Some random driver tests

[Reet]

Before Christmas I had been building a test rig. Over the holidays I was sick and without energy to do much and didn't want to leave the house, what better time to run through a bunch of driver tests. I ended up testing everything I had sitting on my shelves, as well as pulled a few drivers that I'm very familiar with as a point of reference.

The Janky Test Rig
I made a platform to easily rotate a speaker at various angles. It's nothing fancy, just a front bearing for a pivot point, and a couple castors on the back to spin around on. It doesn't look like much but works surprisingly well.

The test cabinet is a fairly normal bookshelf size, with removable baffle blanks so I can swap out drivers and test under the same conditions. It's not an IEC baffle by any means, but provides a good point of reference for a somewhat typical use case. The cabinet is sized to accommodate most tweeters, up to 148mm waveguide, and woofers up to 182mm. A solid plant stand elevated the speaker to about the right height for in-room measurements.



Cabinet diffraction response, on-axis for typical 1" tweeter:


Diffraction response in the woofer location, 152mm driver shown.


Equipment
Motu M4
Amp is a LM3886 DIY thing, loopback tested <0.01% THD up to 20kHz.
Mic: Line Audio Omni1

Disclaimer
My setup is NOT SPL calibrated. If SPL is shown in any of the graphs, it is not an accurate figure. Distortion testing was completed at common voltages, distortion measurements are not SPL matched, so take the driver sensitivity into account for comparison. The purpose of this testing was not to replicate any laboratory sort of test results, but to gather enough data to decide the best use case for the drivers I have on the shelf, by observing frequency response and distortion profiles.

The Line Audio Omni1 also is not a "real" measurement mic, however I have found it's response to be within 1dB of my Omnimic, it doesn't have a long stick so there is some early reflection influence in the top end creating some extra wiggles there. Keep that in mind, however there is still plenty of information to determine if the tweeter is problematic or not.

Test Conditions
Frequency response measurements were completed at 1m distance, 15 degree intervals from 0-90 degrees.
Distortion measurements were completed at 315mm distance to reduce room influence, voltage and other conditions as noted.

Driver testing to follow in subsequent posts.

Driver index
Scan-Speak D2604/8330
Vifa D27TG-45
Morel MDT29
Wavecor TW030WA08
Wavecor TW030WA11
Wavecor TW030WA13
Vifa TG9
Melodavid CB25
Peerless 810921
Fountek NeoCD3.0
Dayton PT2C
Peerless 831882
Wavecor WF152BD04
Wavecor WF120BD04
Rival R152PK08
Peerless 2" Dome (model unknown)
AE TD6M
Melodavid BE28
Melodavid D28S
Dayton AMT3-4
Peerless/Vifa DX20FB00
B&W FST LF27715
Melodavid CM25Nd (part 1)
Melodavid CM25Nd (part 2)
SB Satori MW16P-8
Peerless/Vifa/Tymphany NE180W-08
Peerless 830883

[Reet]

First was a tweeter to determine a baseline, something common and fairly popular. Scan-Speak R2604/8330 ring radiator.


Nothing really to complain about in the FR. It is smooth, hindered by cabinet diffraction only.



For an estimation of IB response, the on-axis response is divided by the diffraction simulation.


Distortion tested at 2.8V, pretty good.
Impedance, nice low Fs.

[Reet]

Next is an old one. Vifa D27TG-45. This version includes the rear chamber, I think it is the same as Scan D2606/9220 that is still for sale today.

Disclaimer on this one. I purchased cheap used units that ended up to have blown diaphragms. I replaced the dome assembly and ferrofluid myself, so they may not perfectly represent an off-the-shelf unit. Dome assembly was the Scan-speak replacement part 600058.


Frequency response is okay other than a sharp dip at 9kHz. Both units tested the same.

IB estimation.


Distortion tested at 2.8V. The other unit had higher 3rd harmonic <2kHz, unfortunately it doesn't look like I saved that result. Pretty good., 2nd order shows high at 9kHz because the fundamental is low at that frequency.


Here I applied a filter with 2.5kHz LR4 target response. Perfectly well usable tweeter.


Impedance. Fs is fairly high.

[Reet]

Morel MDT29.


Frequency response is very good, perhaps the best yet.


IB estimation.


Distortion tested at 2.8V. Not as bad as I expected, actually quite good.


