Probably nothing new to be learned from my testing, since it's already well tested by Hificompass. That will be it for my driver testing for some time.

Next up will be some waveguide experimentation with a few of the tweeters tested here. I will be milling waveguides in wood with a CNC machine, it may take a bit for me to dial in the settings, but should be able to provide some direct comparison to the stock flat faceplates, in the same cabinet and baffle location.

[theSven]

Coming soon, Satori MW16P. Probably nothing new to be learned from my testing, since it's already well tested by Hificompass. That will be it for my driver testing for some time.

Next up will be some waveguide experimentation with a few of the tweeters tested here. I will be milling waveguides in wood with a CNC machine, it may take a bit for me to dial in the settings, but should be able to provide some direct comparison to the stock flat faceplates, in the same cabinet and baffle location.

Looking forward to see some more waveguide testing. The CNC sounds like a fun time and different from the 3d printing that has been shared lately on the forum.

[Reet]

I have ideas to mill some nice shapes directly into a baffle, it can look pretty cool when done well. Image courtesy of Joseph Crowe.


I did a quick test run as a proof of concept a while back, just to show that I can do it. I made a model in FreeCAD in a way that I can very easily modify the throat, mouth, and curve profile to dial it in just right.



For those without CNC or 3D printers, DIY waveguides are not out of reach, there is always the plaster method:
https://www.diyaudio.com/community/t...eguide.327197/

[technodanvan]

Does one try to smooth that out via sanding or fill it with Bondo (or similar) to get a smooth finish?

[Reet]

5 min of sanding and it's smooth. That is just a "quick test run proof of concept" as I said, router bit with 40% step over. For final cut I would do more like 10% step over, it'll take a lot longer to machine, but require very little cleanup when complete.

[technodanvan]

Interesting stuff, worth looking into in the future.

[Steve Manning]

Does one try to smooth that out via sanding or fill it with Bondo (or similar) to get a smooth finish?

Here you go Danny in terms of needing sanding ....

This is a waveguide test I did for Jon some time ago. It's a two sided machine .... backside for mounting hardware and then flip for the actual waveguide.





This is the roughing pass to remove the bulk of the material.



This is the finish pass, no sanding at this point.



Part cut out, still no sanding.

[Reet]

SB Satori MW16P-8.


Apart from a blip at 1.5kHz that is likely from the surround, the frequency response is very good. There is no damping goo on front or back side of this driver at the cone-surround edge, which I think could alleviate this problem. Unfortunate miss on this calibre of driver.


IB estimation.


Distortion tested at 2.8V. Again, other than 1.5kHz blip, it's top tier.


Again at +5dB (~5V).

[tktran]

That higher order distortion really is outstanding!
That motor is fantastic; and I like the low weight of the driver too.

I wish SB would put an alu cone on the Satori motor. But I wonder if that would cannibalize sales from the Textreme Satori.

[duvixan]

I wish SB would put an alu cone on the Satori motor.

Marten's Parker series uses OEM ceramic 7.5" Satoris with a grill, to make them look Accuton-y

[Reet]

A couple more are coming soon. Vifa/Tymphany/Peerless NE180W-08 will be added, as well as Peerless/Tymphany HDS Exclusive 830883.

[Reet]

Vifa/Peerless/Tymphany NE180W-08. Great looking and performing midwoofer. Similar type of cone to Satori, I like the look of the yellow reeds in the NE180 better. Motor looks somewhat unimpressive as a small lightweight thimble, but performance is high, and light weight is important for some applications. Potentially, the rear wave may be fairly unobstructed for a dipole.


Frequency response is very smooth, with mild breakup.


IB Estimation.


Distortion at 2.8V.


Again at +5dB (~5V).


With a filter applied with 2.5kHz LR4 target.

[Reet]

Peerless 830883 HDS Exclusive. These are Tymphany branded, not the old DST drivers. Also a very nice driver, maybe a bit better performer than NE180W.

Frequency response.


IB estimation.


Distortion at 2.8V.


Again at +5dB (~5V).


With a filter applied for 2.5kHz LR4 target.

