Best Of *Science* of solving N scale diesel mechanism drivetrain vibration noise?

[peteski]

Peteski, makes sense what you are suggesting about quality control testing (or lack thereof) in model railroad loco mass-production.

So that then begs the question:

• What do the Kato or Atlas/Kader *design engineers* do, when they are designing or specifying a new or improved silent running mechanism for a new loco  (such as the Kato UP 4-8-4 or new Kato SDP40F.... or Atlas when they were designing the Alco S-2 mechanism)?
• I would hope they build physical manufacturing prototypes to test out their theoretical designs for silent running, prior to committing capital dollars to production?  And thus, I would hope they have tools to locate and debug in their manufacturing prototypes, any vibration or excessively noisy  issues they find?

I again think that you are over-thinking this. We are talking a toy train here (yes, that is what those things are), not some precision tool or a timekeeping instrument.  I would say most (probably all) of the design is done in CAD on the computer - this assures that things will fit correctly.  I highly doubt that noise concerns even enter the design equation. The design is done strictly for mechanical fit.   If the mechanism is built to higher tolerances and using fine-pitch gears with truck-mounted worm (like the new Katos) then it will be inherently quieter than the older designs.

But just like you can have good and fastidious car mechanics and ones who will fix your car but do a sloppy job (your car will be fixed either way), same applies to the model train design engineers (or to any profession, really).  If the design is done by a mediocre design engineer the final product will work but it won't be as good or precise as the same mechanism designed by a really good engineer.  Looking at Kato models (both shell and mechanism) you I clearly see that they are designed by really good design engineers.  Then comes the manufacturing part: As with the design stage, the tooling and the materials used can vary in quality (usually depending on the funds invested in them).  Expensively-made tools and better quality raw materials will result in a better model.

I don't have any inside knowledge about the manufacturing process but I suspect that most of the prototype testing is done on the shell (which is usually 3D-printed). I don't think the mechanical parts are prototyped at all - the CAD/CAM pretty much assures that the parts produced through that process will work well.

[atsf_arizona]

Peteski, good input, thx.

[atsf_arizona]

OK.  Here's the result of implementing the Ron Bearden  "Rapido spring in the universal" technique:   

===> Some noticeable improvement, moving the sound/noise needle towards 'acceptable' <====  (at least for now).   

Ron Bearden described this technique, applicable for universal-joint long wheelbase Atlas/Intermountain mechanisms, documented here: 

http://forum.atlasrr.com/forum/topic.asp?ARCHIVE=true&TOPIC_ID=43690  - Ron Bearden's post of 2008 January 13 :  3:23:31 PM

His words about the "Rapido spring in the universal":

>This next fix is unconventional.
>I pulled the motor out of the chassis half.
>I placed a spring in the flywheel.
>i.e., I took a Rapido coupler spring and cut it short with my rail nippers. Probably to about 2 millimeters or so. I then dropped it into the flywheel.
(JS:  on the IM SD45-2, I placed it inside the universal joint where it meets the flywheel)
>I then installed the motor, making the hex nut universal going into the flywheel.
>Did that for both sides.
>The goal was to place a SLIGHT SLIGHT amount of pressure on the universal joints / hex nut, pressing the worm outward so that it would not rattle.

John Sing's result:   noticeable improvement. 

Both the IM SD45-2 and the Atlas B40-8/C628 mechanisms stopped being 'grinders', they moved from 'too loud' to 'acceptable low growl'. 
They are clearly *not* silent, but are noticeably more quiet, just barely into the acceptable range.   That's a major improvement especially compared to how it was.   

Tomorrow, I will post YouTube video of these two locos.   Later on, I will post some pictures of the above modification. 

Bearden appears to be right - by putting SLIGHT pressure on the hex nut, it is pressing the worm gear outward, thus the worm gear doesn't rattle as badly any longer, thus quieting the mechanism.

There is an old saying, "the morning is always wiser".   Before declaring victory, I will review this all in the morning.

