Printable View
Hi
Just thought i would try to gