Scale Trains Tier 4 Sound -- Speaker Improvement

[jdcolombo]

I absolutely agree with your statement.  Sound "too real" for a model is a very good description.  Like you, I also would not expect or want a 5-inch-long scale model of a locomotive to have a deep-bass Diesel rumble sound which you can feel in your gut.  Just like the model which is scaled down, I feel that its sound needs to be scaled down too.

Ditto Pete.

We set up our N Trak layout at local train shows, and sometimes there will be an HO-scale layout nearby that has full-sized stereo speakers and subwoofers for sound reinforcement.  It's crazy - it is not a "natural" sound at all, even with the larger HO-scale models.  Instead, the incessant bass drives us all nuts.

Most of us end up 3-5 feet away from our N scale trains when operating them.  3' in N is the equivalent of 480 feet or about 1.5 football fields in length.  It is true that even at this distance, you will get a bit of "rumble" from a prototype diesel going by, but not a lot, and the predominant sound will be the lower-midrange "thrum" of the prime mover.  The horn, however, you can hear well from over a mile away.  On our models, the horn should be three or four times as loud as the prime mover; the bell should be a bit louder than the prime mover, but not nearly as loud as the horn.  Spitter valves and the air pumps would be somewhere around bell volume.  As much as I admire what ESU (e.g., Matt Hermann) has done with their US Prototype sound files, none of them come with the sounds properly balanced IMHO.  The prime mover and bell are always way too loud; so are the air pumps and spitters.  Once you balance the sound correctly, a good N-scale sound install will actually sound pretty realistic if you account for how far away you actually are in prototype terms.

John C.

[wmcbride]

Ditto Pete.

We set up our N Trak layout at local train shows, and sometimes there will be an HO-scale layout nearby that has full-sized stereo speakers and subwoofers for sound reinforcement.  It's crazy - it is not a "natural" sound at all, even with the larger HO-scale models.  Instead, the incessant bass drives us all nuts.

Most of us end up 3-5 feet away from our N scale trains when operating them.  3' in N is the equivalent of 480 feet or about 1.5 football fields in length.  It is true that even at this distance, you will get a bit of "rumble" from a prototype diesel going by, but not a lot, and the predominant sound will be the lower-midrange "thrum" of the prime mover.  The horn, however, you can hear well from over a mile away.  On our models, the horn should be three or four times as loud as the prime mover; the bell should be a bit louder than the prime mover, but not nearly as loud as the horn.  Spitter valves and the air pumps would be somewhere around bell volume.  As much as I admire what ESU (e.g., Matt Hermann) has done with their US Prototype sound files, none of them come with the sounds properly balanced IMHO.  The prime mover and bell are always way too loud; so are the air pumps and spitters.  Once you balance the sound correctly, a good N-scale sound install will actually sound pretty realistic if you account for how far away you actually are in prototype terms.

John C.

Exactly. The ESU spitters and air pumps are way to loud but a good reset, as John says, makes it very realistic. The most realistic of any scale IMHO.

[ek2000]

Ditto Pete.

We set up our N Trak layout at local train shows, and sometimes there will be an HO-scale layout nearby that has full-sized stereo speakers and subwoofers for sound reinforcement.  It's crazy - it is not a "natural" sound at all, even with the larger HO-scale models.  Instead, the incessant bass drives us all nuts.

Most of us end up 3-5 feet away from our N scale trains when operating them.  3' in N is the equivalent of 480 feet or about 1.5 football fields in length.  It is true that even at this distance, you will get a bit of "rumble" from a prototype diesel going by, but not a lot, and the predominant sound will be the lower-midrange "thrum" of the prime mover.  The horn, however, you can hear well from over a mile away.  On our models, the horn should be three or four times as loud as the prime mover; the bell should be a bit louder than the prime mover, but not nearly as loud as the horn.  Spitter valves and the air pumps would be somewhere around bell volume.  As much as I admire what ESU (e.g., Matt Hermann) has done with their US Prototype sound files, none of them come with the sounds properly balanced IMHO.  The prime mover and bell are always way too loud; so are the air pumps and spitters.  Once you balance the sound correctly, a good N-scale sound install will actually sound pretty realistic if you account for how far away you actually are in prototype terms.

John C.

John,

Is there a measurement or quantification of what scale sound is for N? For e.g. if a locomotive prime mover measures at 110-120 DB at 3 feet away, what would that translate to scale level sound for N? I ask because we are going to be 1 feet if not more from the layout at most times which could translate to 160+ feet in real life so the sound DB level is going to get lower and lower quickly to the point that it may not be audible at all.

I do understand the point about sub woofers being too loud and unnatural and the comparisons you have laid out between the relative sound levels of various sounds makes sense (bell vs. horn vs. prime mover etc.) , but I'm unsure if scale comparisons and expectations with respect to size translate the same way to sound or if there is a different measurement of what scale means in terms of sound.

[peteski]

John,

Is there a measurement or quantification of what scale sound is for N? For e.g. if a locomotive prime mover measures at 110-120 DB at 3 feet away, what would that translate to scale level sound for N? I ask because we are going to be 1 feet if not more from the layout at most times which could translate to 160+ feet in real life so the sound DB level is going to get lower and lower quickly to the point that it may not be audible at all.

I do understand the point about sub woofers being too loud and unnatural and the comparisons you have laid out between the relative sound levels of various sounds makes sense (bell vs. horn vs. prime mover etc.) , but I'm unsure if scale comparisons and expectations with respect to size translate the same way to sound or if there is a different measurement of what scale means in terms of sound.

Even if we could, we probably wouldn't proportionally scale all the sounds. In N scale modeling we don't evenly reduce everything to exact 160th of the real thing. We often use selective compression, or on the opposite end, exaggerate certain details to make them visible.  I think the same will apply to sound levels. Plus, it is often a matter of personal taste.

However I do think that it would be interesting to find out the actual sound level of various sounds emitted by a full-size locomotive.

[GaryHinshaw]

John,

Is there a measurement or quantification of what scale sound is for N? For e.g. if a locomotive prime mover measures at 110-120 DB at 3 feet away, what would that translate to scale level sound for N?

110-120 dB is painfully loud, like a thunderclap or chainsaw at close range.  I don't think a prime mover is that loud even at 3 feet.  This table gives several examples of noise levels, and it lists "diesel train at 45 mph at 100 ft (83 dB)", similar to a garbage disposal.   Since our ears are not scale models, we could argue that the sound volume should reproduce actual volume at a corresponding scale distance.  Let's see how that works out.

If we view a model from 2 ft away, that is 320 scale feet, or 3.2 times farther than the 100 ft chosen in the example.  Sound intensity falls off like 1/distance^2, which in this case is 1/(3.2^2) ~ 1/10 the volume.  Since dB is a logarithmic scale, 1/10 the volume is 10 dB quieter, so our model should be ~73 dB, about the level of a typical vacuum cleaner.  Who wants a bunch of vacuum cleaners running in the layout room during a session?

Here's the rub: suppose we chose to view the model from across the layout room instead, say a distance of 10 ft away?  This is now 1600 scale feet, so the intensity should be 16^2 = 256 times lower (24 dB lower) than the real train at 100 ft.  This is now roughly 50 dB, comparable to a human voice, and much more tolerable.

Unfortunately, if you set your model volume to be 73 dB at 2 feet, it will only be 5^2=25 times quieter (14 dB lower, so about 60 dB) at 10 ft; much louder than a real train would be at 1600 ft.  The problem is that sound volume falls off like 1/(actual distance)^2, not 1/(scale distance)^2, so you just have to pick a fiducial distance that you are happy with and set the volume accordingly.  But there is no way around the fact that the volume will not scale properly with your scale distance from the model.

[ek2000]

Gary, In your calculation leading to 73 db, you start off with the baseline sound as 83 db and then adjust for distance. I think that the baseline sound for a model as small N should first be set at a certain db level upon which subsequent distance based calculations should be made.

[jdcolombo]

Thanks for the math, Gary!

Yeah, you can't model the fall-off in sound from one scale distance to another - well, at least not unless we invented a variable-aperture earplug that would open and close according to an algorithm that depended on how many actual inches you were from the sound source. 

Just for kicks, I measured the sound pressure level of one of my three-unit sets of GP9's.  The front and rear units both have sound (ESU LokSound with an 8x12mm speaker in a custom enclosure).  From the normal spot at which I would observe and operate the locos (about 2.5' away from the engines), the prime movers hit 65db.  The horn hits 76 db.  According to sources I've read, a 10-db difference in sound pressure level equates roughly to "twice as loud" (this isn't scientific; "twice as loud" depends on the individual's psychoacoustic perceptions, so it's just a rough estimate) - so the horn would be perceived by most listeners as roughly twice as loud as the prime mover.  When the air pumps come on, the SPL is 66db; the bell hits 68db.

All of this sounds about right to me, in a layout room that has acoustic tile for a ceiling and is fully carpeted - in other words, a fairly "dead" acoustic environment.  When I'm doing an op session on the layout, we'll have maybe four of those sets of diesels running at once on different parts of the layout, along with some single engines doing switching at the yard and on local trains.  My operators think the overall sound level is fine.  But I probably should turn the prime mover sound down by another 10db relative to the horn.  I might try that . . . about 55db for the prime mover alone; 58db for the bell, and 76db for the horn.  It would result in a bit less cacophony during sessions!

YMMV.

John C.

[jdcolombo]

BTW, the FRA has a rule that requires a locomotive horn to have a minimum SPL of 96db from 100 feet in front of the locomotive in its direction of travel.  So if the prime mover sound from 100' away on a train passing at 45mph is 83db, then the horn would be a minimum of 13db louder, or somewhat more than twice as loud.

So maybe I've got the ratios between prime mover and horn roughly correct after all . . .

John C.

[GaryHinshaw]

Gary, In your calculation leading to 73 db, you start off with the baseline sound as 83 db and then adjust for distance. I think that the baseline sound for a model as small N should first be set at a certain db level upon which subsequent distance based calculations should be made.

This is kind of my point.  You can pick a fiducial scale distance to set your volume levels, but regardless it will not augment or diminish properly as your model get closer or farther away.  It will be too quiet at closer scale distances and too loud at larger scale distances because the sound intensity you hear depends on actual distance not scale distance (unless you get a pair of John's fancy ear plugs).

...From the normal spot at which I would observe and operate the locos (about 2.5' away from the engines), the prime movers hit 65db.  The horn hits 76 db.  According to sources I've read, a 10-db difference in sound pressure level equates roughly to "twice as loud" (this isn't scientific; "twice as loud" depends on the individual's psychoacoustic perceptions, so it's just a rough estimate).

Neat.  I would say you have the horn to prime mover ratios about right, but I would be tempted to turn it all down by ~5 dB or so.  Definitely personal preference though, and I can imagine wanting different overall levels for solo running vs. ops session running.

BTW, in terms of the physical power propagated by a sound wave, a 10 dB increase in level corresponds to a 10-fold increase in power.  A factor of 2 increase in power is a +3 dB change in level.  But our perception of loudness is better represented by the dB scale since our ears (and eyes) respond logarithmically.  This gives us pretty good dynamic range in our ability to see and hear.

[jagged ben]

110-120 dB is painfully loud, like a thunderclap or chainsaw at close range.  I don't think a prime mover is that loud even at 3 feet.  This table gives several examples of noise levels, and it lists "diesel train at 45 mph at 100 ft (83 dB)", similar to a garbage disposal.   Since our ears are not scale models, we could argue that the sound volume should reproduce actual volume at a corresponding scale distance.  Let's see how that works out.

If we view a model from 2 ft away, that is 320 scale feet, or 3.2 times farther than the 100 ft chosen in the example.  Sound intensity falls off like 1/distance^2, which in this case is 1/(3.2^2) ~ 1/10 the volume.  Since dB is a logarithmic scale, 1/10 the volume is 10 dB quieter, so our model should be ~73 dB, about the level of a typical vacuum cleaner.  Who wants a bunch of vacuum cleaners running in the layout room during a session?

Wary as I am of calling out the physics professor on his math, there's a contradiction in there.  If a train is 83 dB at 100ft, and sound roughly falls off according to the inverse square, then at 3ft the train should be approximately 113 db, which is exactly within the range of 110-120dB.   (Not sure if the 83dB is merely the prime mover, and doesn't include wheel noise and other sources, but point still stands.) 

