Carbide motion automatic homing not a good idea

(Phil Thien) #41

Spelling is not my strong suit, you’d have fooled me had you stuck to just a single spelling (right or wrong).

My spelling is so poor that I’ll often purposely misspell words to make sure my spelling checking is working.

(Phil Gorsuch) #42

I have to agree with @mikep, the thrust of the Carbide3D software tools are to quickly get beginners started, provide robust functionality for the product as shipped, and potentially as a kickoff point to move forward with more sophisticated options if required. We shouldn’t expect Carbide Motion to be as configurable as Mach 4 and similarly Carbide Create should not evolve into something powerful and confusing like Fusion 360 with associated price points. I love my FreeCAD/MeshCAM/CM and CC/CM workflows and really appreciate the easy ramp into CNC but am hitting some limitations given the nature of what I machine and am progressing into a Fusion 360/CM workflow and fully expect to replace CM with another g-code sender (probably to support a custom touch probe and auto levelling) at some point.

That being said if there is a brilliant solution that does not add to user complexity, options, or degrade the out-of-the-box experience I would be all for it.

(Miguel Castillo) #43

I agree with you 100%. Simplicity is good especially for beginners. But the issue we are discussing here is safety which can not be put aside in favor of simplicity.

(Daniel Loughmiller) #44

Simplicity and consistency are both important factors that play into safety. That’s why the specific point I made was about how the machine ‘ends’ the job differently depending on a successful completion or a cancellation and I think either is fine but it should be the same. I can’t imagine there’s a strong argument FOR it being different other than “it’s the way CC/CM have done it in the past”.

(Miguel Castillo) #45

That’s also true and lead me to my other question: Why do we need to home the machine every time we finish a job?. Is it that crucial that the software has to do it automatically?. Just a thought

(William Adams) #46

The machine doesn’t know if it has lost steps or no, and homing again is the only way that we can be certain of starting from a known position.

A lot of this would be a lot easier if we were using some sort of positional encoder — but that’s not a feasible option at this pricepoint/volume. See the LoboCNC for the story of parts availability &c.

(Miguel Castillo) #47

Agree, but we can perform a successful milling operation without homing. Maybe I am wrong, but it is my understanding that all the carving operations are referenced to zero not to home. In any case, if the machine has lost a step or not, the operator should start a homing cycle to “tune the machine” once in a while.

(Jefferson Morris) #48

I completely agree… this should be a user option.

I cannot think of any reason why homing must be automatically initiated after a job completes… can someone else weigh in?

(James Carter) #49

The machine remembers my zero settings, even if I unplug it and move things manually. So yeah, maybe the homing feature is some kind of unnecessary. I hardly notice it though, and like how the machine checks the tool each time.

maybe in the future, we could have “set tool” and “check XY zero location” buttons.

(Jude Marleau) #50

I think it’s not so much the machine (Shapeoko) but the program written for it. CM seems to require it so I would use it, there are other programs that ignor homing completely and they don’t require homing and the spindle stops, raises and returns to zero when completed. I don’t use CM but I am trilled with the Shapeoko’s simplicity and reliablenessability (my kind of word) . The Shapeoko as a kit was the best idea and keeping it as a kit and not some overgrown , takes a college degree to understand over programmed complicated and out of the reach of hobby enthusists is the most important aspect that needs to be maintained for the growth of Carbide3D and the hobby CNC community in general. This allows competition and other hobby programmers to produce alternative cnc programs that work because the machine is simple, reliable and the best machine for hobbists as long as being capable for precision work like that shown by Rich C…Just my thoughts.

(Chaiane Wiggers de Souza) #51

The problem is when you need to machine parts that are already quite “tall/deep”, and with the 3in Z-travel capability of the Shapeoko you don’t get much room to have high enough clearance.

(Dave) #52

+1 on removing the auto-homing on quit, or at least making it configurable.

I often have a bunch of clamps somewhere between the part and the home – and multiple times I’ve finished or quit a job and then had the tool dive right into them on its way back.

I hear you that ideally there’d be no clamp within the safe retract height, but sometimes that’s just not feasible – and having any rapid tool movement that the user hasn’t explicitly asked for seems like a huge safety issue.

Moving to safe Z at current X/Y seems much safer, and I’m not sure I see a downside to the change.

(Tito) #53

What @Dave1 said!

