Zaph's New Test Results for Pro 6.5" Drivers...

[jkim]

Finally Zaph added some pro drivers to his 6.5" test group. The results are very interesting. It turns out that the light cone, high efficiency design of pro midrange drivers is achieved only at the cost of low end extension, but not at the cost of distortion performance.

The use of these drivers should be different from that of others in the group. They are not suitable for a 2-way design. We can evaluate their performance only above 300 Hz---perhaps above 200 Hz for the 18Sound 6ND430-16. The distortion level of the 18Sound is amazingly low. It is comparable to the best hi-fi driver in the group in the 200 Hz to 2 kHz range---the Usher 8945P. Its frequency response is linear. Sensitivity is over 89 dB at 16 ohm nominal impedance! I'm not sure its 8 ohm and 4 ohm versions are available. If so, I'd expect the same level of performance with higher sensitivities.

The B&C 6MD38-8 is also an excellent midrange driver. 400 Hz is the lowest xover point I'd use with this driver, though.

These results suggest that we don't need to use more expensive hi-fi midrange drivers like SS 12M or 15M to achieve ultimate performance.

This time Zaph added two more hi-fi drivers---Usher 8836A and the new Tangband W6-1721. The 8836 result shows Usher's consistently good motor design. The new TB's motor also turns out to be excellent. 2nd order harmonics at 800 to 1000 Hz are a bit high, though---I don't think 2nd order harmonics in this range will be very obtrusive. At $52, it is very attractive. According to Zaph's FR measurement, it seems that it won't need much BSC in a 2-way design, so its relatively low sensitivity won't be a drawback.

Thanks for the work, Zaph!

-jAy

[eyekode]

I have been checking everyday for this update . It is nice to see the new TB. Really nice high order THD numbers. Puts it in competition with the RS180 for value.

Also bodes well for the W4-1720. I think that would make a very interesting MTM (crossed to a subwoofer of course).

[Deward Hastings]

The problems are not just at the low end. That 6ND430-16, for all it's apparent "low distortion", goes off like a kazoo just over 5kHz . . . as bad as any metal cone and maybe worse. You can see it in the harmonic distortion curves (3rd at 1.7k and 5th at 1k) and most notably in the CSD visualization, where it sticks out like . . . well . . . a couple of them might be real nice crossed around 1k to a large mouth horn, with a bass bin under. Except for the sensitivity I don't see it bringing anything to the table that can't be had for half the price elsewhere.

[jkim]

The problems are not just at the low end. That 6ND430-16, for all it's apparent "low distortion", goes off like a kazoo just over 5kHz . . . as bad as any metal cone and maybe worse. You can see it in the harmonic distortion curves (3rd at 1.7k and 5th at 1k) ...

To me, it doesn't seem to be as bad as metal cone drivers. Take a look at L18 and RS180's high 5th and 3rd order harmonics between 1.4 k and 3 kHz. If these metal cone drivers can be crossed at 1.2 k to 1.5 kHz, then no one would argue that the 18Sound can't be crossed at 2 kHz. Those peaks (3rd at 1.8k and 5th at 1.1k) are still no higher than the distortions of other good drivers (e.g., ER18RNX, 830883, CA18RNX, D6.8) at those frequencies. Its cone resonance is not excited by lower frequency signals as badly as metal cones'. The effect concentrates at its resonance frequency alone, which can be easily controlled by a notch filter.

[Deward Hastings]

Take a look at . . . RS180's high 5th and 3rd order harmonics between 1.4 k and 3 kHz. If these metal cone drivers can be crossed at 1.2 k to 1.5 kHz, then no one would argue that the 18Sound can't be crossed at 2 kHz.

You've got it backwards . . . the RS180's breakup is *higher* in frequency, allowing a *higher* crossover. The RS180 has a 5th peak at 1.8kHz and a 3rd peak at 3kHz, both reflecting the breakup peak at about 9kHz . . . (there's also that breakup cancellation null at about 4kHz to avoid, and the slightly smaller peak at around 7.5kHz). It *shouldn't* be crossed much if any above 1600. Applying the same standard to the 6ND430-16 it's much lower 5kHz breakup and the harmonic distortion peaks at 1 and 1.7kHz suggest that it shouldn't be crossed much over 900Hz. Of course it will *work* crossed higher than that (if you like the sound of flapping paper). But that 5kHz breakup is *severe* . . . look at how long it rings. At least the drivers with the thicker felted cones damp their breakup some, and hide it better from the simple distortion tests.