Impedance. Fs fairly high due to lack of rear chamber.

[Reet]

Now for some Wavecor models. First is TW030WA08. 30mm ferrofluid model, this one has a simple motor, I believe is without copper.


Frequency response is quite good. Digs quite low in frequency. Some weird stuff going on >10kHz, otherwise very good.


IB estimation.


Distortion tested at 2.8V. At first glance I was a little disappointed in the 3rd harmonic. Consider however that 3rd harmonic >5kHz will be unheard by most, the 3rd hamonic through the more critical listening range is very good. To top it off, the large dome allows this one to perform very well at the low end of the spectrum.


Tested again with a filter applied with a target of 1.5kHz LR4.


Impedance.
Oops, didn't save that one, I should come back and edit this post later.

[Reet]

Wavecore TW030WA11. I'm really exited to put this one to use, it looks excellent.



Frequency response is very good, you can see that the baffle diffraction has less impact on this one from the small waveguide provided. A bit more directive sound than your standard flat face tweeter.


IB estimation will be skipped for this one, as a diffraction simulation will not accurately represent the waveguide impact.

Distortion at 2.8V. Nothing to complain about here.


With a filter in place, target 1.5kHz LR4.


Impedance.
I should come back and update this later on.

[Reet]

Wavecor TW030WA13. Small format 30mm tweeter performs very well, TW030WA11 above is slightly better however.



Frequency response.


IB estimation.


Distortion tested at 2.8V


Filtered for 1.5kHz LR4 target.


Filtered for 2kHz LR4 target.

[Reet]

A value leader. Vifa TG9 3.5" full-range driver. Low sensitivity and limited power handling is it's downside, otherwise it performs very well. This one was tested in the tweeter location in the cabinet.



Frequency response, merged with nearfield ~400Hz. Of the best I've seen for full-range response.


Distortion at 2.8V, fairly low SPL.


Repeated at 6.2V to be in the same ballpark SPL as the tweeters tested. This driver cannot manage output below 200Hz at this level. for mid-tweet use, pretty good, 2nd order a bit high <1kHz.

[Reet]

Here's a fun one. Cheap AliExpress brand Melodavid CB25. Copper beryllium alloy dome, includes a ferrite motor with copper cap. Visually it looks quite nice, cast aluminum face place, I wish most tweeters would stop making the face plates from plastic.



Frequency response is usable, but not the best. Dome resonance is <20kHz, but the bigger problem is a dip in the response 8-9kHz. It fills in off-axis, so I don't think it will sound as bad as it may seem on the surface.



IB estimation.


Distortion tested at 2.8V. For such a cheap driver, this is staggeringly good. Dome resonance doesn't cause any problems lower in frequency.


Tested with a filter target of 2kHz LR4.


Tested again with a target filter of 2.5kHz LR4.


Impedance. Fs is around 630Hz.

[Reet]

Peerless 810921 HDS tweeter, now being sold as Scan-Speak D2608/9130. Here's what I have called my favourite dome for years. I've used this one for years and am a big fan. Not only does it perform well, but it looks very nice as well with a cast aluminum faceplate.


Frequency response is very nice, a bit over-damped so it rolls off a bit earlier on the low end than many others.


For a point of reference of vertical response, I tested this one on the vertical plane as well. This is positive vertical angles (listener above speaker).


IB estimation.


Distortion at 2.8V. Very good!


I tested this one again at +5dB, about 5V. Still doing pretty well on the low end.

[Reet]

Fountek NeoCD3.0 ribbon.



For frequency response, this was tested with a 50uF cap in series for low frequency protection. I use a dual channel measurement, so the cap doesn't have any impact on frequency response, however SNR is affected on the low end, so please disregard the weird stuff <500Hz. One angle was perticularly bad, but the data is good in the frequencies of interest.

There's a big lump in the response 1-2kHz, but lower than the intended frequency range use for this tweeter so not a big problem.


This one was tested vertically as well. For comparison to a 1" dome, see Peerless 810921 above.


IB estimation.


For some reason I had forgot to his the save button on my REW file for distortion. I did take some screenshots, but at a different resolution than shown for the rest of the drivers here. sorry about that, I may redo at a later date.

Tested at 2.8V, limited to 1.8kHz min. Quite high for all harmonics.


