5 ax, center of rotation.

[crazy^millman]

No reason it wouldn't work for full 5x, the actual math behind it is all jr high school algebra.

 

cool then I need to go back to Jr High and retake that basic stuff, because with all the computer language and other things you need to make it all tie together I still see so many struggle including myself to make everything come out correct be cool to see how you would make it all work so easy for 5 axis toolpath with such a simple Macro.

[Bob W.]

[/size]

 

Bob so you would then program parts one way some of the time then other parts a different way some of the time? Again getting back to the post and machine and control definition you ensure you program the same way all of the time.

 

Yep, I do it all the time. Some parts I program for my 3-axis Haas, others I program for my 5-axis Makino... It really isn't that uncommon in the machining world to machine a part based on the requirements of that part. I have a Makino PS95 that I am adding a Koma rotary table to and I plan to do exactly that. The Fanuc Om control does not support DFO and the lion share of my work will be 3+1. Creating the macro will allow me to probe off the part and calculate the transforms on the fly instead of programming to the axis of rotation and hoping that my setup is dead accurate. I guarantee the accuracy with the macro method will be about 5-10X better than programming to the center of rotation. It also gives me the flexibility to move that part to a different machine without a lot of Mastercam drama. The cost is that one needs to take a different approach to simultaneous multiaxis machining. Small price if you ask me, and it certainly beats transforming the part in Mastercam and reposting every time that job comes up.

[Bob W.]

[/size]

 

cool then I need to go back to Jr High and retake that basic stuff, because with all the computer language and other things you need to make it all tie together I still see so many struggle including myself to make everything come out correct be cool to see how you would make it all work so easy for 5 axis toolpath with such a simple Macro.

 

I wouldn't say it is Jr. high math but it really isn't that difficult. Essentially it is linear algebra with sin and cos commands. 3x3 matrix I believe. Modify the post to call a macro program after indexing. Calculate the transform which would be about 6-8 lines of code, then modify the work offset with G10 command. It would probably take me about 3-4 hours to get this all up and running. One would have to be pretty good with high school trigonometry and have some experience writing macros.

[Guest]

:cough: WSEC and TWP = Part Coordinates, 3+2 AND 5-Ax and WGAR about CL Rotation. :cough:

 

Man that's a nasty frog I've got in my throat. I think I need to get that looked at

[gcode]

You could also have one post/machine def with 2 copies of these settings

 

Part programmed at machine zero location-

#Offset in head based on secondary axis relative to machine base.

#Normally use the tool length for the offset in the tool direction

saxisx : 0 #The axis offset direction?

saxisy : 0 #The axis offset direction?

saxisz : 0 #The axis offset direction?

 

each one set for one of your machines.

 

then set up the post to ask you which set to use

[crazy^millman]

Hum never needed to do that and seemed to cut my share of parts without all the transform and drama you are talking about. Oh well carry on seems I cannot be of much help in this thread. Carry on I am off to Westec for the week.

[Bob W.]

Math goes something like this. Original work offset is composed of X,Y,Z coordinates and the new work offset would be X1, Y1, and Z1. The axes of rotation are A (parallel to X-axis) and B (parallel to Z axis when B is 0.0). This is the configuration of the Haas trunnion.

 

X1 = Xcos(S)+Ycos(A)*sin(S)+Zsin(A)*sin(S)

Y1 = Ycos(A)*cos(S)-Xsin(S)+Zcos(S)*sin(A)

Z1 = Zcos(A)-Ysin(A)

 

(replace 'S' with B, B gave a bunch of smiley faces with sunglasses on...)

 

#550=X1

#551=Y1

#552=Z1

 

G91 G10 L2 P1 X#550 Y#551 Z#552

 

There you have it. 90% done.

 

For more information Google "Euler Angles"

[Bob W.]

Hum never needed to do that and seemed to cut my share of parts without all the transform and drama you are talking about. Oh well carry on seems I cannot be of much help in this thread. Carry on I am off to Westec for the week.

 

That method works no doubt, but there are some merits to the macro method. Namely eliminating most of the fixture inaccuracies and the ability to set the work offset at the beginning of the program through probing. That is where the accuracy improvements come from. I never did this on my Haas with the trunnion because I didn't trust the control to do the calculations correctly. I have had it grab the wrong program with a M98 command...

[Newbeeee™]

Newbeeee question time...

So what is the difficulty in setting the machine as per my previous post ie could you use a combination of parameter changes to grid shift XY, and machine the table for Z?

Would this work for 5ax?

If so, wouldn't it save a huge amount of headache?

[Guyinthedesert]

[/size]

 

cool then I need to go back to Jr High and retake that basic stuff, because with all the computer language and other things you need to make it all tie together I still see so many struggle including myself to make everything come out correct be cool to see how you would make it all work so easy for 5 axis toolpath with such a simple Macro.