Prime mover noise varies of course, but I can tell you that having heard multi-unit lashups climb Tehachapi at notch 8 from a hundred feet away, I definitely did not want to be on the platforms of those locomotives without ear protection. 

[GaryHinshaw]

That's a fair point.  But there are finite size effects to consider at those distances: the prime mover itself is larger than 3ft across, so you don't get the full effect compared to being 3 ft away from a point source emitting the same overall sound power.   (I assume we're talking about being 3 ft from the closest part of the prime mover.)

But the main point is: there is no correct answer to how loud a model should be.

[ek2000]

At the end of the day, this kind of falls into Peteski's line of reasoning of selective exaggeration or compression to suit the scale. There is a standard (NMRA?) for weight for N and HO and I was wondering if there was a similar standard for sound (or should there need to be?). Thank you all for your enlightening inputs.

[jagged ben]

That's a fair point.  But there are finite size effects to consider at those distances: the prime mover itself is larger than 3ft across, so you don't get the full effect compared to being 3 ft away from a point source emitting the same overall sound power.   (I assume we're talking about being 3 ft from the closest part of the prime mover.)

But the main point is: there is no correct answer to how loud a model should be.

It's correct that we can't resolve differences of opinion with mathematics in this case.  (I just interrogated myself with some math in a somewhat philosophical attempt to prove otherwise, and defeated myself.    )   

But that won't stop me from sharing my opinion!     

Honestly, I had to look up 'fiducial' (spell check doesn't know it!), but now that I've done so I'll offer a rather strong opinion that the fiducial distance should be several scale miles.   Why?  Because what is going to sound 'realistic' in a model setting is that once you get an appreciable distance from the model, you don't hear it anymore, or else it at least fades towards the level of background ambient noise and is possible to ignore.    Put it this way; if you can hear your train from far away then you can't experience the real-world thrill of hearing it get louder as it gets closer.  It a club setting, where you might actually have some scale miles to work with, it means that you hear the trains that are near you on the layout, not the ones that are at the other end; this is a good thing. 

So let's look at your table and do the math on a fiducial distance of, say, 3 scale miles, which we might as well round off to 100 real-world feet in N scale.   We'll say at that distance, any non-horn noise shouldn't be audible over background noise.   This depends on your background noise level of course, but let's take 35dB, for in between urban and rural ambient noise levels.  From 2.5ft away (40 times closer, 1600 times louder) we should be about 32 dB louder.  That's 67dB.

Why am I not surprised that John Colombo has his volumes set right in this range?   

I'll just add that the background noise does matter a lot.  I've struggled a bit at our club, which is open to the public on run days.  What seemed like good sound levels when programming without the public there was way too quite once people were in the aisles talking and walking about.  So my advice would be add 10dB if you're programming when it's quiet around but will be running when it's noisier.

Finally, this is all really just repeating what I've said elsewhere before, which is that factory volumes tend to come too loud, and that the horn should be much louder than other sounds instead of the same volume.  Seems like the decoder file compilers are just taking the sound effect files at unity gain and playing them back at unity gain, more or less.  Shouldn't require so much user work to make it seem more realistic, but I guess a few of us care more than most.

[GaryHinshaw]

Perfectly reasonable arguments jb.

I'll just add that the background noise does matter a lot.  I've struggled a bit at our club, which is open to the public on run days.  What seemed like good sound levels when programming without the public there was way too quite once people were in the aisles talking and walking about.  So my advice would be add 10dB if you're programming when it's quiet around but will be running when it's noisier.

This seems about right.  The typical noise level in a restaurant (or a club layout with guests) is around 60 dB.  In your example (67 dB at 2.5 ft), the sound falls to 61 dB at 5 ft, so you're already close to (club) ambient levels even relatively nearby.  As a rule of thumb, I think having the train sound drop below ambient from across the layout room is perfectly reasonable.

[cbroughton67]

The sound on the Tier 4 is a lot "less" compared to an FVM GEVO with a Loksound Micro and enclosed speaker. I'm going to see what I can do in terms of adding an enclosed speaker to replace the one installed by the factory.

@wmcbride Were you able to come up with a new enclosure or some improvement in the sound? I didn't see a follow-up post. I just picked up a pair of these, and the sound is disappointing.