My homemade gear hobber

bazmak · Oct 8, 2018

Im totally lost,when I was at school I thought an Arduino was a faster than light travelling quark

Ross Donelly · Oct 8, 2018

Yes, just set the NTEETH value in the code to how ever many teeth you want. If you want to go to a different pitch, you would need to make a new hob (the cutting tool). The hob I made is for 1/8" tooth pitch.

rlukens · Oct 8, 2018

I'm just stupified. Got my ME degree 50 years ago. Wonder what the next 50 will bring.
Nice work though... I guess.

Corry · Jan 3, 2019

I'm mostly here to just say thanks.
I'm a small r&d company who needs what amounts to an individual as a bunch of gears (around 100) but a minuscule quantity to most gear shops. I'm looking at the cost of just stainless steel spur gears costing about $20/gear. I can buy a lot of tools for that money...perhaps ones that will make wear parts rather than wear parts themselves!

I have a pretty good CNC shop (5 axis mill, 4 axis lathe) so I thought gear hobbing would be a breeze. well, the lathe doesn't have "polygonal turning" capabilities, other than that, it would be the ideal place to hob! So what I'm thinking I'll do is buy a 2Kw servo motor/driver (or make the STMBL open source servo driver), gear it over to a "hobbing spindle", add a servo encoder to the lathe headstock, and the "hobbing spindle" (since I'll be gearing it up since all of the motors I have found are 3000 rpm max or less!) and use PID code to synchronize them.

PID isn't something I really understand though. Linear algebra didn't help, and my college requirement was linear algebra OR diffeq

I should've done both! I know what PID does, but not how to tune it or write it. Your code will certainly help a lot.

If I could make one request, it would be how you went about tuning? General process only. I *think* my non-controlled spindle will have less variance given the extreme amount of inertia it is carrying, as well as the quality of the control system, but who knows. I'm specifically buying a low inertia spindle motor to hopefully make corrections easy and fast, but again, who knows what the reality will be until I get it all done!

I think I can manage the rest.

Like I said, I mostly wanted to say thanks. Your post came up on a search for diy hobbing machine as I had started down the line of thinking of making my own machine for this! (well, evaluating cost/time/chance of success/etc)

Ross Donelly · Jan 3, 2019

Corry,

Glad it was helpful. I'm not the first to do this. I got the idea from some videos I found on the web. Just google gear hobber and you should find some. Here's one:

https://thewikihow.com/video_ZhICrb0Tbn4

I tuned the parameters by starting with just P, then adding some I to allow it "catch up" and synchronize, but not too much or you will overshoot. I also tried adding a little D to improve convergence but that can cause oscillation and I ended up not using any. So it's actually a PI controller, not a PID. I never learned the proper mathematical theory behind this but it's not too hard to understand how to tune a PID reasonably well in practice without the math. I found the wikipedia page on PID to be helpful:

https://en.wikipedia.org/wiki/PID_controller

Corry · Jan 3, 2019

The idea of using an Arduino to do the work is what got me thinking. In my mind, I was stuck on the idea of needing a mechanical linkage. I saw one design where a person used a flex shaft tied to a pulley on the spindle, into some more gears to drive an indexing head. I was really stuck on the mechanical synchronization. Even in considering how to make my own hobbing machine, I was thinking mechanical synchronization. This is far easier, and given they make 5000 ppr in quadrature servo encoders (and I'm guessing the likes of Mitutoyo have even higher precision encoders), its probably able to be more accurate (given all the backlash in the various gear trains needed for a hobbing machine).

I've since learned of polygonal turning (which is cool, but not really useful to me), and seen that CNC hobbing machines are doing something similar (though probably with FPGAs, and even fancier algorithms). Its really funny because professionally, I'm a software engineer. I should see my favorite hammer (software) and see nails everywhere!

Either way, I'm pretty excited for this. I hadn't started updating my designs for the gears, because I really didn't want to spend $2k+ on gears! If I couldn't find a way to make them, I was going to really try hard to design around them (probably would have done chains with sprockets made on the lathe, but chains are noisy....and my project is a jerk, err, has a lot of jerk....Its going to be hard enough on gears. Chain stretching would likely have become a major issue as well.

I'm now working on the design changes

I admit, despite having nice equipment, I'm still a relative newbie with it all. Despite that, if you ever have need of something made, just ask. I'll do my best. Trouble with CNC is that there are billions of ways to get the machine to destroy itself. Inventor HSM simulation isn't very good either, and vericut is rather expensive. (Still considering it anyhow), so that why I say, I'll try. To give you an idea of the capabilities, its a Haas VF-2 with a TR-160 or HRT-160, and a Haas ST-20Y. Pretty good work volumes for hobby/research/prototyping, but as far as "industry" is concerned, they're small machines