200ic - Correcting a Tool Frame by Jog amount ?

Hello,

Please forgive me if this question has been answered before, I've been searching for a while without success. I'm a software engineer but I have only superficial knowledge of the TP language, but enough to work my way around.

I'm looking for a quick way to calibrate our 200ic's tool offset after a tool replacement. They are 3d printed so they have slight variations, but we have pretty tight tolerances and it's a very complex station. We have over 200 points and I'd really like to avoid correcting each of them manually.

Basically what I'd like to do is to have some sort of calibration routine where I would manually jog into a known point, then subtract the position of the stored point and the manual position and apply the difference to the tool offset.

I've spoke to Fanuc Canada but they've been really unhelpful. The've sent me partial code that shows how to enable cartesian representation via $PR_CARTREP=0/1 but with the part in between missing. I figured I'd ask here instead of getting half-answers.

This is what they sent:

How should I proceed from there ?

Thanks!

Quote from Fabian Munoz

As far as I'm concerned you have a new tool. I dont know the shape of your tool but why cant you follow one of the processes that Fanuc offers to teach a tool frame ?

Hi Fabian,

To give a little background this robot was purchased 8 years ago, and my task was to develop the software that controls the station. Unfortunately the company that made the station and did all the robot side does not exist anymore. I ended up taking over the maintenance aspect and learned how to edit and touch-up the TP programs for our needs, but I'm a bit stumped as to what would be the best course of action.

The tool offsets are the same since the station was built and I have no idea if I need to use the same original location where those were originally set, or if I can do that at my desired target point.

To give an idea:

This is our loading station, and we have 120 parking station blocks just like it. As an example the tool needs to drop inside and in between the plates and then retract. As you can see the pitch, yaw and roll are very important. This is what I want to correct for all stations in one go.

Edit:
I've searched a bit more, and I don't think I could use the normal procedures (ie 3-point, 6-point) because of the tool shape and because it needs to be adjusted within the block. The current offset most likely was entered manually:

X: -28.000, Y: 0.000, z: 205.000, W, P and R are all 0.000

So I guess I do need to calculate the difference (in cartesian coords) between the teach point inside the block and the actual jogged position. That's the part i'm not sure how to do.

Is there no way to get the difference or do some arithmetic between two points natively ? Something like PR[n] - $CURPOS or something ?

Thanks!

Sorry, yes I meant the tool frame. I've used the wrong term. This is what I want to adjust:

I've updated the title accordingly.

Quote from pdl

The easiest way to update your frame would be through direct entry, adjusting it till the tool can fit into your fixture.

See, there's ~0.2-0.3mm space between the tool and the fixture's lamellae. I won't discuss the issues of that design here, my job is just get it to work. Someone found a way to jam the original stainless steel tool quite nicely in the past so the new one is made out of brittle plastic and designed to break in order to protect the arm if a collision happens. But with that small of a wiggle room, one false movement and the tool or lamellae breaks, and I then have to start over with a new one that has slightly different specs.

This is why I was hoping to avoid eyeballing and guessing values, and instead copy the fixture "middle" point coordinates from my original Load procedure into a new routine, offset it vertically, and then slowly jog into the correct position for that new tool and calculate the difference to apply to the tool frame.

While it works conceptually, calculating the difference using the actual POSN values does not work because of the coordinate representation is apparently not right for what I want to do, at least according to Fanuc Canada.

Am I grasping at straws ? I'm sure there is a way to do this or I'm missing something really simple. Punching the values into the SETUP Frames panel at the end is no problem - I'd just like to be able to do this basic arithmetic (Last column should be called New Tool Frame, wrong term again):

Again, sorry if am not using the proper terminology. I had to learn as I went and the available documentation is quite lacking.

If there is any other information I can provide please let me know.

Quote from Fabian Munoz

Just to have it clear the repeatability of your robot is +/- 0.02 mm.

Repeatability hasn't been an issue in the past.

Quote from Fabian Munoz

I understand modifying the tool coordinates will save you a lot of time

... And save me the grip of death/locked muscles having to hold that massive pendant, half-press the deadman, hold shift and jog buttons and squint for 50 hours

Quote from Fabian Munoz

you are assuming the Orig Pos in load Bay is correct How can yo prove it to me ?

The robot has over 40,000 running hours using this point. (I've touched it up a few times when the original tool got banged-up). I've come to rely on it when testing the tool integrity after a collision.

Quote from Fabian Munoz

did you try another part (or point) after you did your toll offset

I don't average it out but I usually do a couple repetitions to make sure there's no or minimal contact and then start a full test of all stations picks and drops and observe for a bit. Keep in mind though, in the past I touched-up the points themselves - not the tool frame, which is what I want to try now

Quote from Fabian Munoz

if -0.058, -0.339 and 0.788 are correct all the other 200 points should be perfect. [...] Is there any way that yo can put a bit of a chamfer on the tooling

Unfortunately it doesn't work. I'm applying a transform I got from one frame to the the separate tool frame so the coordinates don't quite line up, as pointed out by HawkME below

-----

Quote from HawkME

the spreadsheet doesn't work because you are using numbers from 2 completely different coordinate systems.

Yeah. That's what I'm trying to figure out how to solve. I thought there was a native or simple way instead of solving the kinematics.

Quote from HawkME

The best thing would be to design a calibration fixture that allows you do do a proper 6 point Tool frame.

I can't think of a way I could do that given that I need to line it up *within* the fixture and that it can only fit one way.

Would it be easier to achieve via a custom application ? I have the PCDK.

In the mean time I guess I'll have to eyeball values in direct entry for now and hope I don't misjudge. I'm still interested in suggestions and reading whatever info I can find on the side.

Thanks Fabien, HawkME

Quote from pdl

Is your tooling orthogonal to the robot's face plate? If so, you should be able to just make a point above your fixture's point (with a z user offset) and adjust the utool there without worrying about snapping off teeth from your tooling. Just adjust your utool until you can safely jog in and out of the fixture without binding.

This tool is, yes. This is how I've been doing so far, but carefully jogging it in place then -> touch up the actual point.

Quote from HawkME

If you go the 6 point TCP method then you teach the TCP first, then touch up all of your points to work with the new TCP. Then if the tool changes you should be able to redo the 6 point method and be good again. But, this assumes you have a good, repeatable process to teach the 6 point tcp.

I'll have to touch them all up anyway. The original tool was machined but hand-assembled with 13 parts and was not even square, or straight. However the technician that taught the original points treated them as being straight (making all of them wrong now that my tool is actually straight) and used linear movements relative to the robot center J1 pivot. It's basically like a stadium:

Worse, the 100 stations were epoxied by hand so some are uneven and slightly rotated, causing audible 'tinks' and friction. I solved the problem by copy-pasting and offsetting all of them to give me more control and compensate the entry/exit orientation.

I'll forego fixing the tool frame. I know my tool is accurate to ‎±0.1mm so I'll rely on that and just get the center point from the CAD model and proceed to fix the whole thing.

My plan was to *never* have to touch them up again as it's such a pain in the *** and hoped I could create a simple procedure that other employees could easily follow, as I might not always be working for this company. Fortunately, touching up the small variations by eye in the future after a tool replacement shouldn't be that difficult.

But for now that means I have to spend a lot of quality time with that damn thing.

Again, thanks for your replies!