But for now, the result of:

a)  the kiwi_bnsf technique of testing the loco upside down on the track, to assure the drivetrain mechanism itself is quiet enough / aligned
b)  thus pointing to the worm gear/truck interaction as the culprit for the sound
c)  then applying the Bearden "Rapido spring in the flywheel/universal" technique, provided enough pressure to prevent the worm gear from rattling

All combined to create a noticeable improvement in the noise, down into the "just barely acceptable" range.  But that's a huge improvement.

We'll see in the morning (EST) if the result is still the same, and see if you all can tell the difference on a YouTube video.

-----------

No, we're not done by a long shot... this is just one set of data points.  We have good guesses, but we still don't know how to discretely measure  exactly why the original disassembly / reassembly resulted in  "excessive rattle in the worm gear area".  (Engineering principle:  if you can't measure it, you can't know exactly how much you have improved it). We are a long way from "assured silent running" after re-assembly.  We still don't know exactly if there is a way to instrument the noisy mechanism to pin-point exactly what are the (multiple?) causes of the noise.   

We are clearly finding/documenting/sharing some tests and techniques that yield improvement. 

I hope that experienced DCC decoder installers might weigh in with their experiences.... do they also see in general "more noise after re-assembly"?   How do they resolve the issues? 

There's more that needs to be investigated and more circumstances, I'm sure, to illuminate.   As Big Train stated on page 2, there's many worthy questions that we don't completely know the answers yet (at least, not yet documented here in this thread).

So please keep the input, comments, experiences coming.    Trust you are finding this valuable and having fun.

See you online in the morning 12/27.   Good night for now.

[narrowminded]

I think you guys are covering most of the issues well.  I agree with Pete that proper parts will always work, first time every time, and that's where to put the focus when designing and building machines.  That's what I do and I don't then try to fix it after the fact.  There's that other saying, "Never time to do it right.  Always time to do it over".  When everything is right it has no choice but to work well.  Being machines, without gender or intellect and free will, they don't know how not to.   One of my often used sayings about machinery is, "It doesn't operate on mood or opinion.  When it's right, it works.  When it's not, it doesn't.  End of story".  That doesn't mean it's always easy but it does mean that getting mad at it or emotionally involved is futile and only adds to the frustration when working on devices.  The answers are always very matter of fact and are best approached using all of the logic and understanding of the device that we can muster.

One suggestion that may seem to be too simple but might be useful to some is, watch what you're doing with your selection and use of tools as they can cause unintended problems.  A pair of pliers or a vise gripping a worm gear, even cautiously, lightly gripping, can cause a burr at the crest.  Hard to see but adding a click or hitch as it rotates.  Thrust bearing faces or any machined face for that matter that are scratched with your tool while manipulating can cause noise.  Select and then use your tools cautiously.  Any tool that point loads can cause a surface blemish. 

Narrowing down noises starts with understanding the mechanism and thinking it through logically.  Run the unit and by hand, load one truck then the other.  Did the noise change?  Try the opposite direction, same thing.  Did a noise change?  Then think through what effects those load changes had on the mechanism.  Loaded one way, which parts of the mechanism are seeing those loads?  Which face is the effected thrust face?  If a change occurred, look at those pieces effected. 

I also check during assembly that each sub-assembly is good.  This would include a truck assembly being free running and further, smooth running if I put a little drag on the gears while rolling it.  Did adding a small load add a hitch in the mechanism?  If so, think it through and find why.  I will often use the side of a toothpick against a shaft or round face, lightly dragging it over that piece and if it's in my hand, no cause of resistance, it should be able to spin that assembly with just that effort.  Did it?  If not, look for why.  If chasing a hitch, place a Sharpie mark on the moving parts and see if there's anything rhythmic about the offending action.  The mark may point you directly at the problem point.  Just good mechanical practice, I suppose.  But isn't that just like asking how long the string in my pocket might be?  

