Improving Tangband W3-1797S Flatcone

Hello all,
The W3-1797S is the 3" sandwiched aluminum flat cone unit from Tangband, which a number of you have experience with. In my application, the driver performs quite well in all but just one respect – the highs come off somewhat harsh and lack fine detail. I wish to know, to what extent could this be improved by a filter or any EQ?

It's really almost near-perfect for my use until the very upper ranges. Dispersion is remarkable as others have noted but lacks the fine grain detail of a quality dedicated HF unit – which is not surprising of course.

Any successful experience among others smoothing these out by some means?

Kind regards,
Luca
 
Last edited:
Are the drivers burnt in?

As I have noticed filters or equalizers just ruin the picture of this flatcone, play with positioning and horizontal/vertical orientation

how do you use them, any pics?

I've used them first in the original packages as enclosures, and that was perfect, then I made a spherical enclosures with minimal baffle, and that's not the near as good as with original packages, maybe there is something in this resonating enclosures, I must research! lol :D

[IMGDEAD]http://ultraphoto.org/images/t52f6ajotegb6rtrl7td.jpg[/IMGDEAD]
[IMGDEAD]http://ultraphoto.org/images/elzceahqyqxb4xxlmfq.jpg[/IMGDEAD]

[IMGDEAD]http://ultraphoto.org/images/lrmjr767uyntq1ha22n.jpg[/IMGDEAD]
[IMGDEAD]http://ultraphoto.org/images/lks58t639xlnt033ml60.jpg[/IMGDEAD]
 
Last edited:

ultrakaz

Member
2004-10-20 11:32 pm
Try sticking a round-ish piece (a bit smaller than a dime) of felt on the front of the built in grill and it seem to smooth the treble a bit. YMMV. Another way is to stack two of them, one on top of the other, like the commercial speaker (I can't remember the name) that uses the Tang Band. Apparently, there is a sympathetic and/or favorable cancellation that smooths out the highs.
 
Boenicke audio.

Pity the driver's such a swine to get hold of now, it was a nice mid-tweet, albeit not perfect, as you've discovered. The flat profile gives that wide HF dispersion, but the price is that very flat cones are also harder to control; you can see in the (smoothed) plots provided by the manufacturer where there's +10dB gain peaking at 15KHz relative to the nominal indicated (not that they're alone in either of those matters)

Two per channel should as suggested take the top end down through the outputs lobing -not automatically a bad thing if gone into with eyes wide open. Ditto for sitting off-axis, which will flatten the top end out. You could try a notch filter and / or stick a Zobel on them which can often help tame problematic units. Beyond that, you're into mechanical changes, either by adding felt / whatever, or physically isolating and damping those parts of the cone where the resonance is insufficiently controlled. It's probably not just the cone, but that's about as far as you can go on that score without disassembling the entire drive unit and modifying components, which is something beyond what most DIYers are able to practically achieve.

Whether any of that will improve 'detail' is another matter, since it depends what you mean by it. 'Detail' can be an absense of distortion, or increased distortion of a particular kind, e.g. an emphasis in a certain frequency band, higher levels of 3rd harmonic distortion etc. YMMV on that score.
 
Last edited:
I currently reside in Taipei where the Tangband head office is located and where access to those units is less of an issue. I've been in touch with TB on a number of occasions over the years, and have had a long-standing invitation to visit them, and intend to do so in the coming months.

Beyond adding a small amount of felt-like material to the center, I'm interested in the more involved modifications, beginning with the filter (what might an appropriate Zobel look like/what frequencies are targeted?) and even – if it were possible and within a certain cost – the very advanced modifications to the driver's design. Don't be hesitate to suggest the impractical. I'm interested in what might be optimum, hypothetically, regardless what that entails.

The driver is housed in a small spherical enclosure, sealed.
 
Less a matter of hesitation and more a case of being unable to give any particularly detailed thoughts regarding physical modifications without having some here to measure and pull apart. All drive units are a blend of different compromises.

For the Zobel, it's just a straightforward shunt resistor & cap. in parallel with the drive unit to flatten its rising HF impedance. Normally they're used for woofers / midbass drivers to ensure the XO has a constant load, but they can (can) sometimes help get some control into a widebander. A notch filter is different in that it's principally designed to attenuate a specific frequency region, e.g. an uncontrolled resonance peak. Ideally, you'll need to measure your drive units in the box they are intended to be used in / are being used in, so you know exactly what they're up to.

FWIW, if you're running them in a sphere, try getting some more damping material into the middle of that box.
 