[Reet]

Directivity comparison. Here is power & di chart of normalized response for NE180W and 830883, very similarly sized drivers, but with different cone profiles, and of course one has a "phase plug". Dashed lines is NE180W, solid 830883.

[tktran]

Thanks Reet

My analysis is that the Satori has the best motor. The higher order distortion products are about ~6-10dB better than the rest. No wonder a certain doctor from KEF thought I was referring to the Satori when I was giving feedback about his coaxial.
The 2nd order is a bit higher than the rest, probably due to the cone. 5th element has been championing for a metal cone Satori for a long time, and I concur.

The 830883 is a little better than the NE180W. Below 250Hz it's better, and it has a slightly smoother response.
How does the silver frame look like in person?
There's still plenty of stock down under (since we're the last to get things, we seem to be the last to run out of things)

Perhaps it will go well with modern finishes like white cabinets...

[Reet]

If the Peerless HDS is goes with your modern decor, then it must have been ahead of it's time

I like it's unique character, you definitely won't mistake it for anything else. Finish is not silver, but rather light grey, almost an industrial powder coat in appearance.

[duvixan]

Both still available on aliexpress

https://aliexpress.ru/item/1005002416110478.html?sku_id=12000027342202733&spm=a2g2w.stores.seller_list.0.41e34cf3kU7ov2

https://aliexpress.ru/item/32889657185.html?sku_id=12000033665130984&spm=a2g2w.stores.seller_list.0.41e34cf3kU7ov2

[Reet]

Scan-Speak D2604/832002 "Limited Edition" tweeter. These are NLA, but as far as I can tell, the only difference from the standard D2604 is that they have a nice cast aluminum faceplate instead of plastic, and the voice coil former is titanium.


Frequency response:


IB Estimation.


Distortion tested at 2.8V, quite impressive.


Again at +5dB.

[Reet]

Melodavid MCF4Nd. Very nice looking carbon fibre midrange driver.


Build quality is a 10/10, as good as any other high end driver I've handled. It measures 115mm to the frame edge, and appears to be built on the frame of a Scan-Speak 12M. Through the gaps between the neo magnets you can see the bottom of the voice coil, and copper layer on top of the pole piece. I'd estimate xmax to be in the 3mm range.

Frequency response for this one was measured in the tweeter location, simply due to the fact that I already had a baffle insert that fit the driver perfectly (115mm diameter). For reference, here's the difference in baffle diffraction signature compared to the other midwoofs tested here. Orange dashed line is the normal location, solid blue is the tweeter location for this driver.


Frequency response exhibits a significant cone breakup, but manageable. The response dip between 2-3kHz unfortunately may limit the high frequency usability of this driver to ~2.5kHz passively, which is probably a bit lower than what most people have in mind when choosing a 12cm midrange.


IB Estimation.


I have a sound level calibrator now, this driver's sensitivity is in the 87-88dB@2.8v/1m range through the midrange, and is a 4 ohm driver.

Distortion is tested at 2.8V, 315mm. Rising 3rd order due to the cone breakup, otherwise quite excellent.


At +5dB nothing really changes significantly, just louder.


See the following thread for more information on how a passive filter implementation can work with this driver:

https://www.htguide.com/forum/forum/primetime-a-v/mission-possible-diy/driver-testing-discussion/953527-another-interesting-melodavid-speaker

[Reet]

Peerless DA32TX loaned from a friend for testing. I think this is the biggest dome tweeter I've ever seen. Massive faceplate might limit usability and visual appeal somewhat, otherwise it's a very nice looking tweeter. Some interesting and unconventional design choises. The rear chamber is an aluminum extruded heatsink, which serves no purpose as a heatsink, but includes banana plug connections. The faceplate is also a rather thin aluminum plate. I think some of the aluminum wasted on the heat sink could have been used to add another 1-2mm of thickness to the faceplate.



Under the faceplate there is a conventional dome assembly, certainly a standard side terminal design was possible.


Under the dome, a generous amount of copper is found, and no ferrofluid in this one. Regular open cell foam is in the pole peice, I didn't remove it to inspect further.


Frequency response is very nice. Big breakup at 25kHz.


IB estimation.


This driver has a somewhat low sensitivity for a dome tweeter at ~88-89dB@2.8V/1m.

Distortion at 2.8V is excellent.



Distortion at +5dB (~5v).

[JonMarsh]

I have one matched pair of these, shipped by mistake from Australian distributor sending me several sets of DA26TX. Behavior looks pretty similar, though obviously a bit lower breakup mode. I've actually built a mod set of the DA25TX using the original FP as a guide to make a thicker one- no change in the behavior, but if one put it in a cabinet with woofer back pressure I think one might feel more comfortable.

These two parts are a very interesting lesson IMO in how to create a very well behaved (overall) design for not much money...

[Reet]

It is a well performing tweeter with a fairly simple design. The rear chamber and terminal connections is quite confusing however. A lot of work apparently went into drilling through the motor to place the terminals on the back side, however the diaphragm plate is a standard design that would bring the terminals to the side like any other tweeter. It's very odd.

[Reet]

Transducer Lab N26CR2-G loaned from a friend for testing. Impressive looking tweeter, larger diameter faceplate and magnet than your average tweeter. The grill is a bit odd, it just sits loosely in between the screws, held in place magnetically. I imagine it would pop off quite easily if something hit it from the side. DA32TX grill is also held in place magnetically, but in a groove so the fitment looks a bit nicer on that one in my opinion.


Behind the dome is what appears to be a kapton voice coil former or some other non-metal material. No alignment pins, ugh! Lots of wiggle and jiggle and retest to make sure this is reassembled properly.

The motor surprisingly enough appears absent of copper. The felt on top is rather gritty feeling, and well glued in place so this is where my investigation ends. There is also no ferrofluid.

Frequency response is very good. Ultrasonic breakup is in the 28-32kHz range.


IB estimation.


This driver tested about 90dB@2.8V/1m.

Distortion tested at 2.8V/315mm. A very good performer.


Again at +5dB (~5V).

[JonMarsh]

I was saddened to see these folks leave the marketplace, but there are different skill sets between designing a very good driver and launching a successful business based on that.

[Reet]

It's an impressive tweeter for sure, but a little rough around the edges, it has a boutique / small-batch / DIY feel. Fortunately, today there are many great tweeter options that are very affordable so it's a tough market to compete in.

[Reet]

Skunkwerks AMT-1. This AMT uses a PET diaphragm, as opposed to the yellow Kapton seen on many other AMT drivers. PET deforms at a much lower temperature than Kapton, so keep that in mind if you want to use an AMT at high output. The tweeter has an odd 95mm frame diameter, and the assembly is completely glued so it's not easily disassembled. It includes an enormous rear chamber for a tweeter. Diaphragm size is about 3cm wide by 4.5cm tall, a fair bit smaller than Dayton AMT3-4.



Frequency response rises with frequency, but is usable. Resonant mode at 1.8kHz, and shelf at ~3kHz will be the problem areas. Top end extends beyond 20kHz, which is nice to see in an AMT.
Horizontal:


Vertical:


On a positive note, the driver is less susceptible to diffraction effect vs a 1" dome.
IB Estimation:


Distortion at 2.8V. Not exactly state of the art, resonant modes contaminate it's performance. It is high sensitivity, so this measurement represents about 90dB at 3kHz, 96dB at 10kHz. 3.5-4kHz would be the best crossover use for this one, but at the price there are certainly better options.

[Reet]

Seas 27TBFC. Aluminum dome appears to have a fairly shallow dome profile, and clear plastic diffuser stuck to the grill.


Frequency Response.


IB Estimation.


Distortion at 2.8V.


Again at 5V.

[Reet]

Gamut "1 magnet" version. These look very similar to Vifa/Peerless XT25, with a phase plug from the Revelator line, and have a Scan-Speak label on them.




Frequency Response. Sensitivity is about 90dB@2.8V/1m.

IB Estimation.


Distortion at 2.8V.


Distortion at 5V.


Impedance.