Also, already mentioned, if there is a "scientific" tool that's useful, it's a milliamp meter.  It is one of the handiest tools ever when working on these things as it monitors pickup as well as any and all changes in load (which binding mechanisms will introduce).  And it reads pretty small changes.  If you can see it or hear it, it will show up in the amp draw.  Again, a mark on a wheel with an amp meter needle bouncing (use an analog meter, not digital) can point to a problem and also because of the occurrence relative to a mark, can help to pinpoint the very spot in the mechanism where the problem is located.  If it's occurring at wheel speed, it's in those final components that are operating at that speed.  If it's in multiples, it's back in the gear train.  Things like that.  If it's a drop in the meter, it's a pickup problem.  If it's a spike in the meter, it's a mechanism binding problem.

Some of this is just thinking out loud and based in good mechanical practice.  There's only so much you can do sight unseen but I  hope it's helpful.  Even if only confirmation of things already known.

[atsf_arizona]

Thx, you're obviously not Narrowminded in real life, quite the opposite      Thx for your excellent input.

[mmagliaro]

Of all the suggestions and techniques brought up by narrowminded (and I like them), to me the king is the milliammeter.  It is one of the easiest things to put in place while running a model, and it never lies. 

- If you have reduced friction or other load on your mechanism, it will always show up in the current draw. 
- The meter isn't some mechanically awkward fixture that has to be attached to the engine.
- It will show you if you have intermittent contact, and if you are over stressing a motor.

[Jim Starbuck]

Interesting thread with much good input.

I've installed many decoders. Drop in and hardwired as well.
One early Intermountain SD45-2 was particularly noisy after reassembly. I found two things contributing to the noise.
1. The frame halves were parallel end to end but not top to bottom. (This has been mentioned in an earlier post) I ended up measuring the flange on the stock round frame spacer/ insulators and adding a styrene strip the same thickness along the bottom edge of the frame and this solved the problem. I think later runs of these incorporated a third spacer and frame screw in the lower part of the frame.

2. The torque to which the frame screws are tightened is critical. I've found that if they feel almost too loose, they're just about right.
Perhaps even though they tighten against the spacer, too tight may introduce a small amount of crush to the motor saddle and bearing blocks causing noise.

I recently installed hardwired decoders in a pair of LifeLike DL 109s. I set the decoders up identically and as odd as it sounds was able to "mechanically" speed match them simply by slightly tightening the fast engine and loosening the frame screws on the slow engine ever so slightly.
I guess you could measure torque of the frame screws as opposed to just going by feel to gain a scientific value.
There is an element of "more art than science" that comes into play I believe.
We used to build marine twin engines for pleasure craft and houseboats. GM 350 and 454 were common. We would build a pair exactly to the numbers (science) and one would always run just a bit better than the other until we tuned and finessed them (art) to run together.

[u18b]

John,
On page 1 I gave 11 sources for quietness.

Here is number 12

12.  Good gear mesh.

I just watched your video of the B40 mechanism with Alco trucks.  You asked what that whine was.

Of course I can't be sure, but it sounds like the worm is riding too tight on the double drive gear in the truck tower.
You said you made some kind of mods in order to make this work on RSD shells.   Did you mod the truck to chassis interface.
If you did, that my best guess is that the chassis needs to be raise about a millimeter.  You could try this by adding a shim glued to the bottom of the chassis at the gear tower opening.

I encountered this in about 1989 when I would swap out trucks on Kato GP38-2 and GP50.  Some trucks worked great-- others not. 

I also encountered this when working on my Kumata book.  I member of this board (Bob Gilmore?) sent me his brass Centipede.  One problem I noticed was the drive gear was pressing its neighbod gear to tight.  When I bent the gear frame to give a tiny gap, the noise dropped significantly.

So this issue may not apply to what you did, but it does add to the discussion in general.
We want the worm TEETH to move the gear TEETH. 
We don't want the teeth of either to hit the the bottom of the gap between the teeth.
 

[u18b]

Interesting thread with much good input.

