ArchivesSyndicate This Blog |
Getting a handle on milling accuracySunday, April 12. 2009
In which your narrator gets a handle on the everyday accuracy of the total tool
chain from Art of Illusion to the milled product...
I've been milling HDPE regularly with Tommelise 2.0. Usually, I am designing parts to fit something made out of metal like a bolt, or a NEMA 17 mounting plate. In my more recent project, however, I have been designing parts that actually have to fit other milled parts. Right now I'm on a quixotic quest to design an extruder that uses a threaded nut instead of a threaded bolt or pinch wheel to pump the filament. Milling is more limited than printing because you can't mill something thicker than the active length of your end mill. With my 1.27 mm diameter fish-tailed, two flute end mill I can handle about 10 mm thick material. What that means as a practical matter is that if I want something thicker, I have to make it out of milled parts thinner than that. That can get tricky. It's not like this is an unusual problem in Reprap. The laser cut variations on Darwin have the same problem. There are a number of ways that you can access the problem. The way the Ponoko and BitsFromBytes Darwins hand it is to use nuts, bolts and spacers to assemble parts made of relatively thin plastic. I've done a bit of that and have been dismayed at how much you wind up paying for all of the nuts, bolts, washers, springs and lock washers you wind up paying for to make anything. A few months ago I knocked out a anti-backlash nut for a lead screw drive. 
It was a beautiful part, but the bloody hardware to hold the milled bits together wound up costing about ten times what the HDPE parts did. There is something kind of crazy about that. Since then I've been trying to design things that fit together with a minimum of fasteners. The particular problem that I've been trying to solve in the few minutes I've had away from my day job in the past few weeks has been how to connect the stubby little drive shaft on the Solarbotics GM-17 onto a spur gear to drive the extruder filament pump. 
The GM-17 was designed to be bolted onto something like a thin metal plate with a few holes drilled in it. HDPE designed for the same purpose tends to be as thick as the stubby little drive shaft of this very efficient little gearmotor. what that means is that you have to put a significant extention onto your spur gear and mill a big hole in your mounting plate around the drive shaft so that the extention can reach up through the mounting plate to engage the drive shaft. That's a pain in the neck, to say the least. The approach I've been taking is to design plugs of 10 mm HDPE that receive the drive shaft and fit into the gear itself. What I have designed looks a bit like this. 
When you do something like that, however, the tightness of the fit between the parts becomes important if you are to avoid a lot of wear. Because of that, I decided to see how close a fit I could get with such a plug using the same hexagonal prism in the open-source Art of Illusion 3D modeling app to make both the plug and the hole that it fits into. I used the same milling settings for both pieces. Here is what the fit looks like.  It appears that I get around 0.2-0.3 mm gap between the plug and the hole. Mind, I know I could design tighter fits by upping the nominal size of the plug just a touch. That sort of thing takes a bit of trial and error, though. Now I know how much of a gap that I get when I just go straight for an design solution. Trackbacks
Trackback specific URI for this entry
No Trackbacks
Comments
Display comments as
(Linear | Threaded)
Actually I have been able to mill deeper than the fluted part of the cutter as long as the shaft is longer.
For example I made my first GM3 shaft coupler by milling the top 10mm of the outline as a trench a bit wider than the bit. I then stepped out 0.1mm and milled the lower 10mm of the outline. You end up with a 0.1mm step in the object but in many cases that doesn't matter.
You could use that technique to make a gear and shaft coupler as one part, although the second shaft would not be possible or the socket for it.
http://hydraraptor.blogspot.com/2007/07/couplings.html
LOL! Yes, I know that I can use stepped trench milling to go much, much deeper than I am now. You've caught me out, though. The bit that I didn't mention was that I didn't buy stock any thicker than 10 mm because I'd already measured the end mill and ordered HDPE on that basis.
Something seems a bit fishy here -- is the hexagon still regular, just slightly small? You are doing, IIRC, a large number of passes to get your complete cut, meaning that it must be *repeatable*, or the edge would look all wuggly, at best, or you'd ram into the cuts you didn't quite make before, at worst.
Is it possible that the conversion between "N mm" and "run the motor for .3 seconds" needs tweaking at some level? For that matter, weren't you milling PCBs with far more accuracy then this?
"is the hexagon still regular, just slightly small?" They were sized to be the same size. What you are seeing is the difference between milling a hole and milling a plug. "You are doing, IIRC, a large number of passes to get your complete cut" That was a while back. These days I am using 1 mm cut depth, so there aren't a lot of cuts necessary. Repeatability is very good these days. "Is it possible that the conversion between "N mm" and "run the motor for .3 seconds"" You're remembering Tommelise 1.0, not 2.0. These days I'm using linear steppers, not gear motors to run the axes. "For that matter, weren't you milling PCBs with far more accuracy then this?" Same accuracy. I'm not trying to fit bits together with PCBs, though.
If you get a difference between hole and plug then it means the value you have for the cutter radius is slightly out.
That wouldn't surprise me in the slightest. I've really got to squelch my tightfisted Scots-Irish nature and buy a good digital caliper.
Hm, you're right re gearmotors vs steppers, of course -- but in that case it simply becomes "run the motor for 20 steps". Repeatability being high but accuracy being low seems like it should be something you solve, rather then live with, generally.
Re PCBs, if something was .2-.3mm off... oh. I was reading .2 as 2. Nevermind.
Nophead's experienced opinion seems to mesh with my wild guess -- I was going to ask if you were certain of your cutter's dia, but figured that'd be a pretty closely controlled parameter, listed on the datasheet of your cutter, and easily verified with a pair of cheap digital calipers. Whoops...
In the interests of completeness, I also misinterpreted the image, and was very impressed with the thin cut!
If there is any eccentricity then the cut can be bigger than the measured diameter.
IIRC I measured my finished objects and tweaked the cutter radius to match. In the end I was milling to an accuracy greater than the resolution of my calipers http://hydraraptor.blogspot.com/2007/06/bob-on.html
Given that tool wear and changing tool diameters are a fact of life for serious milling, sufficient that every CNC we have on campus can probe the geometry of the cutters and compensate at the machine, I'd not be terribly surprised if the diameter of the cutter weren't the tightest-controlled parameter in the world. It'd be a little surprising for it to be a big variation, but you can't get precision in the long term by making the cutter of a specific diameter and trusting it.
|
