Author Topic: Planetary gearbox - an attempt to make sense of how it works  (Read 7053 times)

Offline sorveltaja

  • Full Member
  • ****
  • Posts: 87
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #30 on: April 19, 2019, 12:23:41 AM »
Don, not so silly question. I haven't even thought about tuning the printer. When I got it, I started to print the project parts right away.

To be honest, I really didn't expect the printer to be that precise, after all. With standard settings, there is always some extra material, that is then machined away, to match the actual drawings.

Here is a screenshot of the 'control panel' from the Flashforge Finders manual:


I'll have to admit, that I have no idea, of how to manage those things.
« Last Edit: April 20, 2019, 11:34:16 AM by sorveltaja »

Offline sorveltaja

  • Full Member
  • ****
  • Posts: 87
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #31 on: April 20, 2019, 11:25:29 PM »
I searched the net about how to fine tune the steps/mm. Tried some freeware programs, but wasn't able to get any of them to connect to my printer through USB.
In fact, those programs don't even seem to have a USB option.

I guess that the programs, that allow the user to take 'full control' of the printers settings, are more compatible with 'open source' printers. Obviously mine is just a plug and play, entry level device.

However, what comes to the project, I had to scale it down about x 0,6. At first, at the original scale, I printed one of the brake spools. Then took it to lathe, and soon figured out, that there wasn't enough
room to attach any cutting tools to cross slide :facepalm2:.  That brake spool has about 65mm diameter, and if memory serves, it must be the largest part, that I have ever tried to machine with my lathe.

For size comparison, big one on the right:


While scaling down, I'll also simplify them, and increase the wall thickness of the parts, that are to be threaded for screws, as they are made of plastic after all.

Offline sorveltaja

  • Full Member
  • ****
  • Posts: 87
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #32 on: April 22, 2019, 01:17:41 AM »
After I got the planet gears printed at 'normal' -setting with Flashprint, they were really tight to fit together with ring gears. So I printed one pair of the same gears with 'high' -setting, and still same result.

Tried to fit them by turning the outside diameter down about 0,4mm. Again, too tight fit :headscratch:.

Yet another pair of the gears were printed, and I measured the outer-, and the root diameter. Outer diameter was very close to drawing, while the root diameter, that was supposed to be 7,5mm, was 8,2mm. 

I drew a bogus version of the same gear with 6,5mm root diameter, and printed it. Again, the outside diameter measured ok, but the root diameter was 7,0mm, instead of 6,5mm:


So I decided to take a closer look, of why the Flashprint insist doing that. So far I have found, that it uses 'ffslicer' to do its thing (and create the gcode?).

On the same ffslicer- file folder, there is also a config file, which consist numerous variables, for which I searched documentation on the net, just to find, that there isn't any info available. Got to love the closed systems :smokin2:

As far as I understand, the gcode is, what commands the printer to do, what it does. So maybe there is some hints on the gcode files... I took another look on the net, of how to import gcode files back to 3d-program, to see the actual tool paths, and hoping to modify them, as needed. Results were very sparse. Seems to be yet another 'closed system'.

I installed the Slic3r, as an alternative to ffslicer, but it needs plenty of time to get used to.

Offline dozerdroid

  • Jr. member
  • **
  • Posts: 1
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #33 on: April 22, 2019, 02:35:40 AM »
Sorveltaja, just wanted to thank you for this thread, following closely .. I've always been very interested in transmissions  :D .. Droid. (I have read this site daily for numerous years .. My first post )

Online Vixen

  • Full Member
  • ****
  • Posts: 1358
  • Hampshire UK
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #34 on: April 22, 2019, 12:22:46 PM »
Hello Sorveltaja

I have more experience with planetary (epicyclic) gearboxes than with 3D printing so can only offer a guess rather than a positive solution to your printing problem.

As I see it, you make an accurate drawing, some software slices it and more software creates the g-code, yet more software moves the extruder head to print a filament of plastic. The problem appears that you do not know precisely at which stage in the software process the errors occur. Some calibration test prints may help you.