 

I'm going to look in to this. First snag, I don't think my CP/m fortran compiler will work with Windows 8.

[Guest]

Machien CL Rotation parameters on a 5-Axis FANUC reside in the #19700 parameters. Keep in mind these were set at the factory and will differ once the machine has been installed and has settles over time. Granted, we're talking microns here, but still.

 

Those parameters can be changed at any time. Renishaw Axi-Set software can help you find the actual position dynamically as opposed to the normal static method found most often being used in the factory. One of the options will allow you to "fix" the parametes. Now, these parameters only make a difference when you're running WSEC, TWP, TCP, etc...

[Bob W.]

The problem he is trying to resolve are related to a Haas with a trunnion, so the WSEC options aren't even a possibility. I created a probing routine for my rotary table in the Makino that finds the true position of the axes, stores them in a work offset and uses them in the G10 lines when setting the DFO, WSEC, or TCP offsets. Seems to work really well for squeezing out that last .0001" :-)

[Guest]

Thanks Bob, this topic has wandered all over the pasture and I got lost along the way.

[Bill Henderson]

Been wandering around looking and stumbled across this thread.

 

We have numerous castings that will be coming in. Three datums ar on the NON machined side of the casting. We have four tabs that will be machined on that side to establish points for finishing opposite side. On teh HAAS 5 axis

 

We are trying to see if there is a way to run the probing cycle with a macro to set our A ( datum to be flat) and B rotations (mostly needed to get A datum flat. Once that is set we need the program to run on the skewed B angle. (with just one posted program)

 

From what Bob was saying it is possible.

 

We have a Hermle that will do it easily, but it is bottlenecked.

 

Any help would be appreciated.

[crazy^millman]

Bill short answer yes.

 

Long answer I will let someone with more experience chime in. I think I could figure it out, but would need to be in front of the machine and see what works and what does not work to dial it in. Sounds like Bob has amethod if he does not mind sharing that might work. Might PM him and see if he is willing to share.

[Bill Henderson]

Thanks Ron

[Bob W.]

Probing the A-axis to make it flat will not work but probing the B-axis will work just fine. If the A-axis is askew only something like WSEC can be used to compensate for it I believe. DFO will not work for this and I ran into this exact scenario in my shop. Imagine that the A-axis is off by 10 degrees (exaggeration) and you probe and set it to A0. Well when the machine is now at A0 the B-axis is NOT vertical anymore, it is askew to the Z-axis by 10 degrees. If your part has features that are machined at A90, B90 they will all end up askew by 10 degrees in some direction... It is my understanding that WSEC compensated for this by doing a full G68 type transform in 5-axis which is extremely powerful. The methods I outlined are good for XYZ translations only. The B-axis can always be adjusted as normal no matter what. I found this out the hard way on my Makino when running DFO.

[Guest]

Yes Bob, you are correct. If he has skew in an axis that the machine does not physically have, it will require a compound angle to flatten it. This PLUS linear error is what WSEC handles in 3+2 AND 5-Axis. Then if you have Tilted Work Plane (G68.2) you can get part coordinates going.

[Bill Henderson]

That is what I was figuring... end result would be compound angle.

 

Fixture is mounted...casting secured on fixture. Surface (W) to be flat is in (parallel to table) has to be rotated in B and A . Therefore when (Surface is parallel to table) A may be 5 degrees and the X axis is now skewed to whatever B (10 degrees lets say) became to make (W) flat.

 

If A is "zeroed" at said 5 degrees... the entire program would have to run skewed to the 10 degrees the B has been rotated.

 

Does that make sense?

[crazy^millman]

Bill to me it makes sense, but sounds like you need to be using a best fitting program that can take probed data and then give you a new matrix. From that matrix you need to out put a new index position to run the 3+2 program from. Trying to go to full 5 axis from a shifted matrix and not taking that into account back at the CAM stage requires a control that can handle that or reprogramming based off the new Matrix. Neither one sounds like a viable option here. I vote for a 3rd option. Do a batch of 20 castings. Then run full inspection of the parts. Figure out which one if the best and figure out which one is the worse. Make the program where it takes the best and worse into account. Then figure out the index position that covers them both. Make that the set place and be done with it. If that requires a little more work on the prpe operation to make sure you are control things you were not before then implement that process to allow for that quality upfront. There is where PCMM pays for it self in weeks, yet I am sure you guys do not have one. I know of companies that do casing that are cutting days off machining their big castings, because they are taking PCMM to the castings and getting real world number and then working with real world number to know what direction to go in. Bill I am sorry, but it sounds like you got a poker game going on right now and someone has dealt you all 2's and sold a customer a hand full of Aces. Predictability needs to become your friend in this case and right now not sure how predicable any of this is.

[kool]

This is how I m doing it on 2 Haas TR310.

 

write you center of rotation for each machine (x,y,z) in custom macro variables (#501, #502, #503). you can pick any variable that are not use by any other program.

Load center of rotation using a G10 line on top of program and call these variable (G10 L2 P1 X#501 Y#502 Z#503) etc.

 

You can use the same program on both machine using this method.

[Bob W.]

If you have to skew the part in both A and B during setup it will not work. If you adjust A (to 5 degrees) to get the part flat, B is then askew to the Z-axis by 5 degrees. When machining at A90 the B-axis is NOT horizontal, it will be askew to the XY plane by 5 degrees. You will get bad parts and only WSEC can correct for this, which I don't think is an option on Haas machines. WSEC accounts for XYZ translation and ABC rotations and does all the calculations to get everything right. With the A-axis at 90 degrees the B-axis indexing will be at a skewed angle that is not parallel to the XY plane. You can only do probing to set the B-axis and still get good results. The part has to be set up flat (parallel to table) and you can't adjust A to get it there.

[PROBE]

This is my first post here. I am not programmer, and am not using Mastercam system. I am probing specialist. As  I highly appreciate the skills , knowledge and experience  of members of this forum, I would like to share it.

 

I've been approached by customer with substantial number of 5 axis Makinos A51 ans Matsuuras MAM72 to supply the solution to of center of rotation of B and C axes. As machines are equipped with standard touch probes from several suppliers and not Renishaw Rengage, the use of Axiset wasn't the option.
I wrote the routines, which execute the measurement of calibration sphere in bunch of B and C axis positions. All calculations are made in macro (no need to use external PC), at the end the 19700-19705 parameters are updated, or alarm is raised if mechanical intervention is needed.

The job is done. Now I would like to understand how these parameters are executed in the control. Are the linear (XYZ) movements compensated when rotary (BC) are moving? Should any G function be called to enable this compensation ?

[absolute technologies]

This is my first post here. I am not programmer, and am not using Mastercam system. I am probing specialist. As  I highly appreciate the skills , knowledge and experience  of members of this forum, I would like to share it.

 

I've been approached by customer with substantial number of 5 axis Makinos A51 ans Matsuuras MAM72 to supply the solution to of center of rotation of B and C axes. As machines are equipped with standard touch probes from several suppliers and not Renishaw Rengage, the use of Axiset wasn't the option.

I wrote the routines, which execute the measurement of calibration sphere in bunch of B and C axis positions. All calculations are made in macro (no need to use external PC), at the end the 19700-19705 parameters are updated, or alarm is raised if mechanical intervention is needed.

 

The job is done. Now I would like to understand how these parameters are executed in the control. Are the linear (XYZ) movements compensated when rotary (BC) are moving? Should any G function be called to enable this compensation ?

The functions that use #19700 parameters are G43.4 TCP, G68.2 TWP and G54.4 Work Piece Setting Error Compensation. I've often wanted to do what you have done with the probing of a sphere, however my thought would be to set the pivot points with mechanical means (dial indicators) then run the probe routine to verify the probe results coincide with the mechanical check such that the operator can run the probe routine anytime for verification. The reason I say set the pivot points with dial indicators it can be way more accurate than a probe with less uncertainty. Then from there do your circle, diamond square test cut or better yet, machine a 5-sided block with various features using TCP & TWP that isolates the pivot points better than the Circle Diamond Square test and that can be inspected using surface plate, height gage and dial indicators with extreme precision, make your final adjustments to #19700's and then verify the test cube on the CMM as a reference. This way you have a solid static mechanical check that is backed up by probing and the 5-sided test piece will give you a dynamic look at the volumetric accuracy.

 

Back in mid 80's I was involved with the installation & commissioning and was the operator on the 1st 5-axis machine in the US. We didn't have probes back then. One of the first jobs was a drill jig with true position of .001 on compound angled holes. The 1st part came out about .020. We went back to the old method of setting the spindle angle, pick up a tooling ball and bored the hole on location. After that job we re-vamped the machine and it got about 10 times more accurate. Here we are in 2016 and we still need to be about 10 times more accurate than where we are today IMO.

 

I used to work for the Mori dealer and was involved with their 5-axis machines when they 1st came out. Mori had no clue on how to get the pivot points down to a tenth or less. Up until a just a couple years ago they still had no clue. At the same time we also sold high end Euro machines like Jobs and Fidia. If you have a chance, check out how Fidia sets their pivot points. Its' amazing there is no industry standard for this and customers are left to the expertise of the machine builder.

 

Cheers!

Len Dye

[Watcher]

https://youtu.be/iCNAGPsWM4w