[Reet]

Gamut "2 magnet" version. This one has a Gamut label with "Made in China", and an attractive cast metal faceplate. It has a double staked magnet, similar to Scan-Speak R2604/8330. Unlike the R2604, it does not have the rear chamber lump. It appears to be quite over-damped, making for a low Q impedance bump at Fs, and early roll-off, this one should be "easy to use".




Frequency Response. Sensitivity is in the range of 92-93dB@2.8V/1m.

IB Estimation.


Distortion at 2.8V. Quite impressive low end performance.

Distortion at 5V.


Impedance. (really)

[Reet]

Atlas AR120. This is effectively a Bozhen CQ76 with a larger face-plate with 6 mounting holes. It's not a real ribbon, but rather a rigid diaphragm shaped like a ribbon, with a voice coil and suspension like a traditional cone.


Frequency Response Horizontal. Significant on-axis dip, however off-axis it disappears. Listening window looks okay.

Vertical.


Distortion at 2.8V, 315mm distance. Sensitivity of this large tweeter is a bit on the low side for a tweeter, in the range of 88-89dB@2.8v/1m.
Not the lowest distortion in the world, but better than most ribbon I've seen, and 2nd order is lower than any dome tweeter, no doubt a benefit of the large surface area of this tweeter.

Again at 5V.


2.8V measurement with 2kHz high pass.

[Reet]

Melodavid MR350S. Very nice looking 2" carbon fibre dome midrange. Includes a removable metal faceplate with child protector grill.

Frequency response (this one was mounted in the tweeter location in my test baffle). Breakup is large, but very high in frequency at 12kHz making this a fairly good usable midrange. This driver is 4 ohm with a good 92dB@2.8v/1m sensitivity.


IB estimation:


Distortion measured at 2.8V: (oops typo in the label should be MR350)

Same with basic digital filterBW24 HP at 700Hz:


Impedance is a good match between the pair:

[duvixan]

Are they any better than Dayton CF50 ?

[Reet]

I don't have Dayton CF50N, send me a pair and I'll find out.

On visual inspection, Melodavid looks more attractive to me. It's breakup is 12kHz where Dayton is 8kHz. Impedance otherwise looks very similar. Melodavid is also a bit cheaper, at least for me.

[duvixan]

Any interest in that diamond tweeter?

https://aliexpress.ru/item/1005009673656538.html?sku_id=12000049827341836&spm=a2g2w.stores.seller_list.5.3dcf3b5eScjgCj

[Reet]

Sure, if you're offering to send me one to test. I'm also interested in this one:

https://aliexpress.ru/item/1005009554823897.html

But, I've collected enough drivers to last me for a few years, so I need to refrain from purchasing more "out of curiosity".

[5th element]

Any interest in that diamond tweeter?
https://aliexpress.ru/item/1005009673656538.html?sku_id=12000049827341836&spm =a2g2w.stores.seller_list.5.3dcf3b5eScjgCj

If you look at the provided frequency response you'll see that the tweeter apparently has its first breakup at about 26kHz. I know they show the dome being shattered further down the page, so it's unlikely to be an aluminium dome, but this isn't particularly encouraging.

SEAS cites 45kHz for their dome, B&W 100kHz, and Accuton around 45kHz too.

To me the whole point of diamond domes is to shove the breakup as high as possible. Preferably to 100kHz so that the 5th harmonic from distortion amplification is pushed up to 20kHz.

If it's anything like their other HIEND tweeters then it'll probably have decent performance but I'm not sure it's worth anything Vs their Be dome. The waveguide is worth having though.

[Reet]

I only care about tweeter performance at frequencies heard by humans. With that in mind, the diamond dome is mostly a gimmick, only to undercut the overpriced "name brand" alternative from Seas and make for some pretentious talking points about your speaker. I also don't put much trust in any measurement data provided on Aliexpress, but if we're commenting based on what's provided. the WG arrangement for CVD104 appears not well optimized for this tweeter, with a significant dip in the response at 8kHz. I'd be a bit disappointed to have that response in any tweeter, let alone a pricey one.

[5th element]

With that in mind, the diamond dome is mostly a gimmick.

Yes exactly my thinking too. There are much cheaper options, if you want to go for a rigid dome, that have superb performance. My comments before were aimed at meaning if you are going to go to the trouble, and expense, of a diamond dome, then you should be doing everything right to achieve as many benefits as they are capable of, otherwise why bother?

Alu diaphragms usually have their issues around 25kHz. Be tends to push this out to ~45kHz and with less ringing. If you're going to go diamond there better be a good reason Vs the other two! If you aren't improving upon them then you might as well throw it in the bin!

[Reet]

If none of it results in an improved auditory experience <20kHz, it's all pointless gimmicks to drive up the price, so I will start at "why bother" and we can work our way back from there towards something tangible. The "trouble and expense" is already not worth the effort if we are working towards actual audible performance. It appears to me that speaker R&D is directed by marketing these days and not much more. What do loudspeaker engineers do all day in 2025?

Here's another new snake oil product from Seas, cost is a mere CAD$825ea:

https://solen.ca/storage/media/nqUclAo2UccrGq2yCy4rg0KkU4SwHT8d0uxs15Qt.pdf

It features a glass diaphragm marketed with only the best meaningless fluff like "very fast sound propagation speed" and "lightning fast transients".
Compare to Seas H1499 DXT which retails at $169ea....

[technodanvan]

I feel like you purposely did not tell us which graph goes to which tweeter given their relative similarity, but I'm guessing the bottom graph is the original DXT?

[Reet]

That's right. It's the only data Seas shares on this expensive part, detailing two decades in RND progress, and the deplorable state of high end audio in general. Speaker driver manufacturers need to be shown a datasheet from a $3 IC.

[tktran]

The reality of the matter is that humans like new things, and neuroscience tells us the brain likes novelty (it forms new neuronal pathways). So who could blame you, (or me) if I want a new speakers with a fancy looking shiny dome, or a woven textured cone made of carbon fiber -- they're used for Formula 1 race-cars aren't they? WOW!

Real engineering takes work, but it doesn't matter how good your product is if nobody's buying, right?
Sadly this is the way it's been for 50 years.

I discovered this when I took a history lesson in transducer design and manufacturing. Most illuminating --
you really need to need it on your tablet, bedtime reading style, or have it as a coffee table book.

Some snippets: (bold italicized is my emphasis)

"...unfortunately for Peerless, the loudspeaker industry in Japan, Taiwan and China was very competitive and making parts inhouse was not always a profitable solution. In such a situation, vertical integration can create a lack of flexibility and for example when cast magnesium (later cast aluminium) frames became more fashionable among customers, Peerless stayed with their sheet steel frames. Peerless had to reduce their activities and close down factories. They moved to Karlslunde in 1983..."

...The market demanded products with new technology and a different visual appearance and in 1987 Scan-Speak launched new versions of the 18W and the 21W woofers with membranes in semi-transparent Polypropylene that had a milky white appearance. These products were popular for many years and continued in production until the new millennium..."

i.e. Scan-Speak 7" 18W ("classic) mid-woofers--

18W/8542: nylon reinforced paper
18W/8543: polypropylene cone
18W/8544: Kevlar cone (1988) https://www.scan-speak.dk/datasheet/...e/18w-8544.pdf
18W/8545: carbon reinforced paper, with the SD-1 motor (more economical -> cheaper) designed/patented by Lars Goller in 1993 -- after Ragnar Lian's earlier SD patent had run out (1973-1993) (even though Philips already had products with it )
18W/8546:Kevlar cone is back, now with cheaper SD-1 motor https://www.scan-speak.dk/datasheet/...8w-8546-00.pdf

Inventions or developments are sometimes by pure dumb luck (e.g. penicillin, in 1928 by Alexander Fleming)
Here's how the ring radiator tweeter was developed at Scan-Speak by Peter Larsen
from https://audioxpress.com/article/spea...ft-anniversary
".. whilst working on this at Scan-Speak, a dome tweeter project was not going as intended, and repeated efforts did not yield the desired results. Peter is a resourceful fellow and decided to take extreme measures, which consisted first of consuming a couple of beers (it was the end of the week after all), after which he decided to punish this tweeter that would not behave and stuck a pin in it!
Peter discovered that the "pinned" tweeter not only sounded better, but it even measured better when clamped in its center!..."