Tested again with a filter for 3kHz LR4 target.


Tested again with a 4kHz LR4 target. This is probably the best way to go if you plan to use this tweeter.

[Reet]

The last tweeter for now. Dayton PT2C planar. Wow, this thing is big! I think it's kind of ugly too, but beauty is in the eye of the beholder.


Faceplate is attached with 4 screws, so there is a potential to give this one a facelift and directly rear-mount it to a baffle.


This one was a bit too large for my tweeter plank plates, so it got surface mounted to a board. Keep that in mind, that normal installation should result in a smoother response than what I show here.


Vertical. Since this driver is so very directive vertically , I took a couple extra measurements at 5 and 10 degrees. So measurements shown here is 0,5,10,15,30,45,60,75,90.


Distortion tested at 2.8V. I didn't have high expectations for this one and had expected worse honestly, this is quite usable.


Cranked up +5dB to ~5V it starts to fall apart on the low end.


Back at 2.8V with 3kHz LR4 target filter in place.


Again, with 3.5kHz LR4 filter in place.

[Reet]

On to some midwoofers. Here is an old NLA Peerless HDS Exclusive 831882 5.5". Excellent driver all around, and a great pair for the HDS 810921 tweeter.


Frequency response is good, very usable driver.


IB estimation.


Tested at 2.8V, not much to complain about here.


Tested again at +5dB, about 5V.


Tested with filter applied for 3kHz LR4 larget for typical use. With BSC, this is about 5V at low frequencies, 2.8V at 1kHz.

[Reet]

Wavecor WF152BD04.


Frequency response has a bit of a step at 1700-1800Hz, still quite usable.


IB estimation.


Distortion at 2.8V. Sorry, some background noise during this one, result is still usable. Very good performer, I think possible something buzzing at 220Hz, not the driver.


At +5dB, about 5V.


Tested with filter applied for 2.5kHz LR4 larget for typical use. With BSC, this is about 5V at low frequencies, 2.8V at 1kHz.

[Reet]

Wavecor WF120BD04. Smaller 4" version of above. I think this one is just excellent for midrange use.

Frequency response. I realized once I started testing that I hadn't chamfered the back side of the test baffle. For a driver this small, I probably should have. May be the cause of the bump ~1.5kHz.


IB estimation.


Distortion at 2.8V.


Again at +5dB, about 5V.


Tested with filter applied for 2.5kHz LR4 target for typical use. With BSC, this is about 5V at low frequencies, 2.8V at 1kHz.


For a more stressful test, I also tested at 7.8V with a 3.5kHz LR4 target filter. Again, probably something buzzing around 300Hz.

[Reet]

Rival R152PK08. NLA driver. Uses a Wavecor balanced drive motor, but with different soft parts. Compared to the WF152 above, this one has a larger voice coil at 1.5" diameter, and a larger magnet as well. Cone is paper with kevlar fibres.


Frequency response. Big breakup, but manageable.


IB estimation.


Distortion at 2.8V, very low through the critical midrange. I probably should have re-run this measurement, the 4th and 5th harmonic <600Hz doesn't show up in subsequent measurements.


Again at +5dB, about 5V. The big breakup actually clipped the input signal at 6kHz. I didn't want to adjust the input gain setting, so you're going to have to disregard the distortion peak there.


Tested with filter applied for 2.5kHz LR4 target for typical use. With BSC, this is about 5V at low frequencies, 2.8V at 1kHz.

[Reet]

A "vintage" Peerless 2" dome midrange.


Not much to look at, the big plastic faceplate makes it somewhat unappealing. I did disassemble this one, it could perhaps use a bit more stuff in the pole piece. Unfortunately, the dome and voice coil assembly is glued to the faceplate, so giving this one a face lift might be challenging.


Frequency response. Limited low end, otherwise quite usable.


IB estimation.


Distortion tested at 2.8V. Not great, fairly high 3rd hamonic.


Again at +5dB, about 5V.


Filter set up for 600Hz high pass LR$, 2kHz low pass LR4 target. Tested at 5V.

[Reet]

Acoustic Elegance TD6M. High sensitivity 6" with a massive motor structure, these are about 5dB more sensitive than the other midwoofers tested above. I'm not a huge fan of the basket on these, very small lip leaves your screws very close to the mounting edge for such a heavy driver. It would also be great to hide the voice coil leads under the cone for a cleaner visual appearance.


