Turning adaptive roughing - Tool and settings help

Turning adaptive roughing - Tool and settings help

anders.lonnegraff
Explorer Explorer
2,516 Views
10 Replies
Message 1 of 11

Turning adaptive roughing - Tool and settings help

anders.lonnegraff
Explorer
Explorer

How can i use Turning adaptive roughing in this model?

I can't make it work, i am just getting the error "Toolpath is not supported for the given tool and settings.

You likely need to change tool." (I have tried with different tools etc.)

0 Likes
2,517 Views
10 Replies
Replies (10)
Message 2 of 11

akash.kamoolkar
Autodesk
Autodesk

@anders.lonnegraff we are aware of this issue and it will be resolved in an upcoming update.

 

Regards,

Akash Kamoolkar



Akash Kamoolkar
Software Development Manager
0 Likes
Message 3 of 11

Muzzerboy
Advocate
Advocate

I was investigating lathe adaptive roughing again last night and I see that the documentation suggests this is a preview feature (which in itself suggests that it is in active development).

 

I'm still struggling to get this to work in anything other than a very limited fashion. Rather like a party game, it seems to require experimentation to figure out the myriad rules and checks that have been implemented under the hood. For instance, there seem to be cross checks on the settings such as the insert diameter, min radius, optimal load etc. Some of these are fairly obvious hard requirements but there seem to be others that are preventing many of the features from working at all, judging by the cryptic error messages that come up. Given the lack of information / documentation, we are having to go into this blind.

 

A couple of things I've found so far:

  • Adaptive roughing only works with "turning grooving" tools. I could understand why certain operations can only be done with omnidirectional tools but clearly you could use an adaptive path with handed tools, with appropriate limits on toolpath directions. Either way, I don't think "Select a turning tool from the library, or create a new turning tool" in the documentation quite covers it.
  • The (grooving) tool has to have a "round shape", as set in the tool library. No matter what I do, I can't get the "square shape" to be accepted without an error. Yet it's listed as an option in the "turning grooving" tool library.
  • If I enter ANY value in the "finishing allowance" (aka stock to leave in milling terminology), I get an error "Warning: Could not reach all contour areas with given tool and optimal engagement!". For a roughing operation, this is a bit of a major issue!

I hope you can give this the resource required to thoroughly develop and test out the features. Ideally you will also include an update to the documentation, such as it is, laying out the limitations and checks that we need to take into account. Is there a roadmap for this activity that gives us some visibility of what to expect?

 

On a related note, the LinuxCNC lathe post processor on the Autodesk site is still generating G28 U & W move commands which aren't recognised by LinuxCNC - they need to be G28 X & Z but despite several requests, there seems to be little interest in correcting this.

0 Likes
Message 4 of 11

engineguy
Mentor
Mentor

@Muzzerboy 

 

You mean like this ??

Linux G28.jpg

 

Seems to work here !!

0 Likes
Message 5 of 11

Muzzerboy
Advocate
Advocate

Sorry, didn't see this until just now. I see there was an updated Post dated 2021-09-08 with no release note but as you say, it seems to have fixed the G28 issue. The initial response from the Autodesk team seemed to be quite dismissive but it's good to see that they have fixed it, even if it went under the radar. Bottom line is I can now use G28 without having to edit my post from the default.

 

Many thanks for pointing this out - and to whoever at AD sneaked out the update!

0 Likes
Message 6 of 11

Muzzerboy
Advocate
Advocate

In fact, despite what I posted earlier this week, if you download either of the last 2 revisions of the LinuxCNC post you will find it generates the U and W coordinates which will cause LinuxCNC to fail. The version you posted is out of date (it's r43417) and we are now 2 releases past that.

The fix is to paste the // format home positions section from r43417 into the current post r43470. It would be far better for Autodesk to rectify the mistake themselves so that the broken post doesn't catch anyone else out.

0 Likes
Message 7 of 11

engineguy
Mentor
Mentor

@Muzzerboy 

 

Hmmm, looking at the LinuxCNC list of codes G28 is supported, as are the U/W outputs unless they are turned off in Linux settings, where that would be I have no idea but that is what my research on the Linux online stuff.

 

Anyway, can you not use the G53 ? There is option in the PP Properties to set the G53 to a set position, does that not do what you need ??

0 Likes
Message 8 of 11

Muzzerboy
Advocate
Advocate

I think the point you are missing is that LinuxCNC recognises U and W as distinct axes (and different from X and Z), so when it encounters them, it gets upset. A lathe should only have 2 axes in its books.

 

Some controllers want to see G28 moves specified as U and W coordinates, which is possibly how the LinuxCNC post ended up with them, by AD making variations to an existing post where that was expected.

 

The intent behind G28 is different to a simple G53 machine coordinate move. There's a reason why they both exist and why the post offers the option to choose between them. So yes, I've been using the G53 as I had no option but G28 is preferred. Either way, if you selected the wrong option from the post dialogue, LinuxCNC wouldn't run your file.

0 Likes
Message 9 of 11

seth.madore
Community Manager
Community Manager

@Muzzerboy so, where do we stand on the most recent post processor, are there codes that are causing issues for you and other Linux users?


Seth Madore
Customer Advocacy Manager - Manufacturing


0 Likes
Message 10 of 11

Muzzerboy
Advocate
Advocate

Hi Seth - yes, since r43417 was "updated", it generates U & W moves if you select G28 retracts at the end of the job. Anyone running LinuxCNC Lathe with G28 will get an error. The only solution is to either select only G53 or modify the post as I suggested, by pasting in the original content from r43417 into the latest version. It would be better if AD did the latter and released an update.

 

You might question why use G28 rather than G53 and I would point you at a reference like Smid (intermediate moves etc) or (more passive aggressive) ask why offer the G28/G53 as a post option in the first place.

 

It's been pointed out a few times but so far the AD team has resisted correcting the error. I assume somebody on the AD team repurposed a more generic post where U and W are usable for X and Z moves without thinking or knowing better.

 

The issue is that LinuxCNC interprets U and W coordinates as moves in different axes to X and Z, as it's capable of controlling up to 9 axes simultaneously. But for a lathe, it knows you have set it up for only 2 axes. In that context, U and W moves don't compute.

0 Likes
Message 11 of 11

Muzzerboy
Advocate
Advocate

You can see the changes spelled out in this post on the LinuxCNC forum:

https://forum.linuxcnc.org/fusion-360/43561-fusion-360-emc-post?start=30 

0 Likes