Help regarding internal gear

[propclock]

I used Gearotic to cnc machine My Elmers gear engine internal gear. I made it 2x bigger than the plans
because I felt it was too small and the endmill required was also too small. I made it out of Delrin.
I know you are looking for a single tooth profile, Gearotic does this and the earlier version of Gearotic
generated code for producing the cutter. I am not sure if the free version will output
the dxf profile but you can try. The software does provide single tooth data.
At the time, many years ago the Gearotic licence was less than the Boston gear price for this gear ,
I purchased the licence and have used it many times,
including elliptical and helical gears. And the free version of Gearotic
is just fun to play with. Just my 1.414 cents worth.

 

[TSutrina]

I when through Gearotic tutorials. Clearly this is a professional program and can take care of every issue. I was not clear if the two spur gear model that were made with the minimum input actually inserted clearance. Note typically the center distance is increased by some fraction of the needed clearance. The clearances of the bearing may result in a gear tooth clearance. The image shown on the screen did not show any clearance. Tooth thickness I think is the way clearance is added which is a box that is automatically filled but can be adjusted. The imaginary gear capacity actually makes it possible to make gears that can not be back driven or only driven in one direction.

 

[petertha]

Interesting. I looked up Gearotic & perused their forum a bit. This post is applicable to discussion.
http://gearotic.com/ESW/FavIcons/index.php?topic=1197.0

I'm actually kind of surprised there aren't more homebrew gear apps out there written by hobbyists. But like TSutrina says, it may well not be a button click solution depending on the actual application. Some engineering adjustments may still be required.

 

[Jasonb]

I tried the Gear DXF and all it does is produce a cut out of a spur gear. You can see this where the blue spur gear fits exactly into the grey external gear.

Also pictured is the 24T internal gear I use for the post I linked to with a 24T spur gear placed into it, you can see the difference in profiles. This was so that a 12T could rotate within

 

[awake]

I don't know if it would be of interest or help in this situation, but ... I have experimented with modeling the results of semi-hobbing a gear using a couple of open-source programs - BRLCad and OpenSCAD. Both are programmed using a scripting mechanism, but of course they use completely different scripting languages.

By semi-hobbing, I mean cutting the gear in a series of discrete passes, maybe 4 or 5 per tooth, rather than the "infinite" number of passes that occurs in the continuous action of true hobbing.

BRLCad is difficult to use, but can produce a very high-resolution result (not a triangulated mesh), though if tooth count is high or number of passes is greater than 3 or 4, it will not be quick. OpenSCAD is much easier to use, but you have to choose between very low-resolution results that are slow, or very high resolution results that are utterly, painfully, maybe even impossibly slow. Any resolution of results on OpenSCAD will be in a triangulated mesh format.

Important caveat: My experiments have only been with generating external spur gears, not internal gears. With BRLCad, you would have to have the program first generate the external gear, and then in the same script use that gear to generate the internal gear. With OpenSCAD, it should be possible to import a gear in .stl format (e.g., generated by FreeCAD), extrude it, and use it to "hob" the internal gear.

Note that with either program, you can save an enormous amount of time by generating only a few teeth, just enough to be sure that you have at least one or two fully-formed teeth / spaces.

If anyone is interested, I'll try to dig up either or both of my experiments; just let me know which one(s) you'd like to see.

 

[awake]

BRLCad is difficult to use, but can produce a very high-resolution result (not a triangulated mesh), though if tooth count is high or number of passes is greater than 3 or 4, it will not be quick. OpenSCAD is much easier to use, but you have to choose between very low-resolution results that are slow, or very high resolution results that are utterly, painfully, maybe even impossibly slow. Any resolution of results on OpenSCAD will be in a triangulated mesh format.

Okay, let me adjust that paragraph. I just went back and tried the OpenSCAD version again. It actually is pretty quick to generate a gear, even with a high number of passes (I tried as many as 25) ... as long as you are only generating the "preview" version. When you generate the "render" version is where it slows down drastically, and may be too slow to be practical - again depending on number of teeth, number of passes, etc.

As a test, I've been generating a module 1, 20°, 20-tooth gear while I've been typing this follow up message, using 8 passes. (I.e., the "hob" cuts 8 times per tooth profile, rotating the gear 1/8 of a tooth and moving the hob as if in mesh for each cut. Not sure if that makes sense.) For those who know OpenSCAD, I've had $fn = 256, so the triangles in the mesh should be very tiny.

After several minutes, as I have typed this and done some other things, it is still not complete, and the cooling fan has kicked up to high, meaning the CPU is running flat-out on my i5 laptop. I'll wait until it is done to post this message, and I'll include the .stl of the results.

Okay, it is done at last - took 30 minutes and 25 seconds on my i5 laptop with 16 GB of RAM. Attached are two screen shots from OpenSCAD, one from the top and one from an angle; each of these will let you see the tiny "jaggies" in the corners and the tiny faceting on the tooth faces that is the result of the "semi-hobbing." I've also attached the .stl file. Definitely not a perfectly formed gear, but it would be more than good enough to use as a template for grinding a cutter. Again, this is just the external gear ...

 

[awake]

One more post on this topic: I've attached the OpenSCAD program that generated the above to this post. (I had to zip it since the forum software didn't want to upload a file with the .scad suffix.)