Frequency response isn't great. A fairly signficant hole in the midrange is not ideal. Breakup isn't too bad at least.


This driver would be easily dealt with an active / DSP filter. Passively, it's a bit challenging. The low pass rolloff is easily worked to a LR4 target with a simple 3rd order filter, however you'll have to accept that midrange dip.

IB estimation. The small sealed bookshelf limits the low end response on this one.

Distortion tested at 2.8V. Good news is distortion is low and clean throughout the entire spectrum. This also supports the frequency response correction to not have any ill effects on HD products.


Tested again at +5dB, about5V. Quite loud.


With a filter in place, 2.5kHz LR4 target. With BSC, this is about 5V at low frequencies, 2.8V at 1kHz.
It was a bit of a wacky filter shape, but you can see the driver performs very well with it.

[theSven]

It's been a super hot minute since I've seen a post or posts like this of a driver testing marathon. Thanks for sharing all this data here on the forum Reet!

[Reet]

I hope there’s some useful information in there, measuring all these drivers was actually quite fun, posting it however was more tedious than I thought.

I have a few more I could add to the mix, a DX20 tweeter, tiny little bugger, I need to cut a baffle plate for it, and probably not much information to be gleaned as there’s already plenty of reliable information on it online. I’ll also do the full gamut for the Melodavid Be28 at some point. I could add in a CSS LD22 in a “Augerpro” 3D printed waveguide, or perhaps Dayton AMT3-4. Other possibilities could be some cheapies like BC25 or Dayton DA135, again I don’t think there will be much to be learned from testing these.

for the Fountek NeoCD3.0, I have one with a broken ribbon. I thought it would be fun to experiment with just a strip of aluminum foil as a ribbon replacement. The foil will be thicker than the original ribbon, but can do a precision cut on it with a Cricut machine. The assembly of the tweeter itself is quite odd, so installing the ribbon will be the biggest challenge.

Down the road I plan to use the test cabinet for some waveguide testing, either CNC milled or 3D printed. There’s a few tweeters that could be easily adapted to a waveguide I think, which will be interesting to compare to the baseline above.

[Reet]

Here's a power & DI comparison between some different tweeter shapes. Axial response is normalized flat, for comparison of only power and DI behaviour from >1kHz.

810921 1" "normal" dome.


Fountek NeoCD3.0 ribbon.


Wavecor TW030WA11, 30mm dome with small waveguide.

[tktran]

Thanks for sharing your measurements? That was a lot of data to wade through and digest!

One of the favorite tweeters is the Scan-Speak R2604/8330. It's ruler flat frequency response, reduced output at in the top octave, and practically absent high order harmonic distortion are the highlights. Some people have been concerned it's lack of ability to cross low. Reet do you share that concern?

Another take home message is that although the NeoCD3.0 has "higher distortion" when compare to top flight domes. But compared to a good mid-woofer it's OK:

Fountek:

https://www.htguide.com/forum/filedata/fetch?id=950417&d=1704600182

VS

https://www.htguide.com/forum/filedata/fetch?id=950448&d=1704602165

VS

https://www.htguide.com/forum/filedata/fetch?id=950473&d=1704611922

[tktran]

deleted (duplicate post)

[Reet]

Thanks for sharing your measurements? That was a lot of data to wade through and digest!

One of the favorite tweeters is the Scan-Speak R2604/8330. It's ruler flat frequency response, reduced output at in the top octave, and practically absent high order harmonic distortion are the highlights. Some people have been concerned it's lack of ability to cross low. Reet do you share that concern?

For R2604/8330 I don't have much concern that way, I think it's usable down to 2kHz. When I set up again for some more testing, I could run the R2604 at higher SPL and with some filters in place to get a better idea. I believe the commend on low frequency performance relates back to the original Vifa XT25TG30, which I use successfully in a WG148 down to 2kHz. The other issue with XT25 is a fairly high Q impedance peak, so some attention to the actual acoustic response slope may be needed, possibly the comments from the old days were just poorly implemented crossover filters.

[Reet]

Melodavid Be28. I'm not confident this is a solid Be dome, some PVD process with Be is possible. It has a flat appearance like many ceramic type of tweeter coatings, but darker in colour. I really like the look of this tweeter.