I've installed many decoders. Drop in and hardwired as well.
One early Intermountain SD45-2 was particularly noisy after reassembly. I found two things contributing to the noise.
1. The frame halves were parallel end to end but not top to bottom. (This has been mentioned in an earlier post) I ended up measuring the flange on the stock round frame spacer/ insulators and adding a styrene strip the same thickness along the bottom edge of the frame and this solved the problem. I think later runs of these incorporated a third spacer and frame screw in the lower part of the frame.

2. The torque to which the frame screws are tightened is critical. I've found that if they feel almost too loose, they're just about right.
Perhaps even though they tighten against the spacer, too tight may introduce a small amount of crush to the motor saddle and bearing blocks causing noise.

I recently installed hardwired decoders in a pair of LifeLike DL 109s. I set the decoders up identically and as odd as it sounds was able to "mechanically" speed match them simply by slightly tightening the fast engine and loosening the frame screws on the slow engine ever so slightly.
I guess you could measure torque of the frame screws as opposed to just going by feel to gain a scientific value.
There is an element of "more art than science" that comes into play I believe.
We used to build marine twin engines for pleasure craft and houseboats. GM 350 and 454 were common. We would build a pair exactly to the numbers (science) and one would always run just a bit better than the other until we tuned and finessed them (art) to run together.

In both cases of Jim's good examples, the real issue (deeper issue?) would be that when there is a problem with the frame alignment being off, that will place un-due twisting on the worm bearings.

I also have added spacer in some locos.

Thanks Jim.

[narrowminded]

Of all the suggestions and techniques brought up by narrowminded (and I like them), to me the king is the milliammeter.  It is one of the easiest things to put in place while running a model, and it never lies. 

- If you have reduced friction or other load on your mechanism, it will always show up in the current draw. 
- The meter isn't some mechanically awkward fixture that has to be attached to the engine.
- It will show you if you have intermittent contact, and if you are over stressing a motor.

As we speak, I'm watching the milliamp meter to monitor pickup vs: added friction in my newest, smallest to date, N or Nn3 power truck (8.9mm off the rails with 24" wheels on a 44" wheelbase with a 130:1 gear reduction affording 15 scale MPH top speed).  I'm using a 100mA meter (.1 amp) and I'm working with friction added in a 1-1.5mA range from no pickups (not reliable contact) to pickups on four axles and VERY reliable.  Reliable pickup is very much there evidenced by the meter needle sitting fixed as though glued in place while creeping around at just above 0 and up to 5 scale MPH continuously without a hitch (but it's only been 6 hours or so).  So while it's running I'm jumping ahead to how to  fixture this to reliably and simply produce these parts if they continue to function well over the next week or so.  I think they will. 

And this is why I sing the praises of this tool when working on these small engines.  It's a no BS test that gives honest answers, no wishful thinking allowed.   It would prove the "too tight chassis halves" too.

[Mark W]

There's certainly a ton of info here on hardware troubleshooting.  Has anyone experimented with different lubrication?  Could different lube affect noise? 

I'd be curious to know whether stuff like Lebelle 102 oil vs 106 grease makes a difference.  Or what about graphite?

Long ago, there was the "Toothpaste Method" for silencing growlers, since toothpaste has a mild abrasive to wear down tight mesh.  Does that still have any merit with today's equipment?

[mmagliaro]

There's certainly a ton of info here on hardware troubleshooting.  Has anyone experimented with different lubrication?  Could different lube affect noise? 

I'd be curious to know whether stuff like Lebelle 102 oil vs 106 grease makes a difference.  Or what about graphite?

Long ago, there was the "Toothpaste Method" for silencing growlers, since toothpaste has a mild abrasive to wear down tight mesh.  Does that still have any merit with today's equipment?

