"Egyptian pyramid style" CNC (software telling me how to turn handles)

Rich, your site looks very interesting. You probably saved a major amount by doing this on a DIY basis. Right now I am looking at Mach3 website.

i

This is how John Parsons did it in 1946 when he was machining templates for airfoil sections along a helicoptor blades.

He wasn't doing circles he was cutting formula curves. He did it on a Bridgeport and had two helpers. An "X" helper and "Y" helper. :- )

Regards,

Stan-

Very nice to know. I did this thing past midnight yesterday, so I did not have much time. If I get time today, I will try to do it somehow better, sitting comfortably between the X and Y handles, allowing easier control.

i

A rotary table would be nice.

John

Just cut it like a cam. Set X and Y, cut with Z. File finish. For real accuracy, put a carbide insert boring tool in a collet for the finish pass. Find the cut diameter of the bar with a test cut.

Or a DRO with a bolt circle function. Or a boring head. But it's Iggy. Lots a blabla and no use.

Nick

A circle is the simplest curve, which can accidentally be milled on a rotary table. The "mill by turning dial as the script says" approach works for circles, ovals, sinusoid, lettering, or any other curves, whereas a rotary table does not.

I am not at all trying to diminish importance of rotary tables, just to say that there is more flexibility in the described milling method than in just using a rotary table.

i

It's a royal pain and very inaccurate to drill holes to get a round/oval/whatever shape (only works halfway on the inside). It's much better to clamp the part onto a RT, have the RT's axis along the X-axis and feed by the Z-axis (and mill facets with the Y-axis). This way you get polygons and not a toothed-looking surface that needs filing. Of course this doesn't work with all surfaces. But cams (for example for IC-engines) are easily done that way. Manual milling of harmonic cams -with some halfway clever software in the DRO- this way takes about 15 minutes per lobe. Accuracy +/- 1/100 mm. BTDT.

Nick

Pretty impressive, I would never have the patience to move the controls in such increments. I never wrote Pear script, nor have I ever heard of it until now, but I could put a similar thing together in excel in about

1/2 hour. Maybe I will give it a shot just to see if I have that patience. You have 45 steps in your code so Im assuming your running in 8° increments. I would have thought that would have been too course.

A question for the group is how do CNC machines generally account for backlash. This should be accounted for before hand and place into the code. I guess I could find this experimentally by running a test of consisting of plunging an end mill into a piece of steal and running it

1/4" one way then 1/2" the other way. Subtracting the length of this cut from the tool diameter + the 1/2" would give me the peak to peak backlash.

John R> [for information of a.m.c. readers, I recently bought a Bridgeport

How would one use a rotary table to mill an oval? I think the above use was complex curves. Not saying it's true, it's just what I heard/read.

Thank You, Randy

Remove 333 from email address to reply.

Thanks. It is, more or less, my first milling experience. (I installed this mill in my garage a few days ago).

It is not that boring to move the handles. The most time consuming and unpleasant part was finding the next line on the laptop output. (every time I would turn to handles, and then turn my eye back to the laptop, I would have to locate the line again).

Yep.

No, the step is 2 degrees. I traced 180 pairs of coordinates.

I am maybe the least knowledgeable person about this, but, to me, it is not possible to avoid the need for digital feedback (reading position into the computer), unless you want to go to very expensive stuff like precision ballscrews etc. Manually, I dealt with backlash by locking table, carefully turning dials, etc.

i

"jroncallo roncallo @snet.net>"

According to j :

Perl is a language which got its start on unix, but has been ported to many other systems by now.

The name has two expansions, depending on which end of the original man page you believe. :-)

At the beginning, the author (Larry Wall) said that perl stood for "Practical Extraction and Report Language", but at the end, in the traditional "Bugs" section of the man page (it used to be in all man pages in unix) he said "Don't tell anyone that I said this, but perl really stands for 'Pathologically Eclectic Rubbish Lister'" -- eclectic because he included in the language every feature which he found useful in all of the other computer languages he used, including many of the unix shells, sed (stream editor), awk (just the initials of the three authors), and even BASIC,

And, it has grown, adding libraries to accomplish some really amazing things through the years.

O'Reily has published a manual for perl, as they have for many other languages, utilities, and other features of computers, unix and MS-DOS/Windows. They make a practice of putting very good drawings of different animals on the front covers of their manuals, and the animal chosen for perl is the camel, which has often been described as "a horse designed by a committee". :-)

Actually, he was using 2 degree steps, as you could tell by reading the comments in his code, if not the code itself.

Real CNC machines use ball screws, which have almost no backlash (down in the ten-thousandths of an inch). For homemade CNC machines some of which still use the original Acme threaded leadscrews, there is not a single backlash, but a spectrum of backlash, because the greatest wear occurs in the middle of the leadscrew (where the most work is done), and out at the ends of the leadscrew, there is practically no more backlash than when the leadscrew was new.

Note that using ball screws in a manual machine is a serious hazard, because the screws have so little friction that the cutting forces can easily drive the table, and climb milling can result in the table slamming to the end of travel, or breaking off the milling cutter rather quickly.

This is not a problem with a pure CNC machine, as the servo motors (or stepper motors in the less elaborate ones) do an excellent job of holding all the leadscrews in position against cutting forces.

Yes -- but only at one place along the leadscrew. As mentioned above, if you are not using ball screws, you will need to take multiple measurements at different places along the leadscrew.

And even so, backlash can cause unexpected table motion when the workpiece and the table on which it is mounted are driven by the cutter, at least far enough to take out the slack that the backlash offers.

[ ... ] [ ... ]

Here is the line with the comment. The comment starts with the '#' character. Your guess at 8 degrees was because you were expecting a full circle, but in reality he was only printing out the steps (in this version) for a quarter circle -- thus eliminating the risk of the influence of backlash during that one quarter circle.

Enjoy, DoN.

This is how the concept of NC was first tested. I don't know that anyone actually used it to make real parts. But I have a book somewhere in my library that details who exactly did this, when, and where.

Jon

While you are at it, program a metronome function, so that you can enter feed per tooth, # cutters, and get beat to syncronize the turning of the handles.

Wes

On Sat, 16 Jun 2007 11:27:17 -0400, with neither quill nor qualm, Wes quickly quoth:

Yeah, and program in some nice, quiet beats like "Vital Transformation" by John McLaughlin's Mahavishnu Orchestra on the _Inner Mounting Flame_ album.

- Metaphors Be With You -

As long as it isn't disco....