5 Axis Offset for the A Axis not working

[ctintinger]

Hi,

I have a similar issue to this - https://forums.autodesk.com/t5/fusion-manufacture/machine-setup-rotary-axis-offsets-question/m-p/966... and have tried a few suggestions there but to no avail.

My machine is a Bridgeport Clone (Gate Machine) which uses a Nikken 5AX 170 4th and 5th Axis for A and C.

I have retrofitted the machine with a UCCNC controller which does can control up to 6 Axis, in this case 5 are used. The controller does not support TCP.

There is a post processor for UCCNC which I have struggled with. I then found that there is a 5 Axis Mazak PP which works once I changed it from A and B to A and C in the PP. Functionally it works, code runs, machine moves.

What I am struggling with however is that nothing I seem to do allows me to offset my rotary location. I've even built a Machine in the Machine Builder which functionally works and all the Axis move as needed.

However, when I do any CAM work that involves the A Axis, the offset is not taken into account. I've tried to edit the PP to offset there (based on that other thread). I've tried to edit it in the GUI under the Z offset. Ive even tried to fool the machine by making the part a different size.

In my case, the A Axis's rotation sits 50mm below the top of the mounting plate. In other words, when going A0 to A90, the part's relative X and Y should to around 50/60mm whereas it doesnt. The test part I have is only 20mm in diameter and around 30mm high but already with the 50mm offset, you should see X and Y change a lot when A is 90. The A table only does from 0 to 90, it doesn't do the negative direction (it does 5 degrees but negligible, so not functionally useable).

So, if I just watch the simulation (before PP), should that show the offset working? In my case, it doesnt.

Then after PP, it doesnt change, so that's clearly not working.

Should this offset be managed in the Machine setup or in PP?

Watching a Haas video where the guy shows how you would do this without the fancy acronyms that do this stuff for you, he puts the offset in his model and simply has his model point the few MM higher than his plate to do this offset. I've tried similar, no difference. What am I missing please?

https://www.youtube.com/watch?v=hZdXp8Gy_GM&t=189s

[ctintinger]

This is currently what's in my PP. That said, this 'offset' issue isnt solved even before PP, but not sure if it would / should. I expect to see X and Y numbers around 50-60mm, not 5-15mm once the A table goes to 90 degree as the centre of rotation is 50mm lower than the top of the C plate.

if (true) { // note: setup your machine here
var aAxis = createAxis({coordinate:0, table:true, axis:[1, 0, 0], range:[0, 90], offset:[0, 0, -50]});
var cAxis = createAxis({coordinate:2, table:true, axis:[0, 0, 1], offset:[0, 0, -50], cyclic:true});

//var aAxis = createAxis({coordinate:0, table:true, axis:[1, 0, 0], range:[0, 90]});
//var cAxis = createAxis({coordinate:2, table:true, axis:[0, 0, 1], cyclic:true});

machineConfiguration = new MachineConfiguration(aAxis, cAxis);

[ctintinger]

And in terms of the setup of the machine.

Its a fixed head with a spindle that comes down (Z), typical Bridgeport setup.

The C sits on the A which sits on the X which sits on the Y.

Are my kinematic setups correct for this? Should I also be offsetting C? I don't believe so as its the same location in the X and Y.

[ctintinger]

Anyone, please?

Could another solution to be to calculated / transform the Y and X values based on the A angle? We know that the height offset is 50mm. Using some basic trig, we can work out the adjacent distance which is the Y difference when the table leans over. At 90 degrees, its 50mm, at 45 degrees, its 35.355mm etc.

Thanks

[seth.madore]

Couple of things:
1) Don't give much regard to the numbers that are in the "Simulate > Info" tab. Once you move beyond 3 axis they do not reflect real numbers that you would see in the posted code or the machine. 

Ultimately, what we're looking for is "posted code" values. Where do those land?

Can you share a sample Fusion file that would contain your machine configuration? What post processor are you currently using? If it's modified, please share that here as well.

[ctintinger]

Hi Seth,

OK, that's interesting to note (ie, to ignore the X Y Z), that said, I have post processed a few times and run each time on my local machine (demo mode) and it never offsets as I would expect.

Ive got about 3 different post processors Im using to try, one is a Mazak one, one a Haas one and one is UCCNC which now oddly is working again. None of them 'transform' or seem to offset the Y value as I would expect to see as the table tilts. Even programming 50mm offset in the Model via Fusion seems to be ignored (as far as I can tell). Let me attach 3 of the Post Processors I am using (that post and seem to work, albeit ignoring the offset issue). Thanks.

[seth.madore]

A) What is your control on your machine? Go with that post processor. If it's Mazak, use Mazak. If it's Fanuc, use the Fanuc post

B) Machine Configuration is where you're likely going to be solving this, not in the post processor. Please share a file that shows your machine configuration.

[ctintinger]

Here are some files. The simple part, hoping the CAM will come across too.

The machine file itself. Ignore the 'odd' shapes on the side of the A axis, was just double checking the A angles and the distances I would expect Y to offset by. I did make a rotational joint in the CAD, not sure if that will cause an issue when used. I can delete / undo if needed and share the earlier version if needed.

And the post processor (one of them Ive used where I've programmed in offsets etc).

Thanks

[ctintinger]

Also, I have attached the base PP for my controller which ideally I would use as this version is 100% right for my setup, barring getting the 5 axis elements to work.

[ctintinger]

Hi, It's a retrofitted machine using a UCCNC board. I've attached their PP in the last post.

How do I share the Machine Configuration please? Do I export the CAD that has that CAM / Machine setup or can I somehow export my Machine Configuration on its own?

[ctintinger]

OK, looks like I can just export the machine config, attached, hope this works. Please let me know if not.

[seth.madore]

Were you planning on setting your WCS in one common location (perhaps center of your C axis rotation) and taking it from there, or were you planning on one at A0 and another at A90.?

[ctintinger]

I dont know, new to 5 Axis. I was watching a few guidance vids where they show that you have to offset for the Z difference and must match the centre of rotation (50mm up for me). I'm happy to the centre of the C plate as my WCS and go from there.

[ctintinger]

Doing one at A90 is interesting, more akin to 3+2. Would be good to get that sorted at a point however just keen to get the current 'offset' challenge solved around this centre of rotation.

Ive seen some videos call a combined 4th and 5th axis B and C, despite the 'B' axis rotating along the X. Surely this is incorrect - just mentioning it in case ive misunderstood and my offset doesnt work due to that.

[programming2C78B]

I could never get a 5th axis program to work when the machine is set to a90 (part pointing towards operator)

I'd have to program it as the part was vertical, but put notes to locate the part at g53 a90 then then put -90 in g54.

[ctintinger]

Hi, which I think is pretty standard when doing 3 + 2. As, all the A axis is doing is orientating the part in a different alignment / location. That's not what I am trying to solve here however.

[ctintinger]

Could this be an issue? I try and save the kinematics for A by saying its around the X, but each time I reopen this, its gone back to the default. The machine is a local copy and not linked to any active CAM sessions.

 

[ctintinger]

I think I may have solved it. The machine builder offsets dont work at all, no idea why. I found that the Mazak PP which had posted before, without offsets, had TCP enabled. So I think what was happening is that it was ignoring the programmed offsets by saying 'just ignore that, TCP will solve this'. Ive tested a quick file with 50mm offset and Y moves. I dont think its spot on. Just tested with a 150mm offset which is extreme and the Y moves almost 200 in the extreme from - to +, so that seems to be following the offset.

[ctintinger]

This is now the resulting movement, you can see how the Y goes in and out as the part if being spiralled around (C movement not shown)