Frequency response is very good through the audible range, it has a strong resonance around 27kHz.

IB estimation.


Distortion tested at 2.8V. 3r5d harmonic at the noise floor here.


Testing again at +5dB (~5V) shows an increase in 2nd order, 3rd order not so much.


with 2kHz LR4 filter in place, tested at 5V.

[Reet]

Melodavid D28S, a Dynaudio homage. Thanks to tktran for the kind donation. It looks very nice, the dome material I think is a fair bit softer than what you'll find in a genuine Dynaudio however.



Frequency response is very good apart from a peak >15kHz. Easily taken care of with some low cost passive components, or you could leave it alone, your choice.


IB estimation.


Distortion tested at 2.8V. Very good indeed.


Tested again at +5dB (~5V). Low distortion down to 1kHz makes this a very flexible tweeter.


With a 2kHz LR4 high pass filter, tested at 5V.

[Reet]

I should mention, as an upgrade option for the D28S, there is a slightly more expensive model available, the T380Nd which combines the soft dome of the D28S with the motor and chamber of the Be28. Could be an excellent option for those who favour soft domes.

[Reet]

Dayton AMT3-4.

Hard to say what sensitivity this speaker is since it's entire response is on an uphill slope. At the top end its quite sensitive. Main downside to the is tweeter is limited top octave, it peaks around 12kHz. Low end is frequency response limited.


Vertical plane, positive angles (listener above speaker)


IB estimation.

Distortion at 2.8V. Keep in mind the lack of output below 3kHz skews the normalized distortion plot here.


Again at 5V, quite loud at the top end. Not much change in overall distortion.


Filtered how I have used it, crossover is roughly 4th order Bessel at 2.7kHz. Tested at 2.8V


Tested again at 5V.

[Reet]

Vifa/Peerless DX20BF00-04, tiny little bugger.



Frequency response has as high Q bump on the low end, otherwise very good, at least on-axis. Very wide dispersion up to about 6.5kHz where it starts to take a dive. Could be a challenge to optimize power response on a speaker with this small tweeter.


IB estimation.


Distortion tested at 2.8V


Again at +5dB (~5V). Quite a capable little bugger.,


Tested with a filter target of 2.5kHz LR4, at 5V.

[Reet]

Updated first post with a driver index, linked to the specific posts for each driver.

Coming soon - Melodavid CM25Nd, and a B&W FST 5" midrange from model 804D4.

[Reet]

B&W FST 5", model LF27715. This is an interesting and impressive speaker! Big neo motor, the cone edge just folds back and is fixed to a piece of foam on the frame edge. The "spider" is some surprisingly rigid wiggly sticks. There are no screw holes in the frame, just decorative trim. To mount the driver you run a M6 threaded rod through the motor, and then bolt it through the back of the cabinet. The dust cap is a foam plug that you can pop out to install the rod. There's more info on these drivers if you search through the B&W section of this forum.



Back side of the cone is coated, so there are no gaps in the weave like many woven fibre cones. looks similar to lacquer.


It's hard to find a good comparison for this thing that isn't a pro driver. I truly believe something like this is what is missing in the DIY market. It has high sensitivity and low xmax, limited low end so perhaps a dome midrange is a good comparison. It'll have to be a darn good mid dome to compete as you'll see.

Frequency response. I'd estimate sensitivity in the 93dB range through 1-2kHz.


IB estimation.


Distortion tested at 2.8V.


At +5dB (~5V).It was quite loud.


The frequency response has a lot of lift in the top end, but very controlled, so the response is easily shaped. Here I used with a filter with target of LR4 at 400Hz high pass, 3500Hz low pass. Tested at 2.8V.


Again at 5V.


At low level, 0.9V distortion is near background noise levels. TV is on in the basement, room below causing the spikes in the noise floor



Impedance comparison of the pair of drivers, fairly well matched.

[tktran]

It's hard to find a good comparison for this thing that isn't a pro driver. I truly believe something like this is what is missing in the DIY market. It has high sensitivity and low xmax, limited low end so perhaps a dome midrange is a good comparison. It'll have to be a darn good mid dome to compete as you'll see.

in fact, I’ve never seen anything as high in sensitivity and smooth in frequency response in a pro driver.

