I’m curious about the performance of the Tang Band facing upward. Not in comparison to the QSI speaker, but just in general. Is it worthwhile to test it in that orientation, without something like the desktop to bounce the sound off of? The Tang Band install in my sw1500 will be upward facing, firing out of the radiator core, and I am inclined to think that most of my future installs will also be upward facing.
Here’s a plot of the 1931 module facing down and up. Generally I’m finding that downward-firing introduces an audible change in sound, with up-firing my preference. The yellow trace is firing up, magenta firing down.
This down firing measurement looks a bit different from the previous plot. It may have been held at a different height above the desktop.
Interesting, and good, if you think the upward facing speakers produce better sound. I don’t know that I have much choice in speaker orientation except in models with larger fuel tanks. So it may be a moot point. But it’s always better when the moot points go your way.
Looking at the graph, I have some questions. Remember that I don’t know too much about the technical aspect of sound yet. It looks like the bass rolls off more quickly with the upward facing speaker, below 75Hz? Is that relevant, especially at the 30-35 decibel level? I’m trying to get some context here. Do we even hear the bass at the lowest end of the spectrum at that volume? Can we discern the 5-7 decibel difference in volume? Or are we more concerned with the better vollume in the range from 75Hz through 1kHz?
Lastly, I’m curious what constitutes an “audible change in sound”? Is it volume at a given frequency, or does it relate to factors like distortion or clarity?
By the time we get 12 dB down relative to the response at about 1KHz, the bass is pretty much gone. I wouldn’t be too concerned about differences below 70 Hz. Except…
By an audible change, I’m not implying a change in clarity or distortion. It is the relative emphasis of certain frequency ranges over others. It is a bit like holding an empty glass a few inches from your ear. You can hear the air cavity in the glass emphasize certain notes over others.
Don’t get too wrapped up in peaks and dips. They were far worse before I applied 1/3 octave smoothing. A lot of them come from sound bouncing off surfaces in the “room”, which means the physical room and everything close to the speaker, such as an old scanner, ALPS printer, computer monitor, myself, you get the idea. They are walls in a sound measurement cavern, not a proper anechoic chamber (spooky if you’ve ever been inside one, as I have.)
What’s measured does change quite a bit when the microphone or speaker is moved. Speaker manufacturers often make their measurements near-field, with the microphone fairly close to a single speaker driver mounted in an “infinite baffle” (a large flat non-resonant wall.) In my earlier measurements the capsule was aimed at a spot between the speaker and passive radiator from about 1" away.
Bass response of the 1931 module definitely increases when I move the mic above the passive. That begs the question of which element should be directly under the radiator grill, the passive or the speaker? The passive?
To see why I wonder, here are 3 measurements. Aqua shows the mic on a tripod about 1 inch above the speaker. Green is 1 inch above the passive. Finally, magenta is a foot away from the speaker, above it at a 45 degree angle intended to simulate close trackside rail fanning.
I turned off smoothing so you see the raw 1/12 octave data points. Notice how difficult is it to now estimate the overall trends?
And finally, here’s the same data set with 1/3 octave smoothing. It’s a bit easier to see the forest instead of individual trees.
There’s great news in all this. Today’s best speakers for HO and smaller scales are the small rectangular cell-phone types. The best of them, when mounted in a practically small enclosure, have a resonant peak around 1 KHz then by 500 Hz rapidly roll off.
So when you get your locomotive finished, show off the sound to some of your HO and N scale buddies. Now don’t get me wrong, I enjoy all model scales especially 7 1/2" gauge. Just bring a rag to wipe the drool off the floor… Then ask them to build an O-scale engine that they can operate on your layout!
Well, that’s even more interesting. I guess I should consider turning the speaker around on my sw1500.
I already have plans to wow my HO-centric local modeling associates once I get the decoder and speaker wired up. The layout I currently run on has several sound equipped locos, which I enjoy running more than I thought I would. More so the steamers, as I think that HO sound is more successful for them versus diesels. I’m planning to take my chassis and a test track over there and jumper to the rails for initial testing. I don’t expect anybody to change scales, but I would like for them to have second thoughts.
By the way, since you mention the smaller speakers, I had a nice chat with Bryan Vianco two weeks ago at the Indy 2-rail show. We talked about the Tang Band modules, and a lot of other stuff too. He’s a big fan of the sugar cube speakers, but then he also specializes in N scale dcc installs, so it makes sense.
Thanks for all the effort testing different scenarios. All this info should lead to some convincing sound performance.
Turning around the speaker in an SW so the passive gets more grill exposure isn’t intuitive but it might just work. For a quick and not all that scientific test, I set an Atlas O SW8/9 shell on top of the module, which sat on spacers to better approximate the final mounting height. Since I haven’t yet milled off the posts sticking down from the roof, the speaker had to sit a bit low. I also plan to mill out most of the black painted area for a better appearance through new etched grills. Note that this shell is from an early Atlas production run. Later runs relocated the speaker to the fuel tank and improved attachment strength of truck side frames.