I have stories from Bowers and Wilkins loudspeakers. When I posted it on diyAudio, it was removed by a Moderator, saying "This is not the sort of material that we want" B&W started off any other company- people repairing equipment, making new equipment using other company's drivers-- just like people at DIY Mission Possible!

Would anyone like to hear more about the history of development of frames, magnets, cones, domes, spiders and surrounds and the cabinets that make up our beloved speakers? If so, I shall post it here...

In the meantime- enjoy:

Next up after that?
the Pro Max Ultra version (secret unobtainium cone with matte vantablack coating)

[Reet]

Relentless consumerism wins again...

Perhaps you could consider writing an article for publication, instead of just forum posts, Thanh. Jan Didden at DiyAudio may be able to put you in contact with the right people at AudioXpress. See here: https://www.diyaudio.com/community/t...9/post-8129755

[JonMarsh]

But don't forget the Textreme version of drivers! NOT pretty, for the most part, but a really cool idea!

Seriously, I agree you might want to try AudioXpress, especially as Speaker Builder isn't around any more (wrote a few articles for them). But anything along these lines you'd enjoy posting here, we'll be sure to roll out the red carpet!

[5th element]

Vantablack! Now there's an idea for a cone.

[Reet]

Melodavid MW165Nd. Nice black carbon fibre cone, coated on the back side. Neodymium motor with full copper sleeve, and big 2" voice coil.


Frequency response shows an unfortunate resonant issue around 800Hz, similar behaviour to AE TD6M but not quite as bad. Otherwise the breakup is sufficiently high in frequency and the frequency response very usable. This one may necessitate an extra notch filter to tame the 800Hz resonant mode. Sensitivity is around 86dB@2.83V/1m.


IB Estimation. A bit of BSC is built-in to this one.


Distortion at 2.8V. Quite decent performance, and luckily the 800Hz resonant mode isn't represented in non-linear performance, and the breakup doesn't present any distortion problems either.


At 5V.


5V measurement with an EQ filter applied for LR24 low pass at 2500Hz, and a notch for the 800Hz bump.

[Reet]

GHXAmp RF123W - A competent copy of Tymphany/Vifa/Peerless NE123W.



Frequency response is very nice, with mild breakup starting at 6kHz. Sensitivity is measured at 88dB@2.8V/1m


IB Estimation.


Distortion measured at 2.8V.


Distortion at 5V.

[Reet]

Samtronic RT-8260 AMT.



Frequency response shows a fairly significant cavity at 2400Hz, as well as some smaller resonances throughout. Otherwise the response is quite usable, and miles better than the Skunkwerks above. Sensitivity is measured at 92dB@2.8V/1m in the 5-10kHz range.


Vertical response.


IB Estimation:


Harmonic distortion at 2.8V is quite good, with high order harmonics at -55 to -60dBr level down below 2kHz.


Harmonic distortion at 5V.


Harmonic distortion at 5V with a 2.5kHz LR24 high pass.

[Reet]

I find it interesting to compare SPDI for a few different shapes of tweeters.
Yellow is a 25mm dome
Green is Fountek NeoCD3.0
Blue is Dayton AMT3-4, fairly large, wide diaphragm
Red is Samtronic RT-8260 AMT. Size is "just right" I think.


I think this image is quite telling of the directivity problem that a dome tweeter with flat faceplace presents in a simple bookshelf speaker. The DI slope is not very constant, with wide directivity to about 7kHz, then narrowing directivity at higher frequencies. I compared a few different dome tweeters and they all follow a similar pattern. The AMTs and ribbons show a more narrow directivity than the dome of course due to larger dimensions, but the slope is very constant from 2.5kHz and up.

Here's another comparison with Melodavid Be28 with a stock flat faceplate and mounted in a shallow waveguide. This comparison is a bit more simplified as I had only measured horizontal response, while above I measured horizontal and vertical to 90 degrees, however I think still is useful to show how a waveguide helps maintain a more constant DI slope.