Alignment of 4th axis trunnion

[Jacob Justesen]

Hi Guys

I have a brand new Fanuc Robodrill with a Fanuc Trunnion factory installed.
My problem is when I tilt the table.
In 0° I align it perfectly, but when I tilt in 90° and 270° and facemilling it, it does not make those sides parallel to each other.
I believe it is some adjustment of the table that is necessary. The 2 points in the bearings of the trunnion is in alignment, so it must be the baseplate or something of the table itself.
How is right way to make a complete alignment of the trunnion?

Jacob

 

[cwtoyota]

Hi Guys

I have a brand new Fanuc Robodrill with a Fanuc Trunnion factory installed.
My problem is when I tilt the table.
In 0° I align it perfectly, but when I tilt in 90° and 270° and facemilling it, it does not make those sides parallel to each other.
I believe it is some adjustment of the table that is necessary. The 2 points in the bearings of the trunnion is in alignment, so it must be the baseplate or something of the table itself.
How is right way to make a complete alignment of the trunnion?

Jacob

Flex? Did you indicate the trunnion in all three positions?

Clamp a 123 block on that sucker and check the same surface in all three positions.
That will tell you if the table is out, or if it's due to flex.

 

[Tonytn36]

It could also be scaling or backlash. When the table is at 90 and 270, is the face of the trunion parallel to the Z axis? (This should have been checked by the installation technician.)

 

[Jacob Justesen]

The service from Fanuc is non existing. You would not believe the trouble we have had with this brand new machine.
Ok so, my first job is to get some kind of alignment rod/shaft so i can check if its parallel to Z and aligned in X.
Stupid me assumed it was OK from the factory.
If I check the same surface in all three positions I have an error of 0.07mm, which is way to much.

 

[Nerdlinger]

Twist in the casting? Are you sure the machine's level?

 

[cameraman]

The service from Fanuc is non existing. You would not believe the trouble we have had with this brand new machine.
Ok so, my first job is to get some kind of alignment rod/shaft so i can check if its parallel to Z and aligned in X.
Stupid me assumed it was OK from the factory.
If I check the same surface in all three positions I have an error of 0.07mm, which is way to much.

That was my first thought that the trunion and axis of rotation is not parallel to the X axis of the machine/X movement I.e. there is an angle in the XY plane created by a basic mis-alignment.

One would guess (I stress guess) that the trunion is "Flat" to the table (assuming a table),(indicates as "flat" from the spindle) but the trunion axis/4th rotary veers off at an angle. So if you mill a part (face top surface of work piece) it will be "flat" relative to the spindle and probably the table, but then you rotate 90 degrees and the angle in the XY plane now becomes "flipped" into the trunion's XZ plane which in turn is set at an angle in the XY plane of the machine (so you have two coordinate systems, at least one angle of rotation to the other). In essence you have a compound angle when the trunion is rotated (no sh*t :-D ). Then if you flip the trunion to 270° then you have a further "flip" or transformation with opposite angles.

So I would have thought if the trunion is out only (angle wise) in the XY plane then it would be possible to still cut parallel sides but the work piece becomes a parallelogram instead of being completely square (on all faces) which leads me to believe the trunion may be out on two planes/two angles. (from what you are saying... It sounds like you may be cutting a tapered or really a trapezoidal part (instead of "square"), like what you have "Reported", as the trunion is out on two angles/two planes) Like its high-low end to end and not parallel with the X axis/x movement of the machine. Like what folks are all saying here. But on the other hand I can imagine very specific scenarios where an entire part still comes out "right" even if the axis of rotation tilted...

Which is really messed up for Fanuc (if that's the case)... The coordinate system and axis of 4th axis rotary and trunion has to be completely parallel to the XYZ movements of the mill. Being forced to mill unwanted compound angles the whole time is not what you signed up for lol. (Don't worry I'm sure this can be fixed and maybe you get to learn how to do it yourslef).

I wonder if the machine was shipped from somewhere else and with the assumption that final adjustments to the trunion were made, and in fact had not been made? Or maybe machine got slammed in transportation.

Did they adjust or check the trunion for you in the install? Or did they just plonk the machine down and head for the door?

Was there a separate part of the contract for like $4K that they check the machine with Renishaw (metrology grade) type equipment?

 

[Tonytn36]

Sounds to me like the service is so poor that the very basic standard set-up checks for a new machine were never performed. I.e. it was never installed properly (or this wouldn't be an issue).

To the OP:

Is the trunion aligned so as to be A,C or B,C?

Is the 0.07mm measurement along the plane axis (XY) or is it tilt in the table (rotary).

 

[DMF_TomB]

It could also be scaling or backlash. When the table is at 90 and 270, is the face of the trunion parallel to the Z axis? (This should have been checked by the installation technician.)

.
always go same direction cause of backlash. you reverse direction usually why its off
.
some over travel and reverse direction so alawys coming from same side of backlash

 

[Jashley73]

It shouldn't matter if the trunnion is not aligned to the X-axis. (Assuming the "A" axis of the trunnion is setup to be parallel to the X axis.) It could be skewed by 20 degrees. If you're truly indexing in 90* increments, you're going to end up with (3) square sides...



Questions about the rotary unit...

1- What kind of rotary table/unit do you have? If this is a Fanuc DDR, or any other linear/direct motor type, there should be no backlash. If it's a worm drive, or rotary cam drive, there could be backlash, but I doubt it. Even for rotary axes, there's parameters for backlash compensation.

2- There could be deflection...

3- Does this machine have a column riser installed? (Could cause the column to not be perfectly square...)




Things to check...

1- Indicate the table at "A0.0" to make sure it's truly flat. Across the "Y" direction, but also across the "X" direction as well. If it's not flat across the "Y" axis direction in the "A0.0" position, then you may have to tweak the "home" position. If it's not flat in the "X" direction, then either the head is out of tram, or the rotary unit itself is not parallel to the table. Sweep across the trunnion table, and the table itself to see if they're parallel. Rotate the rotary 180* and sweep the same face again - the error should be opposite now, of the same amount.

If the trunnion and table are parrallel, yet when you tram the trunnion table with your indicator (rotate the spindle/indicator, and keep the table stationary) and this check shows the trunnion/table are not square to the spindle, then then the head is out of tram...



1b- Cut the face. Indicate the table again, to see if anything moved.

2- Rotate the table 90*, as you would in the program. As in, mimic the program as much as possible. If the table is at "A0.0" and you have a "G0 A90.0" line in the program, then go to MDI mode, and execute the same "G0 A0.0" line to mimic the direction, and speed of the index. Then, indicate the same face with the machine, by moving the Z-axis up/down to check this face. Write down your results. Cut the face, and then indicate the same face, using the same method again. Compare results to see if anything moved, which would indicate deflection.

3- Index again. Again, mimic the program. Indicate the same face again, using the Z-axis. Write down results. Cut face. Indicate again, and compare results.


By now, you should have some data...



If you're seeing different indicator readings after cutting each face, then something is deflecting.

If the Z-axis is truly square to the table, and the head is trammed square to the table, then you should end up with a square part. If something is not square, then either the Z-axis is not square to the table, or the head is out of square.

Robodrill's are built using very simple construction. There's only 4-feet, so the leveling job should be fairly simple. If the level is off, I doubt it would cause a whole lot of issue, other than maybe a twist in the Y-axis.




If you can't figure it out using the above method, then I'd try to track down a calibrated granite square, and give the machine a once-over to see if the machine is square.

 

[cameraman]

It shouldn't matter if the trunnion is not aligned to the X-axis. (Assuming the "A" axis of the trunnion is setup to be parallel to the X axis.) It could be skewed by 20 degrees. If you're truly indexing in 90* increments, you're going to end up with (3) square sides...



Questions about the rotary unit...

1- What kind of rotary table/unit do you have? If this is a Fanuc DDR, or any other linear/direct motor type, there should be no backlash. If it's a worm drive, or rotary cam drive, there could be backlash, but I doubt it. Even for rotary axes, there's parameters for backlash compensation.

2- There could be deflection...

3- Does this machine have a column riser installed? (Could cause the column to not be perfectly square...)




Things to check...

1- Indicate the table at "A0.0" to make sure it's truly flat. Across the "Y" direction, but also across the "X" direction as well. If it's not flat across the "Y" axis direction in the "A0.0" position, then you may have to tweak the "home" position. If it's not flat in the "X" direction, then either the head is out of tram, or the rotary unit itself is not parallel to the table. Sweep across the trunnion table, and the table itself to see if they're parallel. Rotate the rotary 180* and sweep the same face again - the error should be opposite now, of the same amount.

If the trunnion and table are parrallel, yet when you tram the trunnion table with your indicator (rotate the spindle/indicator, and keep the table stationary) and this check shows the trunnion/table are not square to the spindle, then then the head is out of tram...



1b- Cut the face. Indicate the table again, to see if anything moved.

2- Rotate the table 90*, as you would in the program. As in, mimic the program as much as possible. If the table is at "A0.0" and you have a "G0 A90.0" line in the program, then go to MDI mode, and execute the same "G0 A0.0" line to mimic the direction, and speed of the index. Then, indicate the same face with the machine, by moving the Z-axis up/down to check this face. Write down your results. Cut the face, and then indicate the same face, using the same method again. Compare results to see if anything moved, which would indicate deflection.

3- Index again. Again, mimic the program. Indicate the same face again, using the Z-axis. Write down results. Cut face. Indicate again, and compare results.


By now, you should have some data...



If you're seeing different indicator readings after cutting each face, then something is deflecting.

If the Z-axis is truly square to the table, and the head is trammed square to the table, then you should end up with a square part. If something is not square, then either the Z-axis is not square to the table, or the head is out of square.

Robodrill's are built using very simple construction. There's only 4-feet, so the leveling job should be fairly simple. If the level is off, I doubt it would cause a whole lot of issue, other than maybe a twist in the Y-axis.




If you can't figure it out using the above method, then I'd try to track down a calibrated granite square, and give the machine a once-over to see if the machine is square.

If the Z-axis is truly square to the table, and the head is trammed square to the table, then you should end up with a square part. If something is not square, then either the Z-axis is not square to the table, or the head is out of square.

^^^All nice^^ stuff..

Or... The trunion is (also) high-low along the x -axis where you end up cutting trapezoidal or tapered parts like what OP first reported.

OP: "In 0° I align it perfectly, but when I tilt in 90° and 270° and facemilling it, it does not make those sides parallel to each other."




As you know it's similar to a lathe like work between centers, relative to the cutting tool. So for example to cut a taper you'll move the tail stock laterally. With the Trunion example the equivalent is having a tunion that is high-low (in Z) along the X axis.


Of course the Trunion (especially a new install), should be parallel/squre to all the machine's axes of travel (whether you are using probes and dynamic work offsets and the like or not). On a machine that has a trunion that is at an angle to the X axis, (as I said earlier by facing cuts and 90, and 270 degree rotation you get parallel sides, but the part is not to the correct "dimensions", and then if you start making profile cuts for the other two remaining sides you end up with a parallelogram as the Y axis travel/movement on the mill is literally not square to the X axis (unless you use specific compensation strategies). (As you know :-) ). [As a side note that's why I'm not (personally) very keen on a two axis rotary trunion on a 3 axis mill for 5 axis work... much less hassle to have a dedicated 5 axis machine (removes a lot of fiddling about and worry).

But in general it's a good discussion as it does make you think about how to create a better manufacturing process (by design) that is virtually bullet proof/fool proof and woks no matter what (as "Best" practice).


I have to say the Fanuc "guys" in the US and Methods Machine Tools that handle robodrils/Fanuc products really have their act together and really understand these machines inside and out, and do great "Metrics" and tests for machines in service.

So I'm kinda surprised things went "Pear shaped" in Denmark?

@Jacob Justesen... Do you have a photo of the trunion / or can you post a photo?

If the machine is new surely the guys you bought it from don't want you messing with "Factory" adjustments until they can get there? [I know... "Factory" adjustments really did you a lot of good here.].

Sounds like maybe they forgot to set the trunion in the first place? (If that's the case it could be "out" on two planes/two angles.)

 

[HuFlungDung]

The real amount of the error is probably half of what you measure, so it's not much. I'd suspect the machine tram as much as anything.

But in the long view, it is asking a lot for zero error, but zero error would still be nice to have. I'd think about using different work offsets for those most important positions of the trunnion, and thereby compensating any errors out, because of having a less than perfect machine.