First photo: Measuring the speaker itself.
Second photo: Measuring the passive.
Overall the setup was about 4 dB louder with the speaker firing out the hole, so in the measurement software I shifted the orange passive radiator response curve up by 4 dB for an easier comparison with the yellow speaker curve.
The extra 5 to 9 dB of bass below 100 Hz might be most welcome. My objective is better bass at a quieter overall volume. We will need a pair of SWs to switch the north end of Havens Yard because the switch lead runs downhill as it ducks out of sight. They should sound pretty nice shoving a long cut of cars.
Must build faster!
I’m going to turn my speaker around. The added benefit is that it puts the speaker connections at the end closer to the decoder. Less wire .
Very interesting results. I have all my Tang Band speakers mounted facing up, with as much air space and opened grilles, vents, fans and exhausts as possible above them. I found out by accident that they sound better and seem to my ears to have more bass. This proves it!
Another advantage of this setup is that I can usually get the driver under or somewhat close to the bell which seems to make the bell more convincing. I think it’s because the bell sound has the most directional treble content. The placing of the passive radiator end of the TB enclosure doesn’t seem to matter much as the sound from that end isn’t so directional. It just needs to have the most open areas above it to let the bass out.
Also I have used thin speaker cloth to cover any big holes under grilles and fans to hide the speaker and it doesn’t seem to impede the sound.
Thanks for all the great testing, design and engineering ideas that you guys come up with. I’m continually learning from you!
As we contemplate turning the speakers around in our SWs, let’s verify that there is sufficient vertical clearance above the speaker. That’s the end that has the greatest cone excursion and stands taller due to the domed shape. There’s enough bass boost in many of the ESU sound files to make that puppy really move. The frequency response will change with a shorter distance between the module and roof.
Thanks for your report! Your experience confirms that we may be on to something useful.
Could a similar improvement in bass response occur with a smaller 1925S module installed in a Car & Locomotive Shop Alco? In an RS-36 I might have accidentally located the passive directly under the fan grill. I say might because the photo was taken before the install was finished.
Anyway, there’s a similar C&LS RS-11 sitting here that I can test with realRTA.
Overland C424/425/430 models have open air intake louvers down low, on either side of the long hood, and like the prototype the slots face down. The speaker wants to fire down from a location near the closed rooftop radiator exhaust shutters. Most sound exits through the louvers, some around the truck bolster, and wee bits through small screens on the sides and rooftop dynamic brake air exhausts.
Good points Bob. As I’ve got more into sound quality after learning of Tang Band from you, I find I’m putting speaker location for best sound performance right up there with smooth starting and slow running as priority 1. So now I’m looking more at the fuel tank to house the decoder and keep-alive, leaving as much as possible of the space inside the loco hood for acoustic optimization. With brass locos there’s usually plenty of air outlets, with Atlas and other plastic locos I’m opening up fans and grilles where practical.
Just did a quick measurement of the 1925S module. In the following plot, which was 1/3 octave smoothed, all measurements were made from 1" away with speaker held in a vise.
Green: aim the mic between the round speaker and passive radiator.
Magenta: aim the mic at the center of the speaker.
Yellow: aim the mic at the center of the passive radiator.
The yellow curve was shifted up 3 dB to overlay it with the other two curves. A “speed tradeoff” in the software was set to a slower rate, 20 Hz, and that produced smoother curves than the earlier 1931 measurements.
So it appears that installing it with the passive radiator directly under a screen or grill would offer slightly less volume but better bass response, about 3 dB at 200 Hz and 7 dB at 100 Hz. As always, things will change after installing in a locomotive.
I thought I should show you my latest effort inspired by the great work I see on this forum! Basically a Tang Band 1942s with a motorized truck mounted on each end! Actually it’s a P&D F7B with drive upgrade kit from Jay Criswell at Right-O’-Way.
I don’t run it quite this loud for ops sessions, but it does get people’s attention. Turn up the video volume and enjoy!
I’m envious of both the sound install and the drive that Jay built for you. I wish I could afford only Jay’s drives in my O-scale locomotives, much less one! Both are top quality.
That’s some great rumble machine. What is your thinking to mate that awesome sound quality with a narrow hood diesel?
How do you like the Photo Throttle? I’d love to run a train with one, but it seems about 2x too big in every dimension, especially the depth. BTW I’ve run a GP9 and an SD9e, the latter with a train. Not to brag, but I loved the throttle response of the old cylindrical air throttle stands. Especially with 14 cars in tow.
Glad you like it and thanks for leading me to Tang Band in the first place. So much fun!