I’ll have to do some Polar Maps, but as far as I can glance, it will be definitely be better than the 6” FST. But it will be interesting to see how it compares to a 3” dome mid.

Made in England without costing a small fortune, unlike some other European drivers. Definitely a hidden gem.

Now, to go to find some IKEA Blanda bowls, or visit your local gardening store to hunt for a spheroidal pot…😂

thank you to 5th element to starting us on the FST journey a decade ago…

[5th element]

thank you to 5th element to starting us on the FST journey a decade ago…

Well you know it was Zaph that originally did that, but as soon as I saw the measurements I knew I needed a pair!

The 5" looks to perform very well. The only fly in the ointment would be the rising third order that seems directly linked to the rising frequency response. The 6" kevlar doesn't suffer from this at all and the 6" continuum doesn't either. The distortion rise is fairly mild when you consider the frequency response rises by 20dB though!

It looks like B&W just need to refine the cone a little more, as with the 6" versions, and it'll be perfect.

[tktran]

My back of the napkin calculations show that the 5V measurement blasting up to 98dB! VERY LOUD!

With a 93dB/2.83V driver, I'd be interested in seeing how it does at 94dB (2.83V) with the filters in place, and again at a lower level, say 86dB
(or 0.283V -10dB; whatever is convenient for you) so we can compare with ancient 3" dome with a ferrite magnet...


best regards,
Thanh

[Reet]

My back of the napkin calculations show that the 5V measurement is a blasting 99dB. With a 95dB/2.83V driver, I'd be interested in seeing how it goes at 94dB, as well as 0.283V (85dB), with the filters in place.

Looking back and comparing to my other measurements, I'd put this driver at more like ~93dB through midrange 1-2kHz range. I've edited the post above to reflect this. Still quite loud. Just to clarify, you are asking for a measurement with the bandpass filter in place? At lower level 2.8V and 0.28V?

Let me know if you want me to move this into the Directivity testing thread.

Yes please.

[Reet]

theSven JonMarsh can you approve above post #31, too many edits..

[theSven]

I approved the post for you.

[Reet]

Melodavid CM25Nd. "Ceramic Alloy" dome, the motor appears to be the same as the more expensive Be model, but with a different rear chamber, same as D28S. Cost is about half of the Be model so I am hopeful that it can be a real gem. Thanks again to tktran for the kind donation!



This one is interesting. I have 2 samples of this driver that were ordered a couple weeks apart, so they were not ordered and shipped together. Same is true for the D28S above, which I have not yet tested the 2nd sample. First one showed a bit of a double peak impedance, which indicates some back-pressure either due to leakage or from a reflection.



Frequency response looks okay however. What is interesting for this dome (unlike the Be model) is the lack of sharp resonant peak. There is a bit of a dip and peak in 17-18kHz range, that is the breakup.



Distortion shows a fairly sharp rise on the low end, otherwise excellent. Perhaps we can improve on this.



When the second sample arrived, I was surprised to find that they were not a very good match. The second sample is even worse, showing some significant swing in the response at the low end. Dashed line is sample 2.


First, I thought that perhaps the chamber is not well sealed, there appears to be visual gaps between the neo discs that have not been filled. (both tweeters appearance was similar.



I glopped some glue in the gaps, but it didn't make any difference, so we can rule that out.

Next, I noticed that the amount of ferrofluid appeared to be lacking, there was not a complete bead bridging the gap, the second sample had nearly none at all.





Luckily, I had some ferrofluid from an AliExpress shop leftover from previous tweeter restoration (D27 and MDT29 above), so I added some in. This did improve things, making the impedance look a bit more "normal", however the frequency response still didn't match well, and the distortion didn't change at all.


Moving on to the chamber and damping. Both tweeters had some felt in the pole piece, the second sample appeared perhaps a bit more mushed in there, but not a big difference. However, pulling out the felt, it appears the rear chamber is left empty on these tweeters!





The chamber void is quite small. I grabbed some fluff from around the house and was able to stab it into the chamber void with a chopstick. I then reinstalled the felt, only just to the end of the pole piece flange. With everything reassembled, I played pink noise for a minute to make sure the ferrofluid is well distributed and settled.



Impedance is changed once again to a broad lump, however not a perfect match between the two. I found that the ferrofluid had the greatest impact on the resonant peak, so very likely that one tweeter has a bit more fluid than the other now.