(Blah blah blah to get 21)

(Charley Yancey) #54

late to the game:
soon to enable homing, upgrade to CM4/grbl1.1, etc…
want to echo a couple thoughts of wiser heads before: have spent way more time futzing around with ‘stuff’ than actually making stuff… (NOTE: NOT due to carbide/shapeoko so much as the rest of the computing world) there are so many unsettled links in the chain of production that end up taking hours/days/weeks to figure out, or simply give up and work around it some way…
(or how many simply give up, period ? ? ?)
I don’t mind knowing all the ins-outs of what is a semi-DIY machine, etc; but MAKING thingies was the point of getting the machine, NOT the machine itself… I get a drill, anything more than zero time spent in getting it to drill holes, is a crappy drill… (realizing a DIY CNC is a drill cubed, but still…)
got to say, been around computers my whole adult life, and the -to me- sad state of affairs in CNC software is puzzling and weird… I guess it could be that there is decent proprietary s/w for ‘institutional’ CNC, but my brief exposure to that where a friend worked with a bunch of high-end CNCs, was that their software sucked, too…
I really don’t get WHY there is not a basic, standard function where I can map out where -say- a clamp is so the software takes that into account when computing the (often inefficient) toolpaths… WHY isn’t it SOP that I map out an area of 2"x4"x2" high where a clamp is located, and the s/w automagically routes the toolpaths around ? ? ? I don’t get it, aren’t these computing machines supposed to make OUR lives easier and be able offload all these simple tasks to them ? ? ? seems like a CNC 101 requirement for doing this stuff, and no one every thought of that, or it only comes on zillion dollar s/w ? ? ?

really, this CNC stuff has been around for hundreds of years, starting with jacquard loom punch cards, etc, and THIS is as good as its gotten in over two hundred years ? ? ? yikes…
again, no specific criticism of carbide or its software at all, just a general comment on the disappointing state of the art in the CNC software world…

(Daniel Loughmiller) #55

It’s up to the software to support it, and Carbide Create unfortunately doesn’t.

(Phil Gorsuch) #56

That’s a really interesting point. In some ways I suspect it has to do with the way the toolchain breaks down. When it comes to CAM software which defines the toolpaths it basically knows very little about the machine it is going to run on. In some ways it’s an advantage as the same CAM program can run on just about any machine of the same axis type. The downside is it makes it hard to tailor around certain restrictions of a particular CNC machine or work holding. Similarly the gCode sender/controller knows very little about the geometry of the stock, the workholding, or part it is milling (only step by step directions) and it definitely doesn’t come into play for automated functions like homing. It would be interesting to have some sort of geometrical ‘awareness’ built into the gCode sender (like defining an exclusion zone diagram via carbide create and having carbide motion halt if the gCode indicates a move into one of the exclusion zones) but it sounds tricky.

(Daniel Loughmiller) #57

The thing is if you specify the exclusion zone in CC then CC wouldn’t create gcode that would move the bit into the area, so there’d be no reason for the sender to be aware of it.

There are some common sense checks I think CM could do such as ‘this project will send the bit outside the work area and hit a limit switch, you SURE you want to continue?’

(Phil Gorsuch) #58

Perhaps I expressed myself poorly. You define the exclusion zones in a separate file outside of the main gCode program with a coordinate system that relates to the machine geometry, not the stock geometry of the main gCode.

If machine geometry awareness is all put in say Carbide Create CAM the problem is if I bring gCode from something else (F360) it loses awareness. CC will never match Fusion360 (what I normally use) functionality and arguably it would be crazy to try. Agreed interesting thought on the gCode x-y limits prescreen on the sender once homed and then zeroed on the stock (though subsequent tool changes in one program might be a bear for figuring out Z limits) given CM should/could have a better idea of the approximate full travel limits of their stock machines.

(Charley Yancey) #59

Good point, Phil. in my case, using ArtCam 2018 s/w, and in my twisted way of looking at it, I would define the 2-3-4 hold-down points in the material setup, and the resultant g-code ArtCam generated would -of course!- take those no-go zones into account and automagically route around them… i find things are a lot easier to tell someone else how to fix it, when you don’t know anything about it ! ! !
(and, yes, I have done stoopid newbie tricks like running my v-bit through a nice rockler aluminum clamp arm… that is the very moment I wondered why there aren’t exclusion zones… 8^)


I’ll throw in my nickel here (two cents doesn’t go far anymore) and note that, with these machines, you are trading price for features.

The controller (GRBL based) doesn’t have any extra space for additional functionality, so it would need to be provided either by the sender or the toolpath generator. For enough money, you can buy a CNC machine that does it by itself. CM and CC are pretty lightweight, and exclusion zones can be quite complex to manage, even rectangular ones, as the tool size, shape, and extension come into play.

Fusion, Inventor, Solidworks, etc all have CAM components that will do it easily, but have a heavier learning curve and (except for Fiusion) come at a price