Firstly, many thanks for your input. Both the amount and type of damping material was sorted through earlier, certainly to noticeable benefit. The Zobel would be the next area of experimentation. I've been reluctant to opt for a FR over a quality coax, but another round of trials with the W3 is called for. The coax used earlier is a hybrid of parts, and does not exist ready made (after years of attempting to source it anywhere on globe) Before exploring a ground up design for such a driver in collaboration with a local OEM here, the limits to any improvement on the W3 should be found.
 
Last edited:
If I proceed with this, it would be in the context of a very small satellite-style speaker, where the Tangband handles everything above 120hz, approximately, except for the very top frequencies, covered by a small dedicated tweeter.

I simply do not care for the way the TB performs at the top end, finding it lacking in various ways. The use of a diminutive 1/2 inch or 3/4 inch HF unit crossed over exceptionally high is under consideration, a textile dome type, with all units managed by an external miniDSP. It's an experiment. The TB may be coupled as well to a passive radiator or used as a pair in a push-push configuration, attempting to gain greater extension in a very low volume enclosure.
 
Last edited:
Keeping it above 250hz? I believe you might be thinking of one of the two 4" aluminum sandwich flat cones from TB, which is a purpose-built midrange driver. The others they offer do perform quite well without any high pass filter on them whatsoever.

I haven't yet access to the proper equipment to measure and adjust for baffle step, but will have in the coming month. Can you estimate where the baffle step will be?
 
Last edited:
Those 4" have an Xmax of 3mm, and this 3" has just 1mm, and also the problem with this 3" is that it has limited cooling capability... if you want to listen at normal or higher volume, I wouldn't do that at 120hz, not just because of the fact that it can't do that, but also because the Image would be thin and there is no weight and "Grundton"(German) with such a small driver... Bose drivers are capable of more excursion, but they have a gap between low range with this BP Sub, and that's not Hi-Fi...

here, I have also those 4" so that you don't think I'm blabbering rubbish :D
[IMGDEAD]http://i45.tinypic.com/6dv2c8.jpg[/IMGDEAD]
 
Last edited:
No, no thoughts of rubbish at all, I didn't mean to imply as much. The photo indeed reveals it's the full range 4" we both have in mind, not the mid 4" identifiable immediately by it's clipped edges on the plate. Seeing them side by side, the 4" really comes off appearing to be significantly larger.

I had experimented with the 3" for some time, finding that its bass performance seemed similar to that of other upper-end TBs, such as the titanium W3-1335SB., despite the very short excursion on paper. It's unclear to me – and I'm sure others will readily be able to explain – why the cone visually appears to have much greater extension than 1mm or 0.5mm in the case of the TI, by at least another 2.5mm or so.

Granted, there's little Grundton to such a small driver, particularly in these tight enclosures, but it almost meets my needs in the intended application. The fundamental oompf will always be lacking, but in order to improve this, even if only a tad, I'm wondering about trying a passive radiator or two, or some combination of push-push drivers. It would become quite involved, something akin to an even tighter version of the Beolab 3, with all the attendant costs. Nevertheless, it may be an experiment I'd go through with, which to some extent will depend on the feedback of others. There are no cost limits, at least for a few prototypes.

Regarding the baffle step and the enclosure, it's a 4" sphere, sealed. Based on that, can you infer what this might entail, in terms of working the BSC? Many thanks for the input...
 
Last edited:
I was joking with rubbish :D Yes maybe 3" has more Xmax than just 1mm, because TB often underrates a little bit the excursion of some drivers... Sphere is good because you can rotate it and try different angles of all directions, 4" baffle, what I know, the Baffle Step may be at about 1000Hz?!
 
Last edited:
Cheers, yes, TB is really underrating the excursion, and to no small extent, I'd say. The xmax is at least double what they're listing; I wonder why they choose to do this. Have you taken and posted any measurements of your speakers' response curves anywhere, (with these drivers of course)?

Xmax doesn't mean a great deal in itself, since there is no universal standard definition as to what it is, nor do any of the popular means of putting a figure to it tell you much of practical value. You really need to measure the distortion figures yourself unless the manufacturer provides them, or a reputable independent with appropriate equipment (as above) does it, if you want to find out what it's actually doing.

Note: Xmax does not define the mechanical limits of a drive unit's operation. To an extent, Xmech (Xlim) does this, but again, there is no standardised definition for that either. Theoretically it is the limit before damage occurs, but is this instant damage, potential damage or damage if regularly driven over some unspecified timescale? Depends on the manufacturer.
 
Last edited: