G68.2 *basic* explanation

[Newbeeee™]

That would be a great idea - G code + practical explanation with a mandatory picture

I'd buy it - could be a good little earner for Jay or Chris or one of the guys who teach here.

[fat_robot]

Well I've worked with both MAM72 and the new MX-520.

 

XYZ of the G68.2 line is new origin and centre of the rotation.

 

IJK is Euler rotation angles. I rotates around the untransformed Z axis, value can be +/- 180 ( or 0-359.999 depending on machine ) J is then rotation around the new X axis, here the value can be from 0 and up to endstop of A( B ) axis. Finally K is rotation around the new Z axis to allow the user to position the X axis in the wanted direction.

 

Always use G53.1 after the G68.2 line.

 

You'll also need to apply G43 after G53.1 and cancel it again ( G49 ) before G69

 

As already mentioned Siemens CYCLE800 can do exactly the same. And so will Heidenhain CYCL DEF 7 + CYCL DEF19 ( PLANE )

Hi everyone. I'm from Russia so dont blame me for bad english.

I work now on 5-axis machining centre with rotate table and trying to understand how to rotate workpiece coordinate system. The problem is I cant clarify for myself how it works. For example on machining centre with rotate head workpiece stands steel so it is easy to rotate coordinate system around axis. But in my particular case workpiece should be rotated around A and C axis so distance between machine reference point and A and C centers should be considered. I can make app for android or program for fanuc to calculate different coordinate systems using machine rotate centre but I suppose there should be easy way.

Hope someone will answer me.

[scooke4073]

Not to add more fuel to the fire but....

 

 

This for Mazak but I'm sure it holds true for Fanuc as well.

[Guest]

@ fat_robot,

 

Welcome. What type of 5-Axis machine do you have? See below and pick one.

 

What control does the machine have? FANUC, Heidenhein, Siemens, or Fidia?

 

How you actually rotate coordinate systems will depend on what control you have.

[fat_robot]

@ fat_robot,

 

Welcome. What type of 5-Axis machine do you have? See below and pick one.

 

 

What control does the machine have? FANUC, Heidenhein, Siemens, or Fidia?

 

How you actually rotate coordinate systems will depend on what control you have.

First one with Fanuc. BTW one member of this forum had already helped me by advising to check the sample in the Cimco. So I watched it and found code like this

(FRONT FACE)

G68.2 X0. Y0. Z-50 I0. J90. K0.

(RIGHT FACE)

G68.2 X50. Y0. Z-50 I90. J90. K0.

(BACK FACE)

G68.2 X50. Y50. Z-50 I180. J90. K0.

(LEFT FACE)

G68.2 X0. Y50. Z-50 I-90. J90. K0.

And then G53.1 is used.

It means that angle J defines rotation around X axis and angle I defines rotation around Y axis. Unfortunately I had't time to run it on machine but will do it as soon as I can.

[Watcher]

From Heidenhain iTNC control....

 

[Newbeeee™]

There's just one thing I don't understand.

 

 

 

 

Actually there's a lot more than the one

[Oscar R.]

If one wants to have their part program display in "Part Coordinates" this function is helpful.

 

I just did some exhaustive training yesterday with a customer on FANUC 5-Axis Functions WSEC, TWP and TCP and using them seperately and together. I'm fine tuning my PowerPoint presentation. We're probably going to be put together a class after the first of the year for customers.

 

Hey James, I sure would be interested in that type of training in a class format.

Hope you can give us all a heads up on when and where we can attend when you get it all sorted.

 

Thanks in advance.

[crazy^millman]

Poster was removed.

[Delcandoit]

I looked at this topic last Saturday and it baked my noodle. I went home and read the Wikipedia link from Colin, watched many Khan Academy videos, reread the Wikipedia link from Colin, visited all other links on this thread, then rinsed and repeated. I came back that Sunday and reread this thread... it made sense!!

 

This forum has been an invaluable wealth of knowledge for all things manufacturing. Thank you guys, keep it coming so that we can keep moving our industry forward as a whole.

