[multiplex]

So i was brainstorming one night a couple months ago on how to build the ultimate edge finder for my cnc router. It had to be small, low cost and easy to interface.

and then in a moment of genius?, i thought - what about those little pointing devices found on IBM laptop keyboards, otherwise known as trackpoints.

essentially these devices are 2 axis (sometimes 3) strain gauges, with a on board micro controller to interface the analog signal into a mouse movement.

I made up a mount using an air cylinder and linear slide bearing, which allows me to bring the probe up and down about 2 inches.

On the end of the strain gauge, i made an adapter to attach a standard indicator probe

I stole the trackpoint from an old IBM laptop i had (you could buy used keyboards on ebay for under 10 bucks), and wired up a mouse cable to it. It also allows you to plug in your standard mouse - so you don't loose that capability. (just don't move your standard mouse while the edge find routine is running)

in the end, i'm pretty happy with the outcome. It seems to repeat within +/-.001" or so, and with some tweaking it be capable of less error. Sorry i'm lacking many pictures at the moment, but i did put up a quick video

general operation is.. get the probe close to the object (.030" or so). lower it. tell it which axis and direction to move in. it then moves in .001" increments until mouse movement is detected. zeros axis, and moves away from edge in opposite direction.

[ame]http://youtube.com/watch?v=1yN3G4da6VI[/ame]

feedback welcome

[Cartierusm]

very interesting, I don't think I quite get it but very creative. I used to have a laser center finder mounting next to my router and had a button with a vb script offset. So I would line up my x axis zero out then my y zero out and then hit the button and it would offset to the center of my router. But I realized after doing a test cut that I had to move the router mounts a little to make sure the router bit was square to the table. This made the laser slightly skewed making it line up at different points depending on the height of the z axis. Plus I found out that router collets are not that accurate. Mine has .003 run out. So now I just chuck up the laser, which it was made for, and find the general edge rotate the chuck until the runout is equodistant to the edge I'm finding. Works very well for now. I have a pretty good eye, and I now the Laser Center Finder manuf. says it's .001 accurate it does take a lot of eye squinting to make sure you're actually there, whick is frustrating. Like on my bridgeport manual mill I use an electronic edge finder, when it just touches the edge it lights up, a much more positive action than squinting at a little itty bitty laser. I always double check the electronic edge finder by backing off and redoing it and it's very accurate. Since I'm using wood on the CNC I can't use it becasue it's based on touching metal....actually I just thought I get it now your device does the same thing but using motion instead of completeing a circuit to find the edge...i like that.

[jdell42]

Very nice... can we see the macro?

[multiplex]

thanks guys.

i'll post the macro up in a day or so. It needs some fine tuning. last night was the first test run.

but i need to have it in production by the weekend!

on these items i'm making at the moment, they require 2 setups and require finding the center from four sides. Thats 8 edge finds x 300 units = 2,400 edge finds!

so far i've been doing it by hand using the edge of the cutter and my eye and its killing me

[jdell42]

I like this concept.

I am thinking of a way to incoporate this into the Z zero to have an all in one edge finder and z zero.

I currently use a touch off plate for both the z zero and edge finding using the bit and some metal angle.. as completely accurate edge finding is not too important to me. But this seems to be straight foward enought to give it a go.

[multiplex]

there may be a way to include z touch off, but i haven't looked into too much yet.

the trackpoint microcontroller has an input for z axis. google TPM754A for the datasheet, which includes the schematic for the trackpoint circuit.

from the looks of it, x and y are fed into the z input. and this may also be where different versions of the trackpoint have different sensor configurations.

on the driver side, if you tap the trackpoint it can interpret it as a mouse click so you might be able to use that somehow.

what originally drove me to this concept was the strain gauge. to take it to the next level, it might be nice to intercept the analog signals before it hits the microcontroller and processes them directly with and ADC. The circuit board already includes op amps to prepare the signal! that way you could create a much more accurate system by not using the mouse driver.

but the existing mouse interface is some of the beauty

[mcwilsonmfg]

I like the way you think, anytime you can automate thats cool. for those who are straining your eyes...use a different edge finder...very simple, look at this one we use it everyday in our machine shop. have used the fancy electronic lights and find that these work best over 22 yrs.

http://www1.mscdirect.com/CGI/NNSRIT...MT4NO=36565124

just load it in the spindle at @1000 rpm and creep over to the edge and you can see it run off center very easily.

mark

[mcwilsonmfg]

I'm very intigued by your idea, "don't let my last post fool you". Actually I would be willing to work on an idea with you to incorporate your "edge finder" into possibly a low cost coordianted measuring device or parts checking on the machine. It has been a while since I wrote any macros, but I used to do quite a bit with the reneshaw probing system on 4 and 5 axis machining centers.

[multiplex]

sure i'd be interested in some ideas / help in taking it to the next level.

did some more work on it tonight, and its pretty much ready to do what i need it to (for now)

here is a little video

touch two sides of a slot, and move the router into it

[ame]http://youtube.com/watch?v=3q58oVZCmB0[/ame]

[evilgenuis]

now that is just too cool!!!

