Author Topic: Planetary gearbox - an attempt to make sense of how it works  (Read 6886 times)

Offline ddmckee54

  • Full Member
  • ****
  • Posts: 256
  • We're having fun now --- or so I've been told.
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #60 on: May 20, 2019, 06:00:31 PM »
sorveltaja:

If you are still interested tuning your printer, here's something you might think about trying.  When I got my Utilimaker and wanted to tweak the steps/mm settings I couldn't find them listed in any of the available menus on the machine.  I downloaded Mattercontrol from Matterhackers to be able to directly control the printer from my PC.  It's a free 3D printer interface, I checked and it's compatible with Flashforge 3D printers.  Using Mattercontrol I was able to connect to my printer through the USB, send the M503 command, and the printer replied to Mattercontrol with the results.  It's kind of silly but the only purpose that Mattercontrol has been used for by me was to get the settings out of the printer using the M503 command.

Don   

Offline sorveltaja

  • Full Member
  • ****
  • Posts: 86
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #61 on: May 21, 2019, 08:22:59 PM »
Thanks, Don. But I couldn't get the Mattercontrol to install correctly. Problem is the newer version of the .net framework, that it requires to run. It just doesn't want to install. There is always errors.

I spent most of the day trying different ways to get it to install. I don't seem to be the only one with problems of installation of .net framework version newer than 4, though.

I ended up removing older versions of it, and managed to make a mess. Programs that required those older versions, didn't run anymore. So I tried to install them back again, but that didn't work either.

As much as I'd like to test Mattercontrol, I had to give up. I had some older system image, which I used to replace the screwed windows partition.
   
« Last Edit: May 21, 2019, 09:14:14 PM by sorveltaja »

Offline ddmckee54

  • Full Member
  • ****
  • Posts: 256
  • We're having fun now --- or so I've been told.
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #62 on: May 22, 2019, 03:20:10 PM »
I feel your pain.  Don't you just love software upgrades that fix some things and yet break others?

At work, we've got a software package that collects utilities usage - gas, steam, electric, etc.  This software requires a version of Java that is well...  let's just say several versions out of date.  Replacing this software would be a massive project since the new version is not compatible with all the hardware that we have installed, and there's a LOT of hardware installed.

There are several key players in the company that require this software for the monthly financial statements.  Whenever those people get a new computer we have to be sure and install that old version of Java.  The computer doesn't even have to be using the older version of Java, it just has to be installed.

Yup, I know what you just went through, been there - done that!

Don

Offline sorveltaja

  • Full Member
  • ****
  • Posts: 86
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #63 on: May 23, 2019, 01:23:56 AM »
Yes it can be quite frustrating, when those things happen. Something like 15 years ago, things were a lot simpler, what comes to windows. As the technology gets more and more advanced, so does the problems involved also, unfortunately.

What comes to Mattercontrol, it seems to have a Linux version also. In the past I used to fiddle with linux just to see, how far I can get with it.

Mostly to see, if I can get all the devices to work, that work in windows. And to test different 'windows only' -programs with wine(emulator).
Luckily there are plenty of documentation, but it sure needs a serious dedication to dive in to that deep subject.

After all these years, I might give it a shot, and try something like Linux mint.

But anyways, back to the original subject. I have scaled the transmission down to half the size, which makes the gears to about mod 0.5. To compensate the printer's tendency to exaggerate the outer dimensions, I've scaled the planet gears down also, so far, that they just fit in. With scale factor 0.97, there is still some tightness.

Scaling the planet gears down seems to work a bit better than using offset, for some reason.

Here is the comparison:


Maybe I finally get my head around the idea of using smaller tooth size instead of bulky one. If the gears are ideally seen as plain rollers, what can be closer to that, than the smaller tooth size?

Basically, I guess, that the tooth size could be as small as possible, as long as the gears mesh together, and therefore provide required ratios.

The original version of the printed transmission uses mod 1 gears(picture above), but should be a breeze to redraw it with mod 0.5 gears.




Offline ddmckee54

  • Full Member
  • ****
  • Posts: 256
  • We're having fun now --- or so I've been told.
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #64 on: May 23, 2019, 07:43:19 PM »
What's the actual size of the 1/2 scale transmission?  Past a certain point it doesn't matter what it looks like on the screen, our FDM 3D printers aren't physically capable of printing the object.  I've got a feeling that your small gears may be approaching that limit.

Our glorified hot glue guns can only do so much.

Don

Offline sorveltaja

  • Full Member
  • ****
  • Posts: 86
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #65 on: May 24, 2019, 01:12:54 AM »
Here are the smallest gears and the largest spool:


Offline ddmckee54

  • Full Member
  • ****
  • Posts: 256
  • We're having fun now --- or so I've been told.
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #66 on: May 24, 2019, 10:01:55 PM »
How thick are the gear teeth at the minor diameter for the 6.54mm gears?  When objects get small enough your slicing software will start taking liberties with how the part is actually printed.  The software will simplify the perimeter it thinks the perimeter's too small, sometimes it simplifies it a lot.

I've printed 12 sided shells that held M3 nuts at about the dimension of your gear and they came out OK.  Then again I've tried printing features smaller than that and they just came out looking like a mostly unrecognizable blob.

Good luck on your gears.

Don

Offline sorveltaja

  • Full Member
  • ****
  • Posts: 86
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #67 on: May 25, 2019, 03:01:07 AM »
Here is the thickness of the teeth at the root diameter, which of course isn't that precise, when the gear is printed out:


Reason for me going to that half size transmission is simple - testing. Printing all the parts with 100% infill takes a lot less time.

Why 100% infill then? To lessen the noise of the gear box. No matter how well you 'post process' the printed parts by machining, or by other means, there seems to be always hint of eccentricity involved.

To minimise that, I've played with scaling and offsets, for the gears to have a smoother meshing, and less friction.

All that matters less, if you are aiming for a hand-cranked model of the transmission.

But that's not, what I'm after. I'm in for a model, which could hopefully be operated by something more constant power source, like an electric motor, and perhaps some load on the output side.

I'm looking forward to get to the gear changing mechanism, once I get the other parts of the transmission finished.
 

 

Offline AOG

  • Full Member
  • ****
  • Posts: 672
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #68 on: May 25, 2019, 06:12:53 AM »
I just noticed this thread and I had some of the same issues as you are having. First of all I have a flashforge creator pro. It uses something called x3g which is not G code. I suspect your printer does to. The other thing I found out is that the line width is set by the diameter of your nozzle. .4 mm seems the be the standard size for most nozzles. As I understand it, with the extruder set to the normal rate you will get a line at nozzle width with the height that you selected. Supposedly you can vary the width somewhat by increasing or decreasing your flow rate.  With the size of your teeth and a .4 nozzle, you are going to have problems approximating the shape. You might to consider changing from involute to cycloidal gears. They may be easier to reproduce at that size.

Tony

Offline sorveltaja

  • Full Member
  • ****
  • Posts: 86
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #69 on: May 26, 2019, 12:15:45 AM »
Thanks for the tips, Tony. I found an online 'Cycloidal Gear Builder' http://hessmer.org/gears/CycloidalGearBuilder.html

So I generated a 0.5 mod pinion with 12 teeth, and printed it out. Its size is quite close to that 'normal' 12 teeth gear. But the printed pinion had some of the spaces between teeth filled, and also had some blobs on the outer surfaces.

Then I took a look of flashprint's settings on the 'expert mode', and found the extrusion ratio, which is 109% by default. Maybe that's the flow rate adjustment ?


I changed the extrusion ratio from 109% to 100%, and printed the pinion once again, But still the same results. Reason for that might be, that the pinion has a lot thinner teeth at the root diameter:
 

I have just reached the limit of my printer, I guess.

