Tang Band speaker folly


I have a few questions (so far) regarding using Tang Band speakers. I am working on a kitbash of an Atlas mp15dc into an sw1500. I plan to install a Tang Band speaker under the radiator core a la Bob’s scheme for his Atlas switcher.

My first question is pretty reasonable, I think. The Tang Band speaker modules have two “speakers” in each, one a round cone and the other an oval shape. Which one is the primary speaker and which one is the passive radiator? Does it matter which one is positioned where in regard to the opening that the sound will come out of?

Do both speaker and radiator exhibit excursion, and to the same degree? I know Bob has indicated that fairly significant excursion happens for the larger Tang Band modules.

Now to the silliness. I know almost nothing about the science or rules behind speaker design, so this may be the dumbest question ever regarding speakers. I have both 1931S and 1925S modules on hand. I would prefer to achieve the extra range of bass response possible with the 1931S, but I must concede that it will not practically fit inside the shell, constrained by the width. There is plenty of room lengthwise for both the 1931S module plus a Loksound decoder. If all else fails, I will use the 1925S as a straightforward solution.

However, I’m curious whether the bass response is a product of the passive radiator, or is it a product of the overall design of the module, including enclosure. Would it be feasible to cut the face of the module out of the existing enclosure, with the primary cone and passive radiator still intact, and then build a new narrower enclosure around those parts. I have no idea if there is anything other than air inside the current enclosure, something like baffles maybe? I have no idea if the volume of air inside the enclosure is critical to the sound produced, or if dimensions are critical with respect to sound wavelengths or similar. I could see where these things might matter, but I don’t have the knowledge to know if they really do.

A 1.25" wide enclosure would definitely fit inside the shell, and possibly something as wide as 1.375" would fit. I want to get the best sound possible if I’m going to bother with it at all, and I think the bass response is critical for that. I don’t have to get the 1931S into this switcher, but I’d sure prefer to do so. If what I’ve proposed above is foolish, I can always use the 1931S in another project where I have more room.

Any ideas, tips, or admonishments are welcome and appreciated.



Jim -

In a rectangular TB speaker module, the round unit is the actual speaker, and the oval, the passive radiator. Most of the sound comes out of the round unit. The passive radiator and the enclosure have been specially tuned to work with the round speaker to extend the bass response about an octave. They work as a team.

The round speaker moves much farther than the passive radiator.

Best of luck fitting anything larger than a TB 2008 in an Atlas SW! It isn’t the width but the height that would be a problem, especially considering that room is needed for an improved radiator core.


As for cutting the module apart, compare the data sheets for the 1931 module and the loose 1931 driver. The driver is 31mm wide and ~19mm tall, not counting headroom for cone excursion. The module is 36mm wide and ~19mm tall. So with no sides on the box it would only be 5mm narrower. Of course, side walls are needed.

As for the interior, I have no idea what’s inside it.

You can, however cut off the mounting tabs on any of these modules, which I usually do with the 1925 and 2008. Then I mount it using clear Loctite double-sided tape, which itself is about 2mm thick.


Appreciate the insight on the speaker modules. I presumed that the round speaker was the primary sound source, but figured I might as well ask.

I never thought to look at the spec sheets for the individual drivers. If I had done so, I would have seen that there is a larger diameter flange that is hidden below the surface of the enclosure on the modules. I had measured the extent of the drivers and radiator faces in the module, and started dreaming based on that incomplete information.

Originally I had been planning on thinning the walls of the Atlas shell as well as the sides of the speaker module in hopes of fitting the 1931 module. That won’t be practical due to the recess in the left side of the shell for the handbrake. With respect to height, I have 3/4" from the bottom of the radiator core assembly to the top of the mounting platform that I intend to install from front to back, over the motor. That would be enough for the module itself, but not enough once speaker excursion is accounted for. The solution for excursion would be to only model the sides and tops of the radiator cores, with a hollow recess below that would net an additional 1/8" to 3/16" clearance. I have the advantage with the sw1500 that the cores are mounted closer to the top surface of the hood. In the case of the early switchers where the cores sit down 7", the 1931S could never work.

While I want the best bass response possible, the work required to implement the 1931S module has clearly reached a dysfunctional level. I’ll do the install with the 1925S and hope for the best. One of the other members on the OGR thread a while ago said that he actually preferred the 1925 to the 1931 so I’m hoping I can be satisfied as well.

Again, thanks for the help,


It was probably me that mentioned liking the sound of the 1925 better than the 1931 over on OGR. To put it in context so I don’t add to the folly: I was looking at it in relation to the power handling capability. I think the 1925 is rated at 1W and the 1931 at 4W. I am using HO Loksound Select and TCS WOW Diesel decoders (I use coreless motors). They have pretty small audio stages. I found that the 1925 sounds looser and richer somehow whereas the 1931 sounds tight and restrained. I couldn’t really hear any extra bass with the 1931. Of course this is 100% subjective.

So… I have the 1925 in everything, Atlas SW900 - even a Yoder 44 Tonner (a bit of filing needed) as well as my larger Diesels. Except for the 1942 I’m currently shoehorning into a P&D B Unit… :sunglasses:

The 1931 cone takes a lot more pressure to move it by finger than the 1925 so that’s likely part of it. But I did do more experiments and found that with the Loksound V4 L “O scale” decoder that the 1931 sounded better at higher volumes, and I could detect more bass. But it’s a trade-off for me - fit the L with the Pittman and I have less room for the larger speaker. Fit the HO decoder with the coreless motor and I have room for the larger speaker but can’t drive it fully (also could lead to a risk of clipping). Oh well, first world problems… :wink:

My buddies from the HO club still experience jaw-drop at the sound quality either way! :grinning:



Pete -

Do you have any photos of how you shoehorned a 1925 in that Atlas SW900? Did you gain extra room by installing a coreless motor and using an HO decoder?


I’m planning on using the existing Atlas motor and drive setup on this model, since they are known to perform well already. So no coreless motors for this model. Although by the time I’m done I’ll have modified or replaced enough parts that I should have just started from scratch. I would strongly consider a Faulhaber in that situation.

I’m going with a Loksound 4.0 decoder, and the 1925S module would be the smart (as in drop-in) choice at the moment. Although the idea of doing some filing, mostly on the shell but also minimally on the 1931 module will undoubtedly persist, lurking in the shadows til I finally give in and try it. I estimate that I need about .070" extra clearance to make the 1931 fit. There are ribs on the sides of the 1931 enclosures that measure at .0125" depth. Presuming those ribs are simply protrusions and not represented by grooves on the inside of the enclosure wall, I figure I can safely remove them and gain .025" at a minimum. The Atlas shell walls are in excess of .080" thick. Success would depend on being able to remove half the thickness of only one side of the shell so as to leave the hand brake recess intact.

This is the type of work that would be perfectly suited to a small end mill setup, which I don’t have. But I hang around a guy that is a retired machinist and he has both a Sherline and a full size mill, so I will approach him about the possibility of doing the work. He’s already done the machine work to cut down the Atlas frame for me, and he’s been following along with the project. It helps that he is exploring a small p48 layout to go along with his primary HOn3 layout, so he understands the madness.

I could always send off an email to the folks at Tang Band. They proclaim on their website that they will consider custom work. We need to get them to add another spec to their lineup that has the sound qualities of the 1931 but a width of on 1.25" to 1.375", perfect for putting into O scale shells. I probably need to order 1000 of them to make it happen. But I also could explain my situation and ask them if they can offer an alternate enclosure size that I could fabricate that wouldn’t compromise the existing bass response. It usually can’t hurt to ask.



Bob, I confess I misled you guys on the coreless motors a bit, in that for my Atlas SWs and MP15DCs, I still use the Atlas motor, sorry about that. It’s anything Pittman (except C&LS, based on your notes about Henry’s special motor) that gets swapped out for coreless.

I know you won’t approve of this, :wink: but I use Loksound HO decoders in the MP15s and a HO TCS WOW in the SW9. I haven’t had any issues with overloading but I only have up to 25 car trains on level track.

I will grab a picture of the SW9 next time I take the shell off. But I basically ripped out the Atlas board and lights, then made a styrene platform to sit on the four Atlas posts. It had to taper the cab end as I recall. I put the 1925 at the front end of the styrene platform, then the decoder on a brass heat sink, then the keep-alive at the cab end. I use the TCS KA2 with all my installs btw. But of course I have more space because I still have the ugly non-prototypical Atlas grilles, unlike your elegant solution. :blush: