Tang Band speakers


When installing a sound decoder we often must fit a loose speaker into a way-too-small sealed enclosure. Bass response always suffers. Standard speakers want to be in enclosures far larger than we can fit in our models.

A small Taiwanese company, Tang Band, has been producing a line of very high quality, small multimedia speakers, including some already in small sealed enclosures with passive radiators. In each the passive radiator, essentially a weighted speaker cone without a voice coil, is tuned to resonate at a frequency lower than the active driver. Set up this way, it extends bass response by about an octave.

The 1925S and 2008S modules are more expensive than the typical model railroad speakers, but at the moment they are the best sounding I’ve installed, putting out much stronger bass and cleaner sound. The 2008S will distort at high volume, but is still plenty loud. It is rated at 1/2 watt, and the Loksound V4L decoder has a 3 watt amplifier. With this pair you do want to program the volume considerably lower than factory default.

Here’s a recycled photo of the two modules side-by-side. The larger one easily fits in Alco and EMD hood units such as an RS11 or GP30. The smaller speaker is not as tall and fits inside Atlas SWs, under the radiator cutout.

I overlaid the data sheet frequency response curves of both modules for comparison purposes. The larger module has an extended bass response below 200 Hz.

A good friend has speaker characterization equipment and measured a 1925 module. He confirmed the surprisingly-strong low bass response.

These speakers are also available without an enclosure. Unfortunately, the passive radiators are not.

Do note that these are 4 ohm units. Some brands of sound decoders require 8 ohm speakers.

Sound samples can be heard in the Loksound Full Throttle thread.


I just received permission from friend Larry Hanlon to share some of his recent TB speaker measurements. To look at this graph you might want to click on it to enlarge.

These measurements were made in a room, not an anechoic chamber, so they have some extra ripples especially in the midrange.

Green: TB 1925S speaker module shown in the previous post.
Purple: TB 1925 speaker mounted in a commercial 28mm x 28mm x 12mm commercial enclosure, filled about 25% with acoustic fill. It does remarkably well, losing only about 1/2 octave of bass.
Red: Larger TB 1931 speaker in a commercial 31mm x 17mm deep enclosure, filled about 25% with acoustic fill.
Aqua: TB 1815S speaker mounted in a 31mm x 17mm deep commercial “high bass” enclosure. This speaker is very similar to the 1931 except that the cone is concave instead of convex.
Yellow: The star of the set, TB 1931SB speaker module. This unit is too large to fit inside O-scale hood units, but can be shoehorned into many steam tenders. It is also too thick to fit in a P&D F-unit. Unfortunately, it is no longer available in quantities less than 200 pairs!

If you contact Parts Express, please lobby them to keep the 2008, 1925 and 1931 modules and speakers in open stock!

Another module I’m looking into is the 1828SD. This one has a cylindrical enclosure and a passive radiator on the bottom of the cylinder, if we mount it facing up in a locomotive. It is too tall to fit in an unmodified P&D F-unit, firing up into the dynamic brake fan exhaust. I would have to cut a 1.75" diameter hole in the brass locomotive frame and mount the passive radiator end below the frame, firing down to the track. It would require making a new motor mount. And it is quite expensive. But I’m listening to one right now, sitting on my mouse pad, driven by a V4 decoder and it sounds awesome. What to do?

Here’s the rated 1828SD frequency response. Note that the left edge of the graph starts at 20 Hz. In the chart shown above in my previous post the graph starts at 100 Hz. The 1828 is down about 7.5 dB at 100 Hz, whereas the 1925S is down 20 dB at 100 Hz, and down about 7.5 dB at 250 Hz, almost 1 1/2 octaves deeper bass.


These speakers do a fantastic job, and with F0 (minimum frequency response) in the 75-80Hz range, they are well worthy of installing even in the lowest-fidelity sound systems. Personally, just swapping out the speakers, have seen a 2-decade drop in the low end of the sound response.

Here’s the factory speaker:

And the Tang Band speaker:

Sorry, there is no sound of the speakers. Look for other videos that do show the difference.


CentralFan -

Checking the Parts Express web site today (Nov 1 2016) shows that they will be stocking the 2008S, 1925S and the 1931S speaker modules in single quantities instead of pairs. That’s great news for model railroaders who work in larger scales, and you will be able to get more of the larger '31s with their extended bass response for your speaker upgrades.


I saw this! And very excited…


  • Mario


Mario - The 1931S pairs are showing back in stock on 13 Dec 2016. Parts Express are still waiting on single units.


FYI single units of T1 1925S are now showing in stock at Parts Express. 4 or more for a discount. :slight_smile:


Woo hoo! Thanks, Pete. I’ll be ordering more.


You’re welcome! It’s thanks to you that I found them. But I came in after the gold rush in the fall of last year. I think I got the last 4 at Parts Express and then found 2 more at Solen in Quebec (I’m in Ontario) and that was it for almost 4 months.


In the thread Car & Locomotive Shop RS11 Sound Install Frolin asked whether the Tang Band 1925S module was that much better than a Railmaster bass reflex speaker. Earlier in this thread there is a measurement by Larry Hanlon showing the 1925S module.

Larry recently sent his measurements of the Railmaster DSM-8 18x35 mm and DLG-8 20x40 mm modules, along with one of the new “sugar cube” or cell phone speakers, a CUI CDS-18138A. If this or the previous plot appears difficult to read, click on it to enlarge.

Larry’s notes:

Railmaster DSM-8 and DLG-8 speaker modules. Neither module has any low end. Red: DLG-8 20x40mm speaker “bass reflex” module; Cyan: DSM-8 18x35mm speaker “bass reflex” module; Green: CUI CDS-18138A micro speaker in “minimum” 2.8mL enclosure. This enclosure is .030” styrene, .5” deep, built right around the speaker which forms the front face, and is about 50% filled with polyfiber. The moving plate of the micro speaker is facing out, and the enclosure is otherwise completely sealed.

These response curves were measured with 260mW driving the speaker, and the calibrated measurement microphone spaced 2.5” from the center of the modules in order to “see” the “bass reflex” port. Spacing from the CUI micro speaker was 1cm, which is why its output appears greater.

In my opinion this graph explains why HO and N scale enthusiasts are quickly switching to the new technology cell phone speakers.

That said, the Tang Band 1928S module puts out a lot more bass for O-scale than the sugar cube, usefully extending down to about 200 Hz, and can handle considerably more power.



I am going to give these speakers a try in an upcoming project. Following your lead sir. Thank you for the data and research.



Erik -

If your secret project is steam, you might look into a 1931 module fitted in the tender, with the floor opened up with a lot more holes.

This module really rocks in the bass department. Useful bass extends down to about 120 Hz. The Loksound L series decoders have two speaker outputs, so it might be possible to fit a second, smaller speaker in the smokebox at the cost of two more wires between the engine and tender.



You guessed it, I’m a sucker for steam. I want to start with my L-105 it’s large and would really benefit from Loksound and your installation. I am planning on giving it a try!

Thank you for sharing this valuable information with the community.

Eternally grateful



I should add I will most likely be picking your mind on this as I go along.


A good external microphone arrived today for my iPhone, a Shure MV88 stereo unit. I had to give it a try.

This is a near-field recording, made about 3 inches away from a Tang Band 1925S module. The sound file is ESU’s new 2000 HP 12-cylinder Alco 251 that I wanted to audition for an RS-32.


The A&O is “O-scale in your face” so at times your ear can get up-close and personal with the equipment. I can’t wait to railfan the RS-32! And hopefully this recording will give you an idea of what can be accomplished in O-scale.



Thanks for the lesson on Tang Band speakers, especially by providing the FR charts. Really nice that those are self-contained and produce much lower bass than stock speakers.

Had you guys ever previously used QSI speakers, perhaps the 1.77" version and done any FR graphs with those in the past that may be compared to the TB speakers? I am looking at replacing the speakers in my 3rd Rail SD9’s. Maybe even try out a Loksound decoder in one of these.

Some thoughts on speakers…

the QSI (TDS) 1.77" speaker would fit exactly where the original 45mm speaker was (facing down at the truck under the cab) and use the same cavity in the hood as its chamber. It is however 8ohms, not 4. (http://www.traintekllc.com/tds-high-bass-speaker-1-77-diameter/) Since its 8ohms, I would get some other 8ohm speaker (maybe a railmaster bassreflex speaker of some sort?) and have it point up the fans too. Perhaps the combined sound of those would perhaps be better than a tangband speaker? There is also this nice looking speaker on ebay with no datasheet information behind it… http://www.ebay.com/itm/2pcs-1-75-inch-4Ohm-4-3W-Neodymium-Full-range-Audio-Speaker-Loudspeaker-45MM-/232041246454?hash=item3606bc6ef6:g:gYsAAOSwU-pXpzHR

The other option would be to try one of these tang band 1925S speakers and have it point up and out of the fans in the top of the long hood. The F0 of this speaker is actually higher than the TDS speaker and the power is the same, although it can get louder since its 4ohms.

I am worried that the sounds of the 1925S would be muffled quite a bit by trying to go through the small fan openings in the roof, but since the frame of the locomotive is already milled out for a speaker, perhaps thats the better option? Any opinions? Or maybe its worth trying it both ways and see which one is better…which is what I am trying to avoid.


Dave -

We have no frequency response charts on the 1.77", but I can offer anecdotal evidence. In my ABA set of F3s, one has a 1.77" QSI (now TDS from Tony’s Train Exchange) and the other two have the no-longer-available 2.07". Subjectively the 1.77 has a poorer bass response than the 2.07 during a run-by. The B unit has a much bigger enclosure and that helps bass response quite a bit, but it still isn’t as deep as the 1925S module. Here’s the only (lousy) photo I have of the 1.77" during its initial install 7 years ago.

Here’s the 2.07"going in the B unit.

The TB 1925S module has the same rated resonant frequency, 150 Hz, as the 1.77" in free air. The impedance chart shows two resonant peaks, one for the speaker and a much lower one caused by the heavily-weighted passive radiator. Once I installed the 1.77 in the biggest enclosure I could, f0 went way up. Comparing a speaker in free-air against another in a sealed enclosure is apples to oranges.

Although I didn’t measure f0 with the 1.77, I did measure the 2.07 once I put it in an enclosure. QSI stated that the 2.07 has a free-air f0 of 150 Hz. For the sake of illustration let’s assume that is correct. After putting it in a sealed enclosure for the F3A, roughly 2" x 2" x 1.5" deep and stuffed with fiberglass batting, f0 went up more than an octave when driven by a 50 ohm function generator and measured with a digital oscilloscope. If I recall it was now somewhere around 350 Hz, but that memory could be faulty.

I have 6 more F3 kits to build, with TB 1828SD modules waiting for them (f0 = 75 Hz.) This ABA set will probably be left as-is, because I have so many more engines of all kinds to prepare for operating sessions.

As for the 1.77s, if they are an easy drop-in, that might be a compelling reason to try them. Switching to a 1925 would be a no-brainer if the SDs were produced with brass screened instead of solid plastic radiator intake grills. Locating the round driver directly under a pair of the fans might not be too bad. Bass will still find its way through the body shell and out the bottom, particularly through the now-empty speaker A-frames.

For a quick test of a 1925S, without a decoder transplant, you could put one in series with a 4 ohm resistor and drive it from Scott’s factory decoder. There won’t be as much damping of the cone and that will have some influence over the frequency response.

All the best, Dave. Stay in touch as you proceed.


Looks like I’d skip the QSI speaker, okay. Will probably try to see what the difference sounds like between 2 SD9’s with different speaker setups.

That 1828SD is a beast. Good luck shoehorning that guy into even an F unit. I’d probably stick with 1931S for ease.

And speaking of uneasy, have you guys experimented with removing width from the side of a 1931S to make them fit?

Looking at some measurements, looks like if I could take another ~4mm or so off the width of the 1931S, I could make that speaker fit in anything I have…Weaver VO1000, AtlasO GP35, AtlasO RS1, etc.

Does anyone recommend any tools for easily removing material from the side of an enclosure? I’m pretty sure a file ain’t gonna cut it, lol. The process would need to be easily replicated and consistent. I was thinking maybe a bench grinder or a standing band saw with very careful movements…thoughts? It’s a shame they aren’t just slightly narrower.


Oh and for those placing orders…$100 gets you free shipping…they also have a 10 off 100 that expires in 3 days with code S16EVT.

Good deal if you’re buying 7 1931’s I suppose.

Then it appears there will be a new code for 10 off 125 that expires in March. Code PCJQ17

Either are a good deal if you want 8 1925’s.


Dave -

The tool for that job would be a vertical mill. However it is highly unlikely that there is 4mm that could be removed without cutting off the sides completely. The module needs to be sealed.

The 1931 module is 36mm wide but the active driver is 31mm across its rim. I don’t know how wide the passive radiator is. Carefully shaving off 4mm would only leave at most 1/2mm of wall thickness on either side of the driver. That wouldn’t work. I suspect the SD9’s fuel tank is too short for a 1931 firing down.

As for the 1828SD, it can be fitted inside a P&D F unit if I mill a hole in the brass frame for the passive radiator. Some of the front motor mount will have to go along with some of the simulated frame rails on the underside. I might need to make new porthole glass so the speaker can clear. With adequate ground and truck clearance the module has some room beneath the roof to breathe.

Being a coal road, the A&O uses dynamic brakes so sound can exit the fan slots and along the side DB air intakes.

A little speaker cabinet damping material glued under the DB hatch may dampen some of the inevitable cavity resonance. In the second photo I was listening to that resonance. It can be “tuned” by holding the speaker at various distances from the roof. Knowles did something like this with a cell phone speaker firing against a padded wall only a few mm above the cone. Sound came out sideways from the module, through a slot.

The HO decoder is only for sound tests. The big Pittman motor would quickly let the magic blue smoke out.

Stick to the 1925s unless you think you can somehow make the SD9 carbody wider inside!