Except for a light oil in the sintered bronze motor shaft bearings, my general position is that lubrication does very very little to reduce friction or noise unless you have two metal gears meshing together, which is pretty rare in N Scale engines.  They are usually a plastic or brass worm meshed to a plastic gear, and the rest of the gears are usually plastic as well.  Brass is a different story, often having more metal gears.  My experience is that oil does little except attract more dirt and gunk, and it's often used as a desperate attempt to solve a more fundamental problem.  Occasionally there are axle shafts riding in metal housings; those could benefit from lubrication.

I use the heavier oil on gears, and the lighter on bearings and axles.

Graphite: I've actually tried this and as I expected, on gears all it does is make a mess (same thing with NeoLube which is graphite in a fluid suspension).  It may seem to help at first, but in short order, it attracts enough dirt and other greasy residue that it becomes a sticky, gummy substance that makes for more friction, not less.  On side rods and valve gear, NeoLube can help as long as you don't already have oil on there.

It's very hard to find anything slipperier than Delrin against Delrin.  That's why oil is of little use on those gears and other surfaces.

[Big Train]

Some of the other things I have thought about is, in the case of flywheel equipped engine, is to apply a heavy grease to lube the hex nut and reduce vibration. Too bad we can't get viscous couplers in N Scale like they use in the constant velocity joints in front wheel drive cars!

The other extreme is apply a very flexible sealant/adhesive to achieve the same result as  the small spring installed between the flywheel and hex nut of the drive shaft as mentioned earlier in this discussion.

I've sometimes found packing the entire truck/bogie assembly with heavy grease helps a long with ample amount in the gear tower. But that means we're no better than some manufacturers that do the same thing during production.

About the bearing blocks themselves. Should they have a little play when seated in the frame? Do they need to be looser to allow some play or
tighter to eliminate play altogether? And the bronze bearing block tensioner: Fingers too loose or too tight against the bearing blocks, Or remove them altogether?

Being on good terms with the dimensional gauge technician at one place I worked, I was given a steel surface block which is absolutely flat. Just laying a few frames on the block sure opened my eyes at the deflection I've seen. Slight to be sure and maybe a contributor to excess noise.

It's at the point right now that when I install decoders, more than a few engines develop some sort of noise to be tracked down as best as possible, so I'm really reticent to acquire engines without DCC already installed even though they cost more.

Someone earlier had mentioned loosely tightening the frame screws and nuts. This is a technique I also use to find the "sweet spot".. But, I had several Intermountain F Units a few years back and after installing decoders and finding the "sweet spot", both trucks/bogies would not stay in the mechanism. Really embarrassing when you pick up an engine and the trucks remain on the track! Even when tightening as I hard as I could, the trucks would never stay in place. After I found the right torque that more or less allowed the trucks to remain attached, the shells were too loose. IM did step up and sent me replacement trucks, though. Now these F units are noisier than my older ones, and will probably stay that way until a workable solution is found.

Working together for a better N Scale experience.

[randgust]

"back in the day" when Trix and Rapido were common, there were a lot more metal gears out there and I used a lot of Vaseline as gear lubricant.  It was very effective on metal gears, quieted things down a lot, also made one awful mess over time.  Didn't seem to affect plastics.  Used to intrigue me because you could actually see some trapped metal particles in it over time.

The only thing I've seen lubricant do that is an 'instant fix' is the notorious Kato Squeal out of the motor end bearings, and one drop of LaBelle 108 fixes it for a year or longer.   Wow, was that annoying.

I've pretty much honed in on the gear train alignment as the source of most noise; universals slopping around, motor out of alignment and even rubbing into the frame/body with the flywheels, worm shaft bearing blocks out of alignment, etc.     I've had the notorious "Atlas Click" out of SD50 chassis to the point where I replaced them with Kato SD40-2 ones.   

One thing I haven't seen mentioned is the horizontal tolerances on the drive train; i.e. taking up slop with washers on either side of the worm to the bearing blocks, moving the universals slightly in or out of the shaft so that the motor isn't binding on them but also isn't allowing the worm to bash back and forth pushing the universals into a noisy spot.   I've come to the conclusion that 'most' racket comes out of the entire universal drive area, second most popular glitch is a loose or misplaced worm bearing block.

