Feature Request: ramp piercing (aka moving piercing)

rodw wrote:

I've forgotten now but I would have thought a 20 inch "hole" is not a hole in terms of smart hole processing. For starters, centripetal velocity limits are unlikely to cause a velocity slow down so no special processing is required.
 

 

So the idea on most of these "industrial" systems is to make them as simple to use as possible. The goal is that from CAM to cutting the user does not need experience to make good cuts. How well this is then executed can be debated.
So in an ideal situation:
- the user gives the size of the plate.
- the user opens up a bunch of dxf, dwg etc. files.
- the nesting software nests the parts to the plate, sets correct type and size leadins and places them in the optimal places and sets any special parameters that is required for a good cut.
- then either in the nesting sw or in the cnc you choose the process to be used to cut the plate.
- the hole parameter stuff can be done in either of the systems, CAM or CNC. Monokrom des it in the CNC but could be also done example in sheetcam, or in any CAM, if that is where you set the process to be used.
- the operator sets the zero point of the cutting program and presses cycle start.

The above situation basically eliminates the requirement to know about how to actually cut with best productivity and quality.

Now if you go the lengths to do hole tuning then you might as well do the largest part where you can use standard cut speed and height (this is the easiest of the things to do). Then you already have the best type of leadin, overburn, arc off etc for all the holes from size y and up. This means that you "optimize" not only the small holes but all of them. The user doesn't need to know anything about setting correct leadins, you already did the work to find the best combo.
So when you find the best hole values where you can use AVC and the standard cut speed, the data does not get extrapolated anymore, it just places the biggest hole data to all holes bigger than that. This is why you would have a "pierce distance" so

"Any" experienced operator can cut nice holes, the whole idea is to make it easy and simple for even the most un-experienced guy.

Think it not from a small hole point of view but in a broader way, how can I help the user in all the holes. Then maybe in the future you can move to slots and outer contours.

It becomes quite clear that a large part of high def plasma is that the machine is tightly integrated with the CAM environment.
Here at Linuxcnc we dont have any control over the CAM process. But we can filter the input files and modify the code sent to the machine
As we start to explore higher end cutting features, we are also constrained by the design of plasmac which "locks" the piercing algorithm away from developers like Joco.

I've gone through my emails and have gathered up the discussion we had with around plasma hole cutting and attached it as a PDF here. Its intersting to note that different manufacturers have different methods to connect the arc to the hole centre. In terms of the gcode generated, there is not any difference between an arc or a straight line regarding code generation complexity. I went with arcs as it looked cooler but straight lines might be quicker to cut.

If you wanted to shorten the leadin as you suggest for 20mm holes, you simply would not generate the arc/line to the centre hole. You have to generate the cut path from the end of the cut, working back to the centre, but the gcode flows in the other direction so that is easy. Of course it is also possile to inject additional code at each of the control points.

So to accommodate what you suggest for all holes, say the largest entry in your hole database was 5 x thickness and you were cutting 10mm plate, 
That defines the biggest high def hole at 50mm. But my recollection was that if the hole exceeded the maximum database diameter 50mm in the eample here) the leadins would be unchanged from the original. Following your suggestion, for holes above that max size (50mm in the example), you just need to generate the arc leadin but skip the segment from the centre.

Enjoy the attachment. That is the algorithim Joco baked into Monokrom and I tested with a few cuts.

rodw wrote:

As we start to explore higher end cutting features, we are also constrained by the design of plasmac which "locks" the piercing algorithm away from developers like Joco.

Funny you should mention that, for some time I have been thinking of doing it differently, I have made a start but kinda got bogged down with other things.

phillc54 wrote:

rodw wrote:

As we start to explore higher end cutting features, we are also constrained by the design of plasmac which "locks" the piercing algorithm away from developers like Joco.
Funny you should mention that, for some time I have been thinking of doing it differently, I have made a start but kinda got bogged down with other things.

Don't get me wrong, as it is now, its all pretty cool but there are some restrictions when people push the envelope. There are a number of big machines now running qtplasmac and some of them have oxy and also sometimes a drill as well.

rodw wrote:

It becomes quite clear that a large part of high def plasma is that the machine is tightly integrated with the CAM environment.
Here at Linuxcnc we dont have any control over the CAM process. But we can filter the input files and modify the code sent to the machine
As we start to explore higher end cutting features, we are also constrained by the design of plasmac which "locks" the piercing algorithm away from developers like Joco.