Could you make a drawing and print a cylinder of exactly 25mm diameter, another at exactly 50mm diameter and a third as large as your printer will allow. You can check the g-code to see if it matches the diameters of the three cylinders. You can then measure the printed cylinders for size and roundness. This may help you identify the source of the errors. The errors may be a constant offset error or the error may be proportional to the size of the object. You could try to compensate for the errors in some way once you have identified the source.

Do you know how the software compensates for the width (thickness) of the extruded plastic filament

Mike
It is the journey that matters, not the destination

Offline sorveltaja

  • Full Member
  • ****
  • Posts: 87
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #35 on: April 22, 2019, 11:47:15 PM »
Dozerdroid, thanks and warm welcome to the forum :ThumbsUp:

Mike, I made a drawing for the 25, 50 and 140mm cylinders(printing area is max.140x140x140mm), height is 5mm. Well not actually cylinders, more like rings, as the biggest one would have taken over 6 hours to print at 'high' -setting, if it was a solid one:


And here are the settings, that I use to print the rings:


I printed them one at a time. Smaller one's outer diameter is spot on 25,0mm, also height is 5,0mm, but the inner diameter is 14,8mm. Bigger one's outer diameter is again spot on 50,0mm, and height is between 5,0 and 5,1mm. Inner diameter is 29,9mm:
 

Biggest one is still printing, and its inner diameter might be even closer. Generally the z-axis(horizontal) seems to be quite precise, also the roundness of the printed parts is quite good.

Due to the printed surfaces not being as smooth as machined ones, mating parts always seem to need some extra in them, which allows machining for closer fit.

And yes, error(s) seems to be proportional to part's size. That's why I chose quite bulky gear size(mod 1), when scaling down the transmission's parts. Even then, as mentioned earlier, there is some kind of an offset jazz going on with the smaller gear's(suns and planets) root diameter. Ring gears, as they are three times bigger in diameter, are somewhat more precise.

I haven't found an easy way to extract the actual tool paths from the gcode files for editing and measuring. I could do it manually, but after editing, the gcode file should still have a certain structure, as it also seems to contain plenty of commands other than coordinates, for the printer to accept it. Maybe I could do it for a single file, just to see the containing dimensions of the small gear.

But again, it might concern only Flashforge Finder, its bundled software, and probably also its firmware. 

What comes to how the software(Flashprint) compensates for the width of filament, I'm not sure, as the manual doesn't mention it.
« Last Edit: April 22, 2019, 11:50:16 PM by sorveltaja »

Offline dieselpilot

  • Full Member
  • ****
  • Posts: 281
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #36 on: April 23, 2019, 02:03:29 PM »
To maintain a high linear velocity, printers often round corners. This is my guess as to what's happening. Running one of the calibration prints would probably give a good idea of what's happening. Printers handle this in different ways and it have to do with material feeding and/or acceleration rates. I have zero experience with printers, only what I read about in terms of how the controls work. It would also be worth reading up to see what others are experiencing with this particular machine.

Online Vixen

  • Full Member
  • ****
  • Posts: 1358
  • Hampshire UK
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #37 on: April 23, 2019, 06:39:10 PM »
When I make my planetary (epicyclic) gearboxes, I use hardened steel for the (annulus) ring gear and for the sun and planet gears. All the gears and the gear centre distances for the planet carriers etc, need to be machined to an accuracy in the order 0.025 to 0.05mm (1 to 2 thou) for the gearbox to work successfully. That is why I use or modify commercially available gears such as the epicyclic gears from a Sturmey Archer bike hub, whenever possible.

Each of the planetary (epicyclic) stages must spin freely, without friction. If any of the gears bind or cause friction, it has the same effect as applying a gear selection brake. The Merritt-Wilson planetary (epicyclic) gearbox can only toreate ONE gear selection brake being applied otherwise the whole gearbox will lock up and damage itself.

Precision is therefor a fundamental necessity for planetary (epicyclic) gearboxes, Achieving the necessary precision is a lot to ask from a filament extruder printer. It will be a challenge.

Mike

« Last Edit: April 23, 2019, 08:52:26 PM by Vixen »
It is the journey that matters, not the destination

Offline sorveltaja

  • Full Member
  • ****
  • Posts: 87
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #38 on: April 23, 2019, 11:41:06 PM »
Dieselpilot, yes there are plenty of factors involved. When I took a look at the gcode file, there were numerous 'ratio' -commands here and there. Are they there for the extruder - possibly.
I have looked also for other's experiences with Finder, but mostly they just print out the sample boat, and judge the print quality by the looks of it.