Regarding drives, I decided on fewer locos but better running. So I’m culling the herd, mainly of Atlas China drives and putting the proceeds into Jay’s drives. I sell off the Pittman motors and Weaver/P&D drives (some Finescale360 chassis and towers as well) plus I save about $50 per decoder as I can use HO. I’ve dropped the power consumption of my layout overall by over 75%, so selling off 2 x 8A boosters as well. Ends up not far out of pocket.
The narrow hood locos are still a challenge speaker-wise. The TB 1931 just fits in the C&LS RS11/36, and the trick seems to be setting it as low down as possible and firing up, to get as much free air space and and vents in the shell above it, just as you found in your recent testing.
As you know, the 1931 doesn’t quite fit in RC GP9 hoods which is a shame. Perhaps it could be machined narrower, but beyond my capabilities. If we had budget, maybe we could get TB to make us a longer, narrower 1931, perhaps with 2 x passive radiators and the longest-throw/highest efficiency driver that could fit.
Or… I did like the look of where you were going with the custom enclosure idea, and with the passive radiator you don’t need a lot of air volume inside the enclosure because you want max air pressure to drive the radiator. Maybe you could devise an enclosure for the 1931 driver and radiator that does fit in a narrow hood?
Otherwise I’m limited to TB 1925s mounted as low as I can, firing up through as many open vents, fans and exhausts as available.
As for ProtoThrottle, I have to say is it’s the most fun and engaging item I’ve ever added to my layout. It makes the engineer’s role and experience as realistic as everything else I’m trying to achieve operations-wise. It is a bit larger than most throttles but I find it easy to get used to. My hands are about 1.25:1 scale mnd you. But everyone who operates here seems to love it so far. I’ve had an operator who was an engineer tell me it’s the most realistic experience he’d ever had operating on a layout.
I highly recommend getting some hands on with it if you can.
That’s neat you drove the real thing. I would love to try that. The closest I ever got was driving Flying Scotsman at Tyseley back in the early '90s. Nothing at all like an SD9!
A couple of pics of the PT in hand:
Glad to hear you’re having fun with the PT. There are mixed reviews at a large local layout, mostly centering around the bulk of the device. Driving the Scotsman must have been a real hoot, man!
As for selling off boosters, one important question to ask is “If I eliminate this booster, what will that do to the length of the DCC bus to the track?” The extra wire inductance can be an important consideration.
I recently did temporary installs of 1931 modules in a pair of Alco FA-1s. They sound great, currently a bit too loud. The rectangular 1828s almost fit. I had to fuss with one of the gear towers that wanted to lock up and bind. Ugh. Jay’s drives would have been most welcome and I fully appreciate his disdain for that drive. But there is an FB-1 waiting to join the consist, so I would be in for 3 conversions, and that would be the tip of the iceberg. The A&O has been running out of motive power.
Thanks for the booster point Bob. I have my booster power districts separated through existing PSXs, just more PSXs per booster now. I only have one buss run over 30 ft’ and I added a NCE snubber to the far end of that one. So far so good… :
Yeah, how come Tang Band doesn’t take the width of O scale cowl- and hood- unit Diesels into account when they design their speaker enclosures - don’t they realize the huge market potential?
Those chain drives can be frustrating for sure. The best I got was to steal all your great tuning ideas, and add the FineScale360 tower kits. That at least gets you a top shaft that’s parallel to the bottom shaft and has ball bearings, plus adjustable shaft centres.
Next up I was thinking about swapping the chains and sprockets for pulleys and belts but then I saw Jay’s concept of a gearhead motor. Having the entire drive-train after the motor run 1:1 with the wheel revs eliminates a lot of noise and vibration. Fortunately for my wallet, my roster is probably 1/10th of the A&O if that!
I just measured the system frequency response of a Tang Band 1925S module installed inside my C425. Sound exits through the radiator intake vents at the end of the long hood, just below the number 475… Here’s the measurement setup. That’s the measurement microphone in the right of the photo.
Next we have the frequency response. It was measured using pink noise and TrueRTA software. The locomotive’s speaker was driven by a Lepai class D amplifier. The original measurement was an average of 10 1/12 octave measurements, then smoothed to 1/3 octave. Bass response is down about 14 dB at 100 Hz relative to 500 Hz.
ESU puts some fixed bass boost in many of their sound files. If we drive the speaker from LokProgrammer and progressively increase the throttle notch, we can measure SPL or Sound Pressure Level plots that give us an idea of what frequencies we will actually hear. First up, a plot from idle = green, run 1 = magenta, run 2 = yellow, run 3 = orange, and run4 = cyan.
Finally run 4 = cyan, run 5 = green, run 6 = magenta, run 7 = yellow, run 8 = orange.
So although the system frequency response is down 14 dB from 500 Hz to 100 Hz, we are still getting significant bass down to 100 Hz. That’s a bit better than Larry’s free-air measurement would predict. In this case, I installed the speaker so that the passive radiator was directly above the radiator louvers.
Of course your mileage may vary.