To be continued...
(15 attachment max per post)

[Reet]

Frequency response is now well matched between the two tweeters, a success!



Full polar response chart of the complete tweeter with added ferrofluid and chamber stuffing.



Distortion is now improved too, this tweeter can now compete with the best! Measured at 2.8V


Again at +5dB (~5V).


With a filter targeted LR4 at 2kHz, measured at 2.8V.



Again at 5V.


So there you have it. If you are willing to do your own QC checks and possibly add some ferrofluid and chamber stuffing, this is a top tier tweeter with mid tier pricing. Ferrofluid can be purchased from Aliexpress for about $12 for 6mg which is enough for at least a dozen tweeters, probably even 2 dozen. You'll need a syringe, application is actually quite fun, the stuff literally gets sucked out of the syringe into the magnetic gap. I tried to use just enough to bridge the gap, probably a bit more than necessary once you insert the voice coil.

I'll find out shortly if the D28S samples are well matched. I can confirm that the Be28 pair were well matched, however were also purchased together as a pair.

[Reet]

The 5" looks to perform very well. The only fly in the ointment would be the rising third order that seems directly linked to the rising frequency response. The 6" kevlar doesn't suffer from this at all and the 6" continuum doesn't either. The distortion rise is fairly mild when you consider the frequency response rises by 20dB though!

It looks like B&W just need to refine the cone a little more, as with the 6" versions, and it'll be perfect.

It's funny you say that. When testing the Melodavid today, I checked and realized I only finger tightened the bolt down for the FST. I gave it a crank to snug it up better and retested. Distortion images are updated above, I think you might be a bit more pleased with it now . Original measurements are now gone for good, no one shall ever know what they looked like!

tktran, filtered measurements at 2.8V and 0.9V are below shown as well.

[Reet]

I inspected the Melodavid other Melodavid tweeters. The CB25 has the smallest chamber, but it did have a thin felt lining. D28S has the same chamber as CM25, also empty, however the larger ferrite motor makes for a longer pole piece which got two stacks of felt which I think is doing a lot of the heavy lifting with that one. It also has a felt ring around the top of the pole piece like you see in similar Dynaudio and Morel tweeters.

Be28 also appears empty with the same piece of felt in the pole piece, however it's chamber design has a more aggressive dimple that may be helping a bit with that one.

Perhaps splitting hairs since their already quite good performers, the greatest benefits from some small modifications would be to the low end of the spectrum <2kHz. Potentially all the tweeters could benefit from a small bit of damping material in the chamber, and I think all could benefit from just a little more ferrofluid too.

[Reet]

For comparison (and to make sure I'm not crazy), Vifa D27 has a felted chamber, however pole piece is left empty.

I also checked Wavecor TW030WA08, no stuff in the pole piece, however an impressive chamber design, and felted. It's no surprise these perform well on the low end. Also, ferrofluid application here is adequate.

[tktran]

The 5” FST does measure well indeed!

You know that blip around 8-900Hz?

Well, with drivers that have such low distortion, sometimes you can measure things like loose screws, nuts or cabling; which can manifest as
”distortion” (all harmonics are raised at the single narrow band of frequencies)

[Reet]

Of course, at these low levels there can be many things that can "contaminate" the output from the driver, we have early reflections, panel resonances from the cabinet, wires, screws, backwave reflections, and whatever else in the room that could be buzzing. My setup isn't exactly a meticulously arranged lab environment, but more of a "quick and dirty" result, which is enough for me to make decisions on which drivers to use for projects and how to use them.

[Reet]

I'll find out shortly if the D28S samples are well matched.

They are well matched. If anything, the second sample is a slightly better performer than the first.

[tktran]

Reet

Re:
The Line Audio Omni1 also is not a "real" measurement mic, however I have found it's response to be within 1dB of my Omnimic, it doesn't have a long stick so there is some early reflection influence in the top end creating some extra wiggles there. Keep that in mind, however there is still plenty of information to determine if the tweeter is problematic or not.


How much better is the Line Audio Omni1 for distortion measurements than the Omnimic?

have you had a chance to compare and contrast with the Omnimic for Really really really ridiculously good looking driver like the FST, where you’re getting distortion around the noise floor of your room (-60dB when playing 98dB)