[cnczane]

Posting this because someone should have thought of it already!

A touch-probe made out of a recycled gel pen and its innards.

Like many of you, I've dreamed about building one of those Renishaw-like 3-point rods-on-balls probes, but it just seemed like too much machining. I wanted something simpler, and I noticed that my clicker pen seemed to have a lot of the necessaries.

Here's what I came up with: <trumpet fanfare><searchlights><fireworks><announcer>"The 2014 edition of the Penticle 5000!!!"<da-da-da-da-da-DA!>
(Pen Tickle)(5000 because its like 5000 times better than just a 1)

What you do: 1. break off a carbide tool to use as a collet mount (choose a diam. that closely fits as a replacement for the clicker in the upper barrel), 2. disassemble a gel pen (ball point might work, but gel has a nice fat cartridge to receive a folded paper clip for half the switch), 3. shorten top and bottom of the barrel and the cartridge (use an exhausted pen to reduce ink mess), 4. glue the pieces back to make a (very) shorty pen (contact cement works well), 5. slot the upper half of the cartridge (dremel: clamp in pliers!), 6. fold up a paper clip for the lower contact and stick it inside the cartridge to ride in the slot, 7. epoxy the shank into the top "at appropriate depth" (kneadable solid epoxy "clay" or "rope" is much easier than liquids), 8. assemble the pieces with the spring in the top next to the shank, 9. fold and wrap a second, stouter paper clip around the upper barrel to form the upper contact (use another pen as a mandrel and needle-nosed pliers as a vise), 10. epoxy upper contact to the upper barrel, 11. assemble with "small gap" between paper clips (hint: they bend), 12. attach "the right type of electrical probe connector" to the free ends of the two separated paper clips, 13. run a probing program (probe output is high until paper-clips contact, pulling it low, so parallel port input is ACTIVE LOW).

As assembled neither paper clip connects to anything so either can be ground or connected to +5V through a pull-up resistor. This means there's no current flowing (no waste) until contact is made, when the parport senses the high voltage is pulled low. (I already made up a probe cable like this with 2 alligator clips for different types of ACTIVE-LOW probes, one clip to ground and one to +5 through a pull-up; the first probe was a copper-pipe center- and Z-finder which also has a place to receive an alligator clip. I'd like to "vacuum-wrap" aluminum foil over the surface of something to be probed and use that as a touchplate too. Anyone ever see this done? 30 years or more ago industry was doing "magneforming.")

It does work, electrically.

How well? I don't know.

Accuracy? Precision? I don't know.

I have to write the LinuxCNC G-code (with its nifty IF-THENs and GOSUBs to allow clever probing), but I'm posting this now, so those of you for whom this prompts a better idea can get on with it!

Hope someone's inspired.

[cnczane]

A couple more views. (It's resting in clay to hold it at a particular angle.)

[cnczane]

Initial probing tests in LinuxCNC shows it does function well enough for my home-built machine's precision. If I ever get around to probing a St. Christopher medal, I'll post it.

[cnczane]

The probe connects to a cable with a male RCA connector on one end, and a pair of alligator clips on the other, with a red LED soldered between them.

The outside of the RCA connector is ground, the inside is routed to 5V through a pullup resistor and the side of the resistor opposite to the 5V connection is connected to an input pin on the parallel port (which is configured in the CNC controller software to be an ACTIVE-LOW probe input). [NB not all parallel port pins can be inputs, and the software has to be told whether the pin is ACTIVE-LOW or ACTIVE-HIGH.]

Because the LED is wired BETWEEN the connectors, it completes a circuit and is ON (if wired correctly: try it before you solder it to make sure it comes on). When the two clips are shorted together (because whatever conductors they were connected to came into contact), the LED goes OFF, because the current that was going through it now goes around it, and comes back ON when contact is broken. [NB you can SEE that contact was made when the LED goes OFF...]

Because the clips can be easily removed, different probes can be used. I've come up with three: a copper-ring center-and-depth finder, an aluminum-plate depth finder, and the current depth prober.

(To anyone who reads "circuit" and goes all shallow-breathing and glassy-eyed: the alligator clips (from Radio Shack) are "screw-on" and I've no doubt that the fine sales force there will be happy to sell you extra connectors so you--still--don't have to solder. The biggest fear is probably burning up something with "too much heat" but these are the simplest and cheapest components you'll find, and anyway you'll find they take more heat than you thought they could.)

[cnczane]

[ABANDONED]
Based on my experience this weekend, I recommend to Just Say 'No' to Z-axis probing, except for "planar" subjects.

I determined to probe, PART of, a 3D object, an area of about 7/8 in. sq. by about 1/2" deep, from a "back-off" height of 0.1" above the highest point.

I decided to probe at 0.005" because I reasoned that no matter what I do, it's going to take a heck of a long time so might as well see if it can DO 0.005"...

After two point three rows, my patience ran out. The nature of a Z-probe is that you cannot afford to NOT retract to the full height, without some programmed knowledge of the part you're scanning, because you cannot afford a collision trashing probe or part or both. My probe spent a lot of time in vertical travel, rapiding up, but probing-speed down.

I watched for a while, then decided watched pots never boil, left, and came back later with the question in my mind whether all the wear-and-tear on my home-built machine's homemade parts was really worth it for all the models I was going to get from scanning with it.

I could've let it run overnight--for several weeks--if I wanted to, but I concluded the guy who wrote smartprobe.ngc wrote it for a reason. (That's an example G-code file that comes with LinuxCNC.)

First of all, he uses a 3-axis probe that senses contact in XYandZ, and his algorithm first tries to move laterally, saving time by not retracting when it can, and retracting incrementally when it cannot.

That's never going to work for a Z-axis-only probe.

I apologize that I was unable to edit the first post to provide this final conclusion.

My two dot three lines of points indicate a respectable model would have been the result.

Elsewhere user handlewanker makes the case that building a second, lighter-weight and higher-accuracy machine dedicated to scanning is something to be considered if scanning is really going to be part of a shop's production workflow. (Having built one machine, and being now too old to suffer to build another, I believe I will put my hope in photography-to-3D-modeling technologies instead.)

Cheers! (Keep thinking, and never quit.)