Nevertheless, the idea of using cycloidal gears is interesting, as there seems to be way more tolerance between the gears(pinions and wheels), than with involute gears.

It could, or should be more forgiving, what comes to slight eccentricity, which the printed parts always have, thus eliminating some of the cumulated friction(and noise), which definitely occurs in every printed, multi-stage geared devices.

To verify all those fancy claims, there is only one way for me to go - print and test. But not with mod 0.5, no. More like 0.75, or better yet, mod 1.0 to make sure, that the printed gears have valid shapes.

The original gearbox has already mod 1 involute gears, so there is no need to start drawing from the scratch.








Offline sorveltaja

  • Full Member
  • ****
  • Posts: 86
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #70 on: May 28, 2019, 12:57:32 AM »
Before going more into the cycloidal gears, there is still the small transmission, which doesn't rotate smoothly enough to make it a good working model:


It is almost impossible to iron out all the eccentricity, that the gears have. But there is one more thing, that I haven't tried yet - hobbing(sort of).

They are mod 0.5 gears, and at least the planet- and sun gears could be hobbed. I just rediscovered the set of cheapo mod 0.5 gear cutters, that I have. I picked the one, that is for 12-13 teeth gears:


Then it was attached to the lathe tool post, so that it can cut a teeth shaped grooves like this:


They are just straight grooves on a 40mm nylon bar. The grooves measure like this:


It was just a test to see how the gears meshed with the 'hob'. So far so good. But I prefer metal, so I'll make another hob out of 25mm round aluminum bar to make sure, that it retains its shape.

But what is a hob without cutting teeth? Maybe not a hob at all, but the one that uses friction to push the material away from the PLA-plastic gears, hopefully


Offline ddmckee54

  • Full Member
  • ****
  • Posts: 256
  • We're having fun now --- or so I've been told.
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #71 on: May 28, 2019, 09:00:16 PM »
You're going to have a hard enough time trying to machine the gear teeth by cutting them, I believe that trying to use friction to push the plastic around is just asking for delamination problems. 

The extrusion ratio is the variable that the slicing software uses to modify the calculated extruder feed rate.  Not knowing how Flashforge handles the extruder feedrate calculation we don't know the exact mechanics of this calculation.  We do know that the feed rate will be based on the volume of material required for each printer move taking the length of filament required for the move, the printer speed, the filament width, and the layer height into account.  With the ratio setting at 109% you were telling the printer to constantly extrude 9% more filament than it thinks it needs.  Over-extruding is handy on the 1st and last layers to get a smoother finish on the surface.  For all the rest of the layers, that extra filament had to go somewhere which might account for some of your dimensional variation.

Your minimum feature size is going to be determined by the width of the filament being extruded, I'm guessing that would be your Path Width setting.  It looks like this is a "fixed width" setting.  No matter what the layer thickness the filament width will always be this value.  For comparison, Slic3r defaults to a filament width that varies with the layer thickness, but it can be set to a fixed width.  With your current 0.4mm Path Width setting, I'm guessing that what you would actually get when you try printing these 0.7mm wide gear teeth will be a single 0.4mm wide straight sided feature.  At a 0.4mm path width, the slicing software cannot fit a continuous perimeter around the object, that would require minimum width of at least 0.8mm, so the teeth would be printed as a single 0.4mm wide line.

I have YET to find a slicing software that fully explains what modifying the individual settings will actually do.  Sometimes you just have to play with the settings to find out what they will do.

Don
« Last Edit: May 28, 2019, 09:05:36 PM by ddmckee54 »

Offline sorveltaja

  • Full Member
  • ****
  • Posts: 86
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #72 on: June 01, 2019, 02:07:32 AM »
Don, you are right. Result of the friction hobbing:


There is one thing, that I noticed, when printing smaller mod 0.5 gears. All of them are ok on the right side, but the left side has constantly a lot of defects, no matter if they are printed one at a time, or three or four at a time.