[jkim]

Applying the same standard to the 6ND430-16 it's much lower 5kHz breakup and the harmonic distortion peaks at 1 and 1.7kHz suggest that it shouldn't be crossed much over 900Hz.

You can't apply the same standard to the 6ND430-16 as you do to metal cone drivers. Harmonic distortions induced by the cone resonance are negligible in the case of this driver, as I already noted above. You should note that all drivers behave differently in the degree of the nonlinear effect of cone resonance. In the case of this 18sound, it shouldn't be an issue. No problem with 2 kHz LR4, or even with 2.5 kHz LR4 with a notch filter.

P.S. We no more want to argue about the nonlinearity of cone resonance. We've already had it enough as you know.

[tktran]

I hate the high sensitivity drivers with 0.5mm xmax and ragged response, but this 18Sound driver measures great.

Available in 16,8 & 4 ohm incarnations.
http://www.eighteensound.com/index.a...roduct&pid=243

Would work well in a 3 way, but with 5mm xmax, would work well as a midwoofer in a 2way crossed at 2Khz. Higher sensitivity is a bonus.

I wonder what box size it wants...

Looking forward to learning more about this brand/line of drivers.

[Pete Schumacher ®]

You've got it backwards . . . the RS180's breakup is *higher* in frequency, allowing a *higher* crossover. The RS180 has a 5th peak at 1.8kHz and a 3rd peak at 3kHz, both reflecting the breakup peak at about 9kHz . . . (there's also that breakup cancellation null at about 4kHz to avoid, and the slightly smaller peak at around 7.5kHz). It *shouldn't* be crossed much if any above 1600. Applying the same standard to the 6ND430-16 it's much lower 5kHz breakup and the harmonic distortion peaks at 1 and 1.7kHz suggest that it shouldn't be crossed much over 900Hz. Of course it will *work* crossed higher than that (if you like the sound of flapping paper). But that 5kHz breakup is *severe* . . . look at how long it rings. At least the drivers with the thicker felted cones damp their breakup some, and hide it better from the simple distortion tests.

Comparing the RS180 to the 18 Sound driver, you can see that the 5th harmonic on the 18 Sound is far lower, across the band, than the RS180. In fact, the only place where the RS180 is better is in 2nd harmonic. The 18 Sound dominates on odd order products, with the RS180's 5th harmonic as high as it's 2nd harmonic in the crossover range between 1 and 2KHz. There's really no contest. The 18 Sound unit appears to offer an even cleaner spectrum across the usable band than almost any other driver in the group, excepting perhaps the Usher 8945P, as JKim has already pointed out.

CSD is merely a reflection of frequency response smoothness, and the 18 Sound is very smooth in its passband. Of course you'd not use it to 5KHz, for a variety of reasons, the main one being that it probably is beaming like a laser at that frequency. However, XO points in the 2KHz to 2.5KHz range would be perfect for this unit.

[eyekode]

I haven't built a ton of speakers so it is hard for me to judge how these measurements _sound_... But with the RS180 why is there an issue with an f5 of -60db at 1.8k?

Shouldn't this be inaudible? I am not exactly sure how the math works but won't this mean a 90db 1.8k tone will have an overtone/harmonic of 40db? Seems like I would not be able to hear 40db with a 90db test signal. Not to mention it would be beaming this pretty fiercely so any off-axes listening position would be even lower.

[Zero the Hero]

Jay, can you explain why you wouldn't use that B&C lower than 400hz? I kinda wish he'd tested an 8 or 10" B&C, but I probably shouldn't look a gift horse in the mouth

[jkim]

Jay, can you explain why you wouldn't use that B&C lower than 400hz?

Despite its low xmax, maybe it can be used lower than 400 Hz because of the high efficiency. But its 160 Hz Fs may be a problem.

[Deward Hastings]

with the RS180 why is there an issue with an f5 of -60db at 1.8k? . . . Shouldn't this be inaudible?

Excellent question . . . and it goes to the root of the difference of opinion that I have with some others here regarding what it is we are (should be) measuring and what there is that we hear. In the view of some that harmonic distortion is all there is, and if that's low (or low enough) everything's fine, because below a certain level it's either masked or we "can't hear it" anyway. My view is that, useful as they are in many ways, harmonic distortion measurements fall far short of measuring *all* distortion, including some kinds of distortion which we *do* hear, and that they are at best a "proxy" for the distortion measurements we *should* be making. In fact the very nature of the way we measure harmonic distortion, our instruments themselves, are designed to suppress other distortions that we are not looking for. So I look not only at absolute levels of harmonic distortion (which is mostly a function of motor nonlinearity, not the cone) but also at the abrupt *changes* in distortion level (which are not caused by the motor but reflect something else non-linear in the system). It is the broad spectrum effect of those other unmeasured distortions which I contend produces the *sound* of "cone breakup" . . . those tell-tale harmonic distortion peaks are merely the signature that we see (with a very limited test).

