Hello Guest it is October 17, 2019, 07:34:35 PM

Author Topic: CNC Gear Cutting  (Read 8165 times)

0 Members and 1 Guest are viewing this topic.

CNC Gear Cutting
« on: October 06, 2008, 07:16:15 PM »
I'm in the middle of converting one of my manual Carroll Dividing Heads to CNC when it occurred to me I may have a problem with resolution so I was hoping someone may be able to provide some insight.

My dividing head has a 20:1 reduction.  I was planning on mounting a Nema 23 stepper with 1.8 degree steps using a timing belt and pulleys with a 3:1 reduction for a final reduction of 60:1 at the spindal.  When I tried to calculate some of the common tooth cuts I would make I found that some would not fall nicely into the resolution of this setup.  While I could change the gearing which would help in some cases it would eliminate others from what I can tell.

For example, to cut a gear with 34 teeth using my existing 20:1 reduction dividing head I would use a dividing plate with 17 divisions and index 10 positions each time.  This would equate to 10.588235294117647058823529411765 degrees per tooth.

Trying to calculate the steps for the 60:1 reduction CNC dividing head I get 3529.4117647058823529411764705882 micro steps.  It's my understanding that I can not move the stepper the fractional part so I would basically have .4117647058823529411764705882 x 34 = 13.99 steps lost which would be 0.041999 degrees lost.  This would create 1 tooth that is either too fat or too thin if I'm not mistaken which isn't good for time keeping.

Am I totally missing something or do I need to take another approach?

Any feedback would be greatly appreciated.

Best Regards.

Carl
Re: CNC Gear Cutting
« Reply #1 on: October 06, 2008, 07:19:41 PM »
You might want to use a servo for that application...one with a high encoder count. 2000 PPR (8000 count) or more per rev.
RC

You would also want to move in absolute, not incrementally as to not accumulate the error.
« Last Edit: October 06, 2008, 07:37:07 PM by Overloaded »
Re: CNC Gear Cutting
« Reply #2 on: October 06, 2008, 07:33:05 PM »
without micro stepping, you would be at 600:1   x  20 = 12,000:1 total
With 1/10 micro step you would be at 6000:1   x  20   = 120,000:1 total
With the 2000 PPR servo you would be at 24,000:1 x 20 = 480,000:1 total and would count pulses exactly.  .0015 deg per count.
RC

« Last Edit: October 06, 2008, 07:41:37 PM by Overloaded »
Re: CNC Gear Cutting
« Reply #3 on: October 06, 2008, 07:45:55 PM »
Try to keep in mind, there's resolution, then there's accuracy.  They're two very different things, and one without the other is kinda useless.  Unless you have an extraordinarily accurate machine, worrying about resolution discrepancies of 0.04 degrees seems a little silly to me.  And the postiion error due to microstepping is probably no better than this.  In fact, it's very likely worse.

Regards,
Ray L.
Regards,
Ray L.

Offline RICH

*
  • *
  •  7,367 7,367
    • View Profile
Re: CNC Gear Cutting
« Reply #4 on: October 06, 2008, 10:46:50 PM »
Carl,
Just for kicks check out motorized positioning tables from Klinger, NRC, Newport. It will quickly give you an idea of what
you get from high end / high priced tables.
Additionally, Davidson Optronics has a nice conversion table relative to angular to linear conversion.
Just some info LOL

RICH
 
Re: CNC Gear Cutting
« Reply #5 on: October 07, 2008, 06:45:52 PM »
I think you guys got me back on track.  Thanks for the input.

If I'm calculating correctly, the max error at the tip of a 32DP gear with 34 teeth would be 0.000104.  So as previously stated, my machine can't come close to that type of accuracy.  As long as I use absolute mode, I should never see an error greater than that along with the accuracy of the machine itself.

How does Mach handle fractional steps that the driver/stepper can't accomodate?  Does it round or truncate?  For example, if I need to move 360/34 which is approximately 10.588235 degrees and the angular movement at the spindal per step is 0.003 then I would need 3,529.411176 steps.  Clear the driver and stepper can't do the fractional part of the movement so what will Mach do with the it?

Back to machining the mounting bracket for the dividing head.

Thanks again.

Best Regards.

Carl

Offline RICH

*
  • *
  •  7,367 7,367
    • View Profile
Re: CNC Gear Cutting
« Reply #6 on: October 07, 2008, 08:17:00 PM »
Carl,
I understand your need for accuracy. In fact, isn't it just amazing what you get for a $10 Timex!
Since your into gears I have attached my Excel spread sheet which caluculates all dimensions you will
need to draw a gear and was used to make some gear cutters I made. To use it just put the pitch and #
of gears, and pressure angle in and it will calculate the rest.

You can do all the calc's you want, but when all is said and done, it goes back to all the components of the
machine system, just like what Ray said. Every component of the system contributes to the accuracy of the
finished piece. A cutter won't run true away from the spindle within .0001" , lucky you get .001".
I think you should just try cutting some gears and see what your equipment will do. The accuracy
of the cut gear is defined by what it will be used for, and in a time piece isn't there some adustment.
Even measuring the accuracy of a cut gear is a challenge without the right equipment.
RICH
« Last Edit: October 07, 2008, 08:21:11 PM by RICH »
Re: CNC Gear Cutting
« Reply #7 on: October 07, 2008, 09:53:05 PM »
Thanks for the spreadsheet RICH.  That will come in handy.

I think my mill can handle the accuracy I need.  I've built one grandfather clock already and have made most of the parts for a second one.  What got me into the whole CNC arena was the desire to easily, quickly, and accurately, cosmetically that is, cross out the gears.  I did the first clock gear crossing by hand and don't care to do that again. 

I think thiis CNC stuff is like dope.  I now want to see if I can do the indexing too.  So it appears that it's feasabile but I will need to run some tests as you suggested before getting too far along.  Even if I can't, it will be nice to have around and besides I've already got the program complete for cutting the mounting plate for the 4th axis which is really neat.

I've also cut pinion wire on the mill which turned out nicely and just parted off the length I needed for each arbor.  I was using drill rod for the pinions so that's the hardest stuff I'll cut on the mill at this point.

Thanks again everyone.  This is a great site and Mach has made this all possible for the average hobbiest like me.

Best Regards.

Carl
Re: CNC Gear Cutting
« Reply #8 on: November 11, 2008, 05:56:06 PM »
Just finished setting up a 4th axis and it seems to be fairly accurate under no load just doing a full 360.  Will need to do some test gear cutting next to make sure I can get the type of accuracy and repeatability I need for gears.

I'm still a little puzzled by how slow it turns even when I set the velocity to 1000 in motor tuning.  I've tried changing the rotational diameter in Settings Alt6 but it still doesn't seem to affect things.  Not sure what I'm missing but will experiment more.

Here's some pics of the setup.  It's an old Caroll Dividing Head.  The mounting plate for the stepper was done with a CNC program I created which was my first real attempt at making something useful.  Kind of cool to see this all come together.  Need to learn more about how to make better finishes on these parts though.

Best Regards.

Carl

Offline docltf

*
  •  353 353
    • View Profile
Re: CNC Gear Cutting
« Reply #9 on: November 11, 2008, 06:11:06 PM »
carl
go to config/toolpath check box for use dia for feedrate.

bill