I've gone through my emails and have gathered up the discussion we had with around plasma hole cutting and attached it as a PDF here. Its intersting to note that different manufacturers have different methods to connect the arc to the hole centre. In terms of the gcode generated, there is not any difference between an arc or a straight line regarding code generation complexity. I went with arcs as it looked cooler but straight lines might be quicker to cut.

If you wanted to shorten the leadin as you suggest for 20mm holes, you simply would not generate the arc/line to the centre hole. You have to generate the cut path from the end of the cut, working back to the centre, but the gcode flows in the other direction so that is easy. Of course it is also possile to inject additional code at each of the control points.

So to accommodate what you suggest for all holes, say the largest entry in your hole database was 5 x thickness and you were cutting 10mm plate, 
That defines the biggest high def hole at 50mm. But my recollection was that if the hole exceeded the maximum database diameter 50mm in the eample here) the leadins would be unchanged from the original. Following your suggestion, for holes above that max size (50mm in the example), you just need to generate the arc leadin but skip the segment from the centre.

Enjoy the attachment. That is the algorithim Joco baked into Monokrom and I tested with a few cuts.
 
 

The big machines are not that well integrated with the plasma or CNC. From a pure cutting point of view they are not so different from example sheetcam, its really nothing that special. If you think about it, probably hypertherm is the most integrated as they sell the plasma, CNC and CAM but not everyone buys the whole package. The big OEMs typically fave their own CNC (based example on bechoff) and they sell CAM systems from lantek, sigmanest etc. Then if you go to the big end users (think about steel service centers etc that have multiple machines with multiple technologies) they already have their CAM and wont switch what they already use on all other machines just because the OEM wants to sell them a different CAM. I mean Kjellberg doesn't even produce/sell a CNC or CAM, they only sell the plasma.

I cant comment on plasmac as I don't have the expertise on how its locked or open.

Sorry if my explanation of the pierce distance was poor, you would not remove the first 2 arcs and only have the tangential one, in your case you would just copy the 50mm leadin geometry (along whit all the other hole parameters) and use that as is for all holes bigger than 50mm. The maxed out center pierce remains the same but the start just stays on one of the center axis. Center pierce would place the leadin start on both X/Y axis but in case the hole is bigger than the max, it would "lock" only on example the hole Y axis.

Actually the 2 arcs + tangential arc is faster that the straight + arc, so I would say its better.

But again most if all of these can be done at least 2 different ways. Everything integrated in a CAM system that generates g-code that the CNC blindly follows. Or you integrate these functionalities to the CNC and use a filter like in monokrom. These are done in both ways all over the world, I typically like the CNC approach better as then you are not locked into and rely in a CAM system.

I hope this this gives a more clear idea whats inside my head, most of this stuff is 100x times more easy to explain in person.

EDIT: Sorry didn't actually check the pdf until after typing in my reply, poorly done by me. I assume you meant the the straigth + tangential arc being faster, pdf J, so yes that is faster. That is kjellbergs "1st gen" hole type, now the Q holes are like your 3 arc center start. The 3 arc way is fatser than the pdf line + half circle.

Shifting back to the original topic, is there an interest in me crafting a pull request against the linuxcnc repo for my ramp pierce work? Or shall I maintain  it off to the side as part of an extension for monokrom plasma?

cheers - J. 

Joco wrote:

Shifting back to the original topic, is there an interest in me crafting a pull request against the linuxcnc repo for my ramp pierce work? Or shall I maintain  it off to the side as part of an extension for monokrom plasma?

cheers - J. 

Let me turn it around, is there a reason it should not be included in the repo?

I vote to include it, and for an updated Sheetcam post that supports ramp piercing to be supplied.

Moving Pierce capability is now available in linuxcnc Master (aka 2.10).  PR reviewed and polished with advice/suggestions from PhillC.   Thanks to all who provided input and ideas into this and to Rod for sending me the hypertherm white paper that got me thinking about how this capability could be made to happen within the constructs we have today.

While its not a "use every day" festure for those who find themselves in a position where they periodically need to be able tp push the limits of their plasma source but can not justfiy more $$ down for more AMPs this facility will support you getting a little our of your gear without having to drill holes to allow edge starts.

And the good news is that there is no need for an updated sheetcam PP.  You crate a gcode tool with the needed mwords in the mix and have that insterted at the start of your programme.

Also note that the QTPlasmac documentation has been updated with a moving pierce section that sets out the M159 usage and what the parameters do.

Cheers - J.