Page 2 of 3

Re: parts with arcs grow

PostPosted: Thu Jan 10, 2019 10:43 pm
by beefy
Hi Balazs,

not my thread or my problem but just want to say a little thank you for that information. I've been haunting the plasma related forums for a few years now and never heard about that rack and pinion issue.

One day when time and money allow I want to build my "ultimate" plasma table and that new knowledge could save me some headaches and frustration :)

Keith.

Re: parts with arcs grow

PostPosted: Fri Jan 11, 2019 12:24 pm
by ger21
cncdrive wrote:Rack and pinions drive is therefor not very good for precise works. 2 pinions can improve it though. And also helical racks are better than straight tooth ones.


They can be. But the spring loaded drive systems are designed to be a low cost, low backlash option, that's easy to implement.

I'm building a machine using helical racks and planetary gearboxes. The main disadvantage is that you need to machine the rack mounting surfaces so the rack is parallel to the linear rails. But with low cost chinese gearboxes readily available, you can get a much better performing system for only a little more money.

Re: parts with arcs grow

PostPosted: Fri Jan 11, 2019 1:19 pm
by cncdrive
Hi Gerry,

OK, let's agree that it is hard to design a rack and pinion system very precise. :)

Chinese gearboxes are usually far out of tolerances. We have bought some samples some time ago from different companies.
All were rated to 15 arc minutes max. backlash and in reality they had over over 1 full degree backlash.
Good quality gearboxes are expensive. There is e.g. Neugart, but what we have used on the BF50 machine costs an arm and a leg. If I recall it costs 1200 Euros, but it is really precise and remains precise in long terms too.

Re: parts with arcs grow

PostPosted: Fri Jan 11, 2019 2:09 pm
by ger21
OK, let's agree that it is hard to design a rack and pinion system very precise.


Yes, unless you spend a LOT of money.
I got some pretty nice Chinese gearboxes, with supposedly <8arcmin of backlash. They were about $125 each.
I've also got some zero backlash planetary gearboxes made by Harmonic Drive. I paid $250 each for them used, but yes, they cost well over $1000 new.

Re: parts with arcs grow

PostPosted: Fri Jan 11, 2019 3:14 pm
by cncdrive
Yes, the issue is the "supposedly". :)
We also got <15arcmin gearboxes from several chinese suppliers, we bought some samples to test.
We tested it on rated torque on test-bench and the backlash was >1 degree. And after some runs on rated torque it became 2-3 degrees.
We bought also some from Neugart (German company) and did the same tests and they were and remained in the defined tolerances.
So, our conclusion was that chinese don't really know how to make precise gearboxes.

Harmonic drives are good, I mean the original ones, but they also cost an arm and a leg, so they are costy unless you can get some second hand ones from a seller who do not know who much those really worth. :)
The size needed for our BF50 Z-axis a harmonic drive in that size costs about $4000.

Re: parts with arcs grow

PostPosted: Fri Jan 11, 2019 3:15 pm
by cncdrive
And the gearboxes we got from the chinese looked also really nice and shinny and professional, but they were not precise at all, they just looked nice, but did not work nice.

Re: parts with arcs grow

PostPosted: Fri Jan 11, 2019 3:18 pm
by cncdrive
Hi Keith,

I'm glad if I could save you some headaches and money.
We have payed our "student exchange" on this, so I thought to share our findings and experiences, so others will not have to pay the same.

Re: parts with arcs grow

PostPosted: Sun Jan 13, 2019 8:52 pm
by balbs
Update. I went through and replaced the spring with solid bolts while applying pressure to force the gear into the rack and tightening the bolts. There is no movement of the gear away from the rack now. I went through and ran the program again and the parts are exactly the same size as before (over sized) I went through and moved the machine and measured with a tape measure and its movement is precise. I can adjust the steps per in based off the part and it comes out to being around 999 steps per in which is 12ish steps off from what it was calculated out to be. Doesn't quick make sense to me. Anything else you'd recommend? Its used the mx4660 driver but I also tested it (albiet awhile ago) with my spare tb6600hg 4 axis board and a uc100 and the results were i believe the same. I also cleaned off the gear rack and applied some lubricant to them to ensure smooth movement. With the waterjet head off the machine runs through the program smoothly. I also halved my accel from 30 to 15 and dropped my rapid speed from 500 to 350 during this test.

Re: parts with arcs grow

PostPosted: Sun Jan 13, 2019 9:15 pm
by cncdrive
Interesting. I'm a bit out of ideas.
And you sure that your steps per values are correct? Did you make soem test measurements?

Re: parts with arcs grow

PostPosted: Mon Jan 14, 2019 1:56 am
by balbs
I am fairly certain that the steps are correct, using a tape measure the machine moves accurately. The parts are consistently off so it doesn't seem like it could be missing or gain steps. I did import the dxf and scaled the length (y value) up to the value that it was actually cutting and then measured the width (x value) and it matched so its scaling it up evenly at least. Going to calculate this scale factor and see if I can just make a scaling adjustment in my CAM software to compensate for this error. I am also going to try and grab an old desktop with a parallel port and try it with a spare tb6600hg board i have using the old fashion way and see if that makes any difference. Really appreciate the help and insight. This has baffled me for quite some time.