My general rule is to try, really try, to put indicator marks on everything when I take it apart, including the orientation of the bearing blocks and numbering them by scratching marks in them if you're lucky enough to have a quiet one. 

As far as for what you're seeing still, I'll line up with the idea that the motor bracket has shifted ever so slightly during disassembly and reassembly.  That's a tough one, but moving it around by pushing the flywheels different directlons may prove the point. 

Having said all that, I have a really nice and quiet original Atlas Classic GP7.   I want to change the motor to slow speed, but that's one of those things that I know my odds of keeping it as quiet as it already is is 'maybe' 50-50.   So I've decided against it.   If I could find the right part number for the NEW universals Atlas has I might risk it, but until then, nope.   And I've messed with these since they first came out.  And I know they are still capable of turning into very noisy locomotives for no good reason at all from dissassembly.   I've completely retired three of them so far just due to racket.   I'd junk the entire mechanism, but the Classic chassis is heavier, wheels are worn to the point they grip better than average, and I'd take a 25% hit in tractive effort on a yard loco that regularly has to yank a 25-car cut on level track.   Nope.

[atsf_arizona]

Hi, folks,

As promised, here are YouTube videos showing before and after implementing the "Ron Bearden Rapido 1/2 spring in the flywheel / universal joint" modification discussed earlier in this thread.

Made a noticeable improvement.  The previously extremely noisy mechanisms are now reduced to "tolerable growl" - which frankly, is a major improvement from where they were.

The modification that Ron talked about is here:


http://www.pbase.com/atsf_arizona/image/164753331

And the words were:

http://forum.atlasrr.com/forum/topic.asp?ARCHIVE=true&TOPIC_ID=43690  - Ron Bearden's post of 2008 January 13 :  3:23:31 PM

 "Rapido spring in the universal" explanation:

>This next fix is unconventional.
>I pulled the motor out of the chassis half.
>I placed a spring in the flywheel.
>i.e., I took a Rapido coupler spring and cut it short with my rail nippers. Probably to about 2 millimeters or so. I then dropped it into the flywheel.
(JS:  in the above picture, that's the shorter spring, being placed inside the flywheel, to put slight outward pressure on the hex nut)
>then installed the motor, making the hex nut universal going into the flywheel.
>Did that for both sides.
>The goal was to place a SLIGHT SLIGHT amount of pressure on the universal joints / hex nut, pressing the worm outward so that it would not rattle.

1)   To appreciate how much of a difference that made, you have to see how BAD the IM SD45-2 was before here (via a TubeChop):  http://www.tubechop.com/watch/8728407

2)  Then watch this and compare to see it is noticeably better now:

/>
Clearly, not silent, so more work to be done.  But clearly a noticeable improvement, indicating reducing the "worm rattle" clearly reduced the noise.  In person, this mechanism is now "tolerable" at scale speeds.  Before it was unacceptable.
-----

Here's the Atlas B40-8 / C628 mechanism, with the same modification.  Here the results are better:

3)   You have to see how BAD the  Atlas B40-8 / C628 mechanism was before (via a TubeChop): http://www.tubechop.com/watch/8728439


4)  In order to appreciate how much better it is now:

/>
Again, not silent, more work to be done, but clearly noticeable improvement.  While not whisper silent, I can live with this amount of noise if I have to.  And perhaps we can find more improvements.

I do  notice that it is more quiet in one direction vs. the other. 

Ron Bearden and Randgust suggested I check and see if the worm gear / gear tower are too close together.

So those are the next areas of investigation.   Time to apply further some of Randgust's and other's suggestions...... 



I will say this.... in less than 48 hours since this thread started,  these two locos have gone from "junk" to "tolerable".  Considering I've been toying with the IM SD45-2 for nearly 6 months trying to quiet it down, this is wonderful.   Thank You to all of you for the crowd-sourced, accelerated results and learning . 

Let's continue the thread as I trust this input continues to be very fruitful.