I am brand new to this, so it is probably both stupid and also posted in the wrong place. I am happy to be corrected!
Is there a reason we can’t define two X,Y,Z 0 points in order to inform the machine the location, thickness, and orientation of the stock (three would even allow for size assuming it is a rectangle). It seems to me this would solve many registration issues for tiling and two sided cuts, would avoid issues of imperfections with the waste board, and would conserve on materials and time.
I’m sure there are very good reasons, but I’m a newb and based on why learning curve thought I would at least throw it out there.
I’m not 100% sure I understand everything you are asking, but here is my best interpretation.
In general, CNC machines do not think. They are tools to reliably and repeatably execute specific instructions (gcode) written by your CAM software, for your specific machine — using a machine-specific post processor. It sounds like you want the machine to “remix” (for lack of a better word) already processed gcode to a new coordinate system, one which may differ from the actual machine orientation. This is simply not possible at this time. Any material positions or properties should be defined in CAM, prior to creating the gcode.
Fences and bench dogs are well-established ways of quickly and consistently placing material, and can make two-sided milling quite easy. Other, perhaps more advanced, methods include using dowels or pins.
Zeroing from the spoilboard surface instead of the material surface can help to increase the longevity of your spoilboard, though I’m not sure this was the actual meaning of your statement.
Thank you for the reply. Your answer to the first quote answers my question. I understand the use of bench dogs and fences, pins and etc. But they don’t strike me as all that great of a way to do things, more like hacks to get a system to work rather than stepping back and thinking about what we are actually doing. I acknowledge that this is due to me coming to this using slicing/tiling and two sided cuts to make large non-symmetrical objects and considering doing multiple pieces per cut for the sake of efficiency. Probably not a very common use case.
The machine dictates the orientation and location (0,0) of the system. We then use the machine to face the spoil board in order to get that in the same orientation. So far so good. If we are doing one sided cuts that’s the end of the story. But if we are doing two sided cuts the x,y plane of the machine has essentially been mapped to the stock. If the stocks orientation or location changes, no good. So we come up with various registration systems to make sure the stock doesn’t move relative to the machine. But that seems like a sludge when you can use the machine itself along with a sensor to measure the orientation of the stock, which is the thing we actually care about.
So I was just wondering why there isn’t a mode (in addition to the current setup, not instead), where we did that. Use a zeroing tool to get the XYZ of the lower left and the XY and maybe Z of the upper right, and map that to the machine? It would seem like it would just* be a transfer function to translate the machines XY into the stocks XY. as long as you have those corners you just do that before each cut.
For 2 sided milling and tiling this would void the issue of using clamps/pins/etc to register the location of the stock. Instead you would just need to clamp the piece for secure cutting and could also open up possibilities in terms of 3-6 sided cuts and reduce the precision of level/flatness required of the spoil board.
But as you said. It can’t be done right now. Fair enough. It was just a thought that popped into my head as I am gearing up to do many, many two sided cuts and I am spending a lot of time making sure the stock doesn’t move rather than machining.
“Just” is doing a LOT of work here. It would certainly take me more than a bit of time to brush up on my math/ coding skills to be able to write the function.
Edit to add. Cut paths might be a traveling salesman problem and if so then that transfer function would be WAY harder than I was thinking.
Your ideas are interesting. And as you say, it’s just math, so it’s certainly something that could be achieved by someone willing to put in the time and effort. I’m just not sold on it being necessary, but I’d love for you to prove me wrong.
For what it’s worth, I have made a lot of two-sided pieces and have never felt any of the methods mentioned to be hackish or a sludge — but maybe I don’t know what I’m missing. With the way I set up my projects, all I have to do to carve the reverse side of a piece is to flip the material over, then align it against the same bench dogs I used for the front side. It’s so easy, I don’t even probe again, let alone twice more. But to each their own.
This maybe what you’re looking for. If your stock has squared corners, then you do not have to use the dowel pin method for a two sided carve and still don’t need to know the exact dimensions of the stock. But you do need fixtures to keep the stock aligned with the machine x and y axis and zero twice. I use right-angle fixtures that can be mounted on the wasteboard using the 20mm hole pattern bored by the CNC.
For example, on side 1 clamp the stock to a fixture so it’s aligned with the x axis and zero off the top lower left corner as usual. To do side 2, flip the stock along the x axis and zero off the bottom lower right corner. Don’t have a probe to zero off the lower right corner? Zero the lower left corner of a precision 123-block in a right-angle fixture and then offset the x zero by the 123-block x dimension. Stick the stock corner in the fixture and clamp. This method can be extended to 3,4, or 5 sided carves.
You’re probably right. It is most likely that I am still developing my skills on the machine and am just not being precise enough with my setup in terms of bench dogs, etc… I think it is also that I am looking not at making single pieces but 100s of them for this project and having an easy way to be able to put 32"x48" stock with 5-10 different parts is attractive.
Really it was just a thought that popped into my head. One of those this is either an interesting idea, or totally stupid that I have on occasion. I’ve learned to throw them out there as otherwise nothing would ever happen. I’m leaning more towards this one being stupid in terms of a work/reward perspective…
I’ve only done one two sided project. What I did was make sure to leave the lower left corner on the project after side one was carved. Then flip it over and secure it making sure the bottom is parallel to x axis. Then I zeroed the 1F on the lower right hand corner (that was the lower left until I flipped it). When I generated the g-code for the second side I set the lower right as the ‘zero’ corner. To zero off anywhere but the lower left you can’t use the touch probe for X and Y unless you buy the upgrade. I did it visually and then used the touch probe to set Z. If I’m setting X and Y visually I use a V bit and a magnifying glass (my eyes aren’t as good as they were 20 years ago). Most people think that they need to get it within .001" but if working with wood (which is all I do) that’s far too precise.