There is one video on Youtube, where the user tests the accuracy by printing single planetary gearset in one go, to see, if they can be detached from each other after the printing.
That kind of tests seem to be very rare. Of course, the printer he had was of different brand, and maybe not so 'entry level', as he had updated the extruder, or some of that stuff, before taking it to test.

Mike, thanks for the explanation. To overcome the printer's lack of accuracy, considering the gears, hobbing is what comes to mind. Print the sun- and planet gears a bit oversize, and hobbing takes care of the rest.

For me, making the actual hob is a big question mark, though. Should it necessarily be a cutting tool, or could it be an abrasive one instead, as there shouldn't be that much material(PLA plastic) to be removed after all, in this scale? At a low speed, with some cutting fluid now and then, to allow the debris to flow away, and lessen the friction.

Something like an iron powder impregnated epoxy perhaps, which is then poured to printed mold. But enough of that plan 9 from outer space.

The project goes on. I can hardly wait to get it ready to go:







Offline ddmckee54

  • Full Member
  • ****
  • Posts: 260
  • We're having fun now --- or so I've been told.
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #39 on: April 25, 2019, 09:57:59 PM »
When tuning my printer I cheat, I use gcode commands.

From my machine control panel in Simplify3D I send the M503 command, you can do the same from what ever you are using to run the printer.  That will return the current Steps/mm values for all axis.  That will return values that will look something like this;  X100 Y100 Z400 E100.  When I want to get info out of the printer is the only time I hook a PC up to it.  I normally print from the SD card and use my PC for other things.

If for instance, your X axis is printing oversized, and your Y axis was printing undersized, that would mean that mean that X needs to take less Steps/mm and Y needs to take more.  I'm not going to go through how to calculate the correct values, I'd just get it wrong and make myself look sillier than I already do.

The next thing we need to do is get that information into the printer's brains so it can be put to use.  If you have the correct version of your software and are comfortable with modifying it that way - have at it.  I've been a programmer for years and since I DON'T have the known correct version of the software, I'm too afraid I'll fix one thing and break 12 others if I try that route.

So what's that leave us?  Yup, BRUTE force, we're gonna use gcode.  In particular we are going to use the M92 command to set the Steps/mm.  I add a line similar to the one below into the start-up script that is run every time a gcode file is loaded to the printer:
M92 X90.01 Y110.95 Z400 E100;
I'll keep tuning those values until I get to a point where it's "Good 'nough for gubmint work".  Currently my 20mm cubes are measuring 20.02mmx20.02mmx19.98mm.  I could probably tune it more, but at this point it's "Crose enough".

It doesn't matter which slicer you use, somewhere they are going to give you the ability to send custom gcode to the printer.  I've also done the same thing to tweak the PID parameters, I just can't remember that M command right now.  By downloading these values to the printer every time a gcode file is loaded the default printer values are overwritten with the "tuned" values.  I'm ensured that if a part don't fit - it's because I screwed up somewhere not the printer.

Don

Offline sorveltaja

  • Full Member
  • ****
  • Posts: 87
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #40 on: April 25, 2019, 10:27:35 PM »
Today I took the transmission to the lathe to spin it for testing. Third and second gears weren't that bad, but first gear was a lot noisier. The reverse gear, where the input shaft's rotational force seems to go through all the other stages or gears, it got really crakling.


As Mike mentioned earlier, tolerances are extremely important. Even smallest errors in gear's geometry (and noises)multiply at every single stage. I can fully understand, why he sticks to readily available gears.
I'd do the exact same.

What comes to 3d-printed gearsets, I think I'll increase the tooth size next, from mod 1 to mod 1,5:


There is a caveat, when increasing the tooth size, when needed to fit to certain(small) dimensions, though. Gears with very few teeth start to lose their meshing ability. Somewhere I read, that 13 teeth is an absolute minimum.

As can be seen on the above picture, mod 1,5 planets- and sun gear have only 7 teeth, and look already too 'fat', when compared to mod 1 ones.

But anyways, I'm going to take liberties with that. No matter what form the teeth are going to have, the goal, or at least the aim, is to make a more 'forgiving', full gearset, to keep the noise level down,
while also allowing it to rotate more freely.

Lesser teeth seems to be an option at the moment. And no, I'm not a dentist.


Offline sorveltaja

  • Full Member
  • ****
  • Posts: 87
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #41 on: April 25, 2019, 10:51:38 PM »
Don, thanks. I have tried Simplify3D also, but haven't had much luck with it. Somehow it just doesn't recognise my printer at all.

Online Vixen

  • Full Member
  • ****
  • Posts: 1358
  • Hampshire UK
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #42 on: April 25, 2019, 11:13:50 PM »
Hello sorveltaja

Congratulations for getting this far. A noisy three speed and reverse epicyclic gearbox using printed gears is a real achievement. With any Wilson gearbox, all the epicyclic stages spin at different speeds, The speed of the output shaft is dependant on which gear band brake is selected. First and Reverse gears will the worst as the overall reduction is the highest and reverse has a direction change as well. Some of the epicyclic stages will be spinning really fast.

What's you reasoning for going from 1.0 Mod to 1.5 Mod?

My instinctive reaction would be to increase the tooth count to reduce the tooth meshing noise, unfortunately reducing the tooth size also increases the precision required.

It is normal practice to have three planet wheels, instead of just the two. This will have the effect of increasing tooth count as well as improving the load distribution and concentricity of the sun and  planets, as they spin inside annulus gears.

Do you have an end use for your epicyclic gearbox or is it simply an exercise in design and 3D printing?

My Merritt-Wilson transmission system was designed for use in a 45Kg 1/6 scale, all metal, model tank (AFV) driven by a 2HP four stroke petrol engine. The Sturmey-Archer bike gear hubs provided the epicyclics for both the Wilson gearbox and for the Merritt double differential track steering. The S-A bike hubs selflessly donated their internal organs for scientific purposes. They proved to be a strong and very robust solution. They were also quite noisy, but then, so is any tank.

Mike
« Last Edit: April 25, 2019, 11:20:17 PM by Vixen »
It is the journey that matters, not the destination

Online Vixen

  • Full Member
  • ****
  • Posts: 1358
  • Hampshire UK
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #43 on: April 25, 2019, 11:58:18 PM »
Hello sorveltaja

I found some photos of the Sturmey-Archer epicyclic gears.

These are the sun, planet and annulus (ring) gear as removed from the bike hub but before any modifications. You can clearly see that there are four planet gears inside the planet carrier housing. The extra planet wheels improve concentricity and load sharing.



Here I have installed the four plant wheels inside a new planet carrier housing and have cut the annulus (ring) gear free using an abrasive cut off disk in a Dremmel  The annulus is mounted on a new aluminium carrier containing a bearing and a toothed belt drive gear. These components are configured as a steering differential on a 1/6 scale all metal model tank.


This is the final assembly of the Sturmey Archer epicyclic stage used as a steering differential on a 1/6 scale all metal model tank.


Your illustrations only show two planet gears for simplicity, three or four planet gears are more normal.

Mike
It is the journey that matters, not the destination

Offline ddmckee54

  • Full Member
  • ****
  • Posts: 260
  • We're having fun now --- or so I've been told.
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #44 on: April 26, 2019, 08:17:06 PM »
I think you said that you're using Slic3r?  I always had trouble with Slic3r and outer dimension stability.  It's been a while since I used Slic3r, but if I remember correctly I had better luck with it when I fixed the filament width at the value of the nozzle width.

My theory at the time was that a nozzle is NOT going to extrude a filament that is less than the width of the nozzle opening.  I think that Slic3r is calculating the filament width based on a percentage of layer height as a default setting.  A filament width calculated on a layer height less than 0.4mm will give a filament width of less than 0.4mm.  If you're actually printing a 0.4mm wide ribbon and the software says the width of that ribbon is 0.2mm that extra material has to go someplace.  I don't know if that's correct or not, won't be the first time I was operating on an incorrect assumption.

Out of idle curiosity, what do you mean when you say Simplify3D doesn't recognize your machine?

Don