For anyone who wants to play with the OpenSCAD program: I didn't put a proper license into it, but just a quick note that I retain the copyright but make it available to be freely copied or modified so long as credit is given to the original. You will see that there is a module, hobbed_gear, which is intended to be the one called by the user. (Other modules help generate the "hob" and carry out the "hobbing.") There are comments to indicate what each of the parameters of this module do. To generate the .stl file and screen shots attached, I used the following:

$fn = 256;
hobbed_gear(m=1, pa=20, n=20, passes=8);

The way the program works: It determines whether the user has entered a diametral pitch or a module gear, and calculates the size of the blank accordingly for the number of teeth specified. It generates a "hob" consisting of a straight rack with teeth of the appropriate size and shape for the specified dp/module and pressure angle. It then positions the blank and the rack at the appropriate distance and "cuts" the blank using the rack. Then it rotates the blank, adjusts the position of the rack (for incremental passes / cuts), and takes another cut. And so on until it finishes.

I'll play with creating a similar program that will take a previously generated gear in .stl format, using it to "hob" an internal gear. May take me a few days ... and if anyone else wants to give it a whirl, have at it with my blessings!

 

[lohring]

Fusion 360 has alot of gear generation programs available. My free version came with a spur gear generator, but lots of others are available in the app store.

Lohring Miller

 

[awake]

I should have offered a disclaimer up front: the programs I am talking about are NOT the best / easiest way to get a good model of an external gear - much easier to use the gear generation built into various CAD programs. But as noted above, those programs may not generate a correct internal gear. And some may generate a less-than-ideal model even of an external gear. FreeCAD, for example, has a module to generate both internal and external gears, but the external gear is not quite involute, and the internal gear is like the one that jasonb showed above - just a cutout of an internal gear, with none of the clearance needed for actual use.

The point of the BRLCad and OpenSCAD programs I reference above is less about modeling a gear, and more about modeling the process of approximating a gear through "semi-hobbing." But it does have the advantage that, with a high number of passes, the result really does approach very closely to the theoretical ideal.

 

[TSutrina]

You may try drawing and calculating software offered by gear manufacture. Manufactures get expert gear designers involved so the likelihood of getting usable gears is higher then the coders that make the gear design listed above except for Gearotic. You need to signup with them. I didn't do that since I am not designing gears right now. https://khkgears.net/new/gear_calculator.html Go to the bottom.

 

[petertha]

For the price of 'logging in', the KHK calculator utility seems like a very good solution IMO.

 

[ninefinger]

I created a log in for the KHK and it does generate dxf outputs for taking into your cad program. As can be seen the internal gear tooth profile gets modified at the root and the ID is smaller as well, the left side was the original gear as drawn by the add-in to Fusion 360, and the right side is the gear as modified by the output from the KHK program, using all the default values (0.5M 72 tooth, 36T mating small gear). The line drawing is from the KHK program and shows the meshing of the 36T with the 72T internal gear.


What I'd like to know is how expert gear designers decide what parameters need to be modified for a given set of gears? I realize this is beyond what Don was looking for but if a pointer can be given to a resource that helps explain why a person would vary certain parameters it would be appreciated. I'm in the middle of re-drawing a set of 48DP gears to use 0.5M and beyond adjusting the center distances am wondering what else to consider.
Mike

 

[Don Pittman]

So I soon realized that I bit off much more than I could handle with this internal gear thing, the level of expertise and understanding concepts are still a bit beyond me.
However I still wanted to try so I did what I do best...I flew by the seat of my pants and tried "poking" out an internal gear with a broaching/slotting device on my lathe.
I had to do the tool point grinding and fitting mostly under a microscope because module 0.5 teeth are pretty small. To my untrained eye it appears the fit might be passable. The external gear was cut with an involute cutter and I believe it should be pretty close to correct. Anyhow it was a learning experience!

 

[Ken I]

I typically generate my own profiles - starting from a rack of whatever module or DP required - this also allows me to play with pressure angles and clearance radii etc.
vis:-

You draw the rack - array one flank linearly in pitches corresponding to 1, 2 or 3 degrees of rotation (depending on how fine you wish to resolve it) and then rotate each flank into position - then draw a best fit radius or series of rads. Mirror profile for the opposite flank. (This emulates the way a Fellows gear shaper generates gear profiles.)
In the case of the epicyclic, you have to generate the pinion tooth profile from a rack and then use that developed profile - arrayed about its centre and rotated back on the internal gear centre. (This emulates the way a Maag gear shaper generates gear profiles.)
Add clearance (if required - generally I don't) - use data *.dwg or *.dxf for EDM wire erode or waterjet (minimum 2 Module because of waterjet kerf)
Sometimes force the rad to the nearest standard cutter if you plan on making a cutter - then regenerate the matching gear to suit. This is very useful for avoiding interference and undercutting.
A PITB to be sure but I find it gratifying to develop my own profiles.
You can also use this method to develop non-standard profiles or non standard centreline distances - such as ninefinger is asking about.

Example - waterjet cut M2 synchronising gears on my radius gauge.
Regards, Ken

 

[TSutrina]

I had the opportunity to have experts design the gears for the few times I needed them. As an aerospace company we pushed gear and bearing hard. Purchase many of them so the manufactures were happy to present seminars which I did attend. Basic understanding is that a gears do not move if they are made exactly to the volute path and at the calculated centers. (Sand blasted the gears to create clearance.) So the approach to get clearance is to offset the profile, thin the gear. That is one parameter on a gear design chart. If the centers can be move then this to will create clearance but increase the contact angle. Thinning is the only method possible for planetary gears since the centers are fixed. The other problems of undercutting and interference is determined by math. Those equations are available or compute models will show the problem. Then the discussion that the experts have put into articles will instruct the solution direction. I have named one such expert above. General solution is to change the angle of contact or change the height or depth of a gear off the circle at the contact angle. Under cut gears require increasing the contact angle. DIY builders have the advantage of being able to make new tools where manufactures adjust their machine because they need to use existing tooling.