I'm tempted to go different route, and machine all the planet- and sun gears from the blanks of pla, or other kind of plastic. But that's not the way to make a working model of the transmission, which should be easily reproduced, as many, who are interested, don't have access to all the machines and the tools, that it requires.

I have machined only few gears in the past out of metal, and the amount of the work needed was/is daunting. My humble greetings to all those, who machine their gears out of metal.

That's why I'm going to move to mod 1.0 cycloid gears. I Did some test prints, and the resulting gears had correct shapes, without visible defects.


 

Offline sorveltaja

  • Full Member
  • ****
  • Posts: 86
  • Finland
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #73 on: June 01, 2019, 10:44:49 PM »
Instead of going straight to mod 1.0 teeth size, I started to wonder, what is the actual smallest teeth size, that my printer can print cleanly. So I printed some mod 0.6 cycloidal gears, and to my surprise, this was the result:
 

The ring gear spool has some pitting on the uppermost layer, but the teeth are just fine. No blobs or other defects on the smallest gears. Of course they have small 'skirts' on the lowermost part, that are easily trimmed away in the lathe. But then again, all the parts, even though they are printed without brim, have skirt in them.

Easier workaround could be to print the gears something like 1mm taller, and then simply sanding them to desired length, thus removing the skirt also.

Little about the shapes of the cycloidal gears - especially ring gears. As mentioned before, I use "Cycloidal Gear Builder" at http://hessmer.org/gears/CycloidalGearBuilder.html

There isn't an option to build ring gears, but I simply built a 'wheel' -gear, which had the same amount of teeth, that the desired ring gear has. It had to be modified a bit, though. I took one tooth apart, and mirrored it to replace inwards pointing teeth:


On the left is the original shape, and on the right side is modified one. The original shape seems to have already some tight spots with pinion gears.

Just another scetch, where the tips of the inwards pointing teeth are replaced with round shapes. But that one could have tight spots as well:


One more thing, that I tested. Does the quality of the prints increase, if you make the objects surface 'finer', by dividing it to more smaller surfaces, like this:


Answer is simple - yes to a certain point. The simpler gear on the left is about 40kb in size, when exported to stl-file. The gx-file is about 310kb.

Then I went overboard with the gear on the right side. I started to divide its surface, resulting larger and larger stl- and gx-files(megabytes). At the end I had a 52Mb stl-file.

But no matter how big and detailed the stl-file was, the slicer(at least on flashprint) created a gx-file, that was never over 7-something megabytes. Again one mighty limit, but it could be about the resolution of the hardware of the printer itself.

"Printer's got to know its limitations", as Clint Eastwood would have said it.
« Last Edit: June 04, 2019, 07:02:59 PM by sorveltaja »

Offline ddmckee54

  • Full Member
  • ****
  • Posts: 256
  • We're having fun now --- or so I've been told.
Re: Planetary gearbox - an attempt to make sense of how it works
« Reply #74 on: June 10, 2019, 06:12:53 PM »
sorveltaja:

The small skirt that you mentioned is sometimes referred to as an "elephant's foot".  This is caused by your print head being slightly closer to the print bed than was intended - "smooshing" the 1st layer filament wider than it should be.  You might for instance be intending to print a 0.3mm 1st layer, but the print head could actually be at 0.25mm.  I'll bet you're getting really great bed adhesion though aren't you.  The simplest thing to do is tweak the position of your Z axis limit, raising the print head a fraction of a mm.  You'll need to find the Z axis sweet spot where you're getting good 1st layer bed adhesion, and yet not getting any "elephant's feet"

As far as the top layer "pitting" is concerned, the fix for that might be as simple as increasing the number of top layers that are printed in you slicing setup.  In order to get a good solid top layer, I usually print a minimum of 3 top layers.  It sometimes takes a couple of layers just to close up the gaps before you'll get a solid infill.

Don
« Last Edit: June 10, 2019, 06:55:37 PM by ddmckee54 »