[joecnc2006]

Here is one done on a shopbot.

[mcwilsonmfg]

joe,

very nice, only thing is, it looks like you are working with a grounding clamp system that has to complete an electrical cicuit to find the edge...what multiplex has going on is a probing setup that gages off presure...any type of material can be gaged...wood,plastic...

[mcwilsonmfg]

sure i'd be interested in some ideas / help in taking it to the next level.

did some more work on it tonight, and its pretty much ready to do what i need it to (for now)

here is a little video

touch two sides of a slot, and move the router into it

http://youtube.com/watch?v=3q58oVZCmB0

Multiplex,
I need to find the hardware and software you are using for this project, then I can get into it with you. I've always run high end industrial cnc machines. So I'm not familiar with your control, new to cnczone and here is the first I've heard of "Mach3"... can you send me a breakdown list of what I need to make up this control? and the touch probe? I don't know if we're allowed to put our e-mail on here or not. But you can find me on the web with my user named o t c o m.

[jdell42]

joe,

very nice, only thing is, it looks like you are working with a grounding clamp system that has to complete an electrical cicuit to find the edge...what multiplex has going on is a probing setup that gages off presure...any type of material can be gaged...wood,plastic...

As mentioned in the previous post I use the same concept using a peice if angle butted up against the wood with my Z touch off plate on the angle.. that allows the circuit to be completed.. I use the same macro and touch offt plate to do x, y, and z. For the x and y you need to compensate for the bit diameter in the macro. This setup works fine if you dont need perfect edge finding because the bit may not touch exactly on a flute giving you "good enough" edge finding.

I got an idea in my head of how to make multiplex's concept work in Z.. but I need to flush it out a bit.

I really like not haveing to attach a ground clip or worry about tool diamerter for this approach.

[multiplex]

yeah the nice thing about this probe is there is no changing out bits for a probe, and its not dependent on cutter diameter.

Z could really depend on the capability of the trackpoint device. If it can do 'tap to click' then i would assume its pretty straight forward to grab the event. older ones may not have this capability.

i'm using a windows dll to access the mouse info from mach using the 'cypress express' vb scripting.

i'll try to document what i've used, with links and pictures a little later this weekend.

for now check out this PDF for details on the trackpoint circuit. Page 9 shows a diagram

http://www.nxp.com/acrobat_download/.../TPM754A_3.pdf

[mcwilsonmfg]

Multiplex,

I downloaded the data sheet. I will look them over. Since you already deciphered the usable pins from the list I should waste time sorting it again, or is there something you need me to help you find on there?

[mcwilsonmfg]

I ordered my keyboard with the mouse in it today...two extra ones just in case I mess something up, have been know to make a blunder once in a while...LOL

[multiplex]

Cool! I just received a couple extra too.

And started to take some pictures of the process - soldering on the wires, etc.

i'll try to get them posted asap.

I had a little oops the other day - my air valve which controls the probe up / down got stuck while it was in the down position. needless to say mach 3 thought the probe was up and it crashed the probe right into the workpiece. Hate when things like that happen (i was in the middle of a big production job too). So i will be adding a limit switch to ensure the probe is out of the way. Thats what i get for taking an air valve from work that was in the junk box - should have just bought a new one.

[mcwilsonmfg]

cool...looking for the pics. I just got my three keypads. are you removing the board that is attached to the red spongy piece in the middle of the keypad?

[multiplex]

Ok finally got some pics uploaded

Here it the victim - T23 Keyboard



On the back you will find this. These screws are a pain to get off. Usually one out of four strips, but i just drill out the screws



Once you get it off, unsolder (or rip off) the orange plastic wiring. You can then solder on your wires for the mouse cables. I recommend lots of hot melt glue over the wiring to prevent them from breaking. There will be two sets of three wires. First set is the mouse out to the computer. The other is mouse in (lets you plug another mouse inline).

I'll call the blue wire Pin 2.

Pin 1. Reset - not used
Pin 2. External Mouse Data
Pin 3. TrackPoint Data
Pin 4. External Mouse CLK
Pin 5. Ground
Pin 6. 5+
Pin 7. TrackPointCLK
Pin 8. Left Button
Pin 9. Right Mouse Button
Pin 10. Middle Mouse Button

I'm sure you could come up with some ideas for using the buttons as inputs =)

A standard PS/2 has the following pins

Pin 1. Data
Pin 2. N/A
Pin 3. Ground
Pin 4. 5+
Pin 5. Clock
Pin 6. N/A

So at this point, you just need to connect the dots.

I had a funky mouse cable sitting around that had both male and female ends, kinda like a splitter. Thats what i used to wire up my board. However you could probably buy a male to female extension cable, cut it in half and wire it up that way



Here is the top side, where you can see the strain sensor. For whatever reason IBM has a plastic cap that fits over the sensor. I made my probe to fit this cap. Then the cap goes onto the sensor. I super glued them all together so it was rigid