[fat_robot]

Finelly I've run example with G68.2 code on machine and error about illegal operation appeared. Searching by number of mistake gave me nothing. So I changed G68.2 to G68 and it kind of worked. Maybe manufacturer changed some codes? For example I have to use G65 instead of G37 for tool measurement but there is no word about it in the manual. It's bugging me whole week.

[Guest]

There's a certain order you need to activate things in, otherwise you get errors that you can't make heads or tails of.

 

Should be like this;

 

(FRONT FACE)

G68.2 X0. Y0. Z-50 I0. J90. K0.

G53.1

(DO SOME STUFF)

G69

(POSITION)

(RIGHT FACE)

G68.2 X50. Y0. Z-50 I90. J90. K0.

G53.1

(DO SOME MORE STUFF)

G69.

.

Blah, blah, blah, blah...

[fat_robot]

Thanks everybody for help. Finally I realized how it works and how to use it properly.

[Bob W.]

I am currently in the process of implementing WSEC on my A51 (horizontal machine with C-axis rotary installed on B-axis) and I have a few questions. If a part is askew relative to the B and C axes it is easy enough for the control to compensate, but can it still compensate if the part is askew relative to the A-axis (about X) even though the machine doesn't have an A-axis of rotation? If it will, this will result in some pretty large C-axis moves to do the compensation, will Vericut simulate this?

[crazy^millman]

Bob gut tells me no it should not. I would think it should only do what it has, start getting into things outside what it has and asking for all kinds of trouble. James, Tim or Jeremy would be my answer guys to ask on this one.

 

Vericut should be able to handle it if it is set up correctly.

 

HTH

[Guest MTB Technical Services]

WSEC is , by definition, a 5-axis function that finds the solution that is closest to the tool position and vector in the current coordinate system.

It is calculated based on the kinematics of the rotary axes established in the control and present on the machine.

It will then rotate both rotary axes and reposition XYZ accordingly.

 

http://www.faqs.org/...app/20090093905

 

As far as Vericut is concerned, It would require the equivalent of the WSEC offset area of the control

to properly simulate the repositioned part orientation. It has to have the numbers that the control is using

for the repositioned part orientation just as it does for simulating G54-G59 work coordinate positions.

[crazy^millman]

Like I said someone else could give a better answer. Thanks Tim for chiming in.

[Bob W.]

I just heard back from Makino and they say it can be done but as I suspected it involves very large adjustments to get the part properly aligned and there can be a huge variance from part to part as opposed to the very small adjustments required to account for B and C axis errors alone. One concern is that there are two alignment solutions for every error scenario and I'm curious how the machine chooses the right one or if it is possible to bias the machine toward one or the other, much like misc integers allows a programmer to set the axis to positive or negative in 5-axis machining.

 

I am looking into this because I am setting up probing routines to find all of the axis locations and setup errors and set these parameters every time I run a given part. With pallet change error and other variables this is the only way to really guarantee the most accurate parts coming off the machine.

[Guest MTB Technical Services]

WSEC will always use the solution closest to the current part orientation in the current coordinate system.

 

In practical terms, think of it this way.

C-90 and C270 will both get you to the same place.

The controls are typically setup to take the shortest path.

 

WSEC works in a similar way in the selection of a solution.

[Guest]

If a part is askew relative to the B and C axes it is easy enough for the control to compensate, but can it still compensate if the part is askew relative to the A-axis (about X) even though the machine doesn't have an A-axis of rotation?

Yes. it can compensate for A-Axis error even though you do not have a physical A-Axis. It will compound B and C to compensate.

 

If it will, this will result in some pretty large C-axis moves to do the compensation, will Vericut simulate this?

It could very easily be very large C-Axis moves. I've purposefully skewed some parts 10 Deg in one direction and 45 in another just to see how it comps. You're either going to need to physically shift the part the known amounts so that Vericut is checking reality or it will do a shift internally. I don't know enough about Vericut to know if it's capable of doing this without a physical shift. I know in CAMplete, I need to physically shift stuff to do the check which is no big deal, a few mouse clicks and type in the values.