A historic analogy would be distortion in solid state amplifiers. In their early incarnation many such amplifiers tested very well . . . you'd see harmonic distortion figures ten times lower than the tube amps of the day. Distortion measured 60dB (or more) down at 50 or 100 Watts, and we saw advertisements with claims like "a straight wire with gain". Those same amplifiers often *sounded* terrible. It wasn't until designers realized that crossover distortion and TIM distortion, *which do not show up in steady state harmonic distortion tests* were *real*, and the cause of the problem, that the problem itself got solved, and now we have $5 "chip amps" that don't have those problems, and they sound great.

My contention is that that cone breakup is like crossover distortion . . . it's real, it's audible, and it is not measured but only hinted at in conventional harmonic distortion testing. Look at the "ridge" in the CSD display, visualize what is happening to the cone when parts of it are moving independently of the rest of it and of the voice coil, and tell me (with a straight face) that has no effect on the music signal that the voice coil is trying to get the cone to follow because it doesn't show up on a single tone harmonic distortion test.

[augerpro]

I'm not all that impressed with the TangBand. The way F3 increases towards the bass implies a fairly bad BL asymmetry? John if you read this did you do your informal Le(x) test on the driver?

[3way]

Tang Band also sells an 8" model W8-1722 of similar under-hung motor construction. I really, really, hope Parts Express will stock this 8" driver and really, really, really hope Zaph will test it when it becomes available.

The specs show it could be used as a "full body" midrange in a 3way, or provide superior 60Hz Xover to a sub in a 2way MTM as common for HT. I favor higher efficiency full bodied midrange speakers. The Linkwitz Orion uses an 8" midrange.

I really, really, hope Parts Express will stock the 8" Tang Band W8-1722, including any 8 ohm versions.

[Zero the Hero]

I'm not all that impressed with the TangBand. The way F3 increases towards the bass implies a fairly bad BL asymmetry? John if you read this did you do your informal Le(x) test on the driver?

looks like built-in BSC

[3way]

I'm not all that impressed with the TangBand. The way F3 increases towards the bass implies a fairly bad BL asymmetry? John if you read this did you do your informal Le(x) test on the driver?

Does anyone know if Zaph performs a suspension break-in before testing drivers? I did not read about break-in on his 6.5" test notes, but I'm also new to Zaph's work.

Has anyone else seen speaker test results change after a good break-in stress?

[Wolf]

Tang Band also sells an 8" model W8-1722 of similar under-hung motor construction. I really, really, hope Parts Express will stock this 8" driver and really, really, really hope Zaph will test it when it becomes available.

The specs show it could be used as a "full body" midrange in a 3way, or provide superior 60Hz Xover to a sub in a 2way MTM as common for HT. I favor higher efficiency full bodied midrange speakers. The Linkwitz Orion uses an 8" midrange.

I really, really, hope Parts Express will stock the 8" Tang Band W8-1722, including any 8 ohm versions.

This is a woofer, probably will not link with most tweeters, and AFAIK is only available in 4 ohms impedance. It is on PE's TB listed items, but I don't know if it's in stock yet.

The Orion uses a W22?? I thought it was the W18. I guess I forgot that.
Later,
Wolf

[augerpro]

looks like built-in BSC

For sure, and with the low unductance a noob could probably use textbook filters and get a passable loudspeaker.

[Deward Hastings]

The Orion uses a W22??

Yes . . . and gets away with it because of a low and steep crossover (1440 Hz., LR4 *electric*, not "acoustic") and a notch filter to boot. Those things combined keep signal off the W22 cone in its problem areas, and without them (with either a higher crossover frequency or a lower electric slope) the ORION midrange and treble would sound like scat. It is not just the "open back" design and its lack of box and cabinet resonances that makes ORION sound so good . . .

[jkim]

I'm not all that impressed with the TangBand. The way F3 increases towards the bass implies a fairly bad BL asymmetry? John if you read this did you do your informal Le(x) test on the driver?

Yeah, I've noticed that, too. I think that the driver's shelving FR is the culprit. If its FR is really like that, we won't need much BSC. And the HD profiles of other drivers with some BSC incorporated will become similar to that of this TB.

BTW, the same tendency is observed with the SB17NRX35, too, though it's a bit less severe. These two drivers behave very similarly. Then, what's the cause of this from a motor design point of view?