 

WSEC will always use the solution closest to the current part orientation in the current coordinate system.

 

In practical terms, think of it this way.

C-90 and C270 will both get you to the same place.

The controls are typically setup to take the shortest path.

 

WSEC works in a similar way in the selection of a solution.

^^^^^^^ This.

[Guest]

I just heard back from Makino and they say it can be done but as I suspected it involves very large adjustments to get the part properly aligned and there can be a huge variance from part to part as opposed to the very small adjustments required to account for B and C axis errors alone.

I'm surprised the girls at Makino even allowed you to get the option on your machine. They always push RTDFO for everything.

 

You're going to need to set the 19700 parameters. If you plan on removing the Rotary Table from time to time, you'll need to reset them every time you re-install it. These parameters tell the control where the CL of rotation is in X, Y, and Z are, as well as the intersection point error between C and X, the distance from the face of C and the CL rotation of B and a couple others. They are all in metric and if you don't set these correctly, you'll never get a good part no matter what and how hard you try. And you don't have to be off much for it to be a problem.

 

What I do just to make things easier on myself is to program from CL of rotation, and have my common work offset be the CL rotation position as well. Now if I have to have part coordinates showing in my program, then I have my work offset be the distance from CL rotation to part zero and use TWP to manage the angle orientations and translations.

 

...I am setting up probing routines to find all of the axis locations and setup errors and set these parameters every time I run a given part. With pallet change error and other variables this is the only way to really guarantee the most accurate parts coming off the machine.

You'll definitely need to have a probing cycle to set your 19700 parameters every time it comes out of the machine envelope. I'd suggest getting Renishaw Axi-Set. It'll save you from having to write all that stuff yourself. Your machine is a little different bird than say a trunion style machine because you are physically moving one of the rotary axes from the machine so you'll need to do it every time that pallet comes out. Just out of curiosity, what's your pallet change accuracy? Meaning if you measure a location, do a pallet change, then pallet change the first pallet in and re-measure the position? 5 microns?

[Bob W.]

Just out of curiosity, what's your pallet change accuracy? Meaning if you measure a location, do a pallet change, then pallet change the first pallet in and re-measure the position? 5 microns?

 

I'm not really sure because I haven't done any testing to figure it out, yet... When things are clean and there is no chance of chip contamination I'd say it is well within spec. I believe chips and fines from surface machining work their way in there and cause accuracy issues and it is hard to quantify. Since my accuracy issues a few weeks ago, things have been 1000% better because I'm not taking anything for granted, like accurate pallet changes. I am probing every cycle on the parts with tight tolerances and running much longer warm up cycles and it has made a big difference. It is still a work in progress however, developing all of this takes quite a bit of time and I do still need to ship parts.

 

I appreciate the input Tim, Ron, James, and Jeremy, and I will look into Axi-Set to see what it can do.

[Zbuilder]

Actually, Tilted Work Planes with Euler angles rotate about the programmed toolpath's Z, X and Z again. Just got a Matsuura MX520 and been doing a LOT of reading.

[MSL]

I hope this help.

Tilted work plane operation.pdf

[So not a Guru]

On 12/17/2010 at 6:00 PM, Colin Gilchrist said:

 

Martin,

 

This may be your particular Machine's implementation of G68.2, but many Fanuc controls will allow 3-dimensional rotation.

 

On these controls you will get a G68.2 X Y Z I J K (with no "R"). The I J K values will use Euler Angles for describing the Rotation about X,Y, and Z.

 

In these cases, I, J, and K will hold a rotation value from 0-90 degrees.

 

Most controls use G53.1 in conjunction with the G68.2 command. The G53.1 controls the Tool Axis direction (Z+ or Z- direction).

 

Wikipedia does a great job of explaining Euler's angles:

http://en.wikipedia.org/wiki/Euler_angles

Colin, what are the pros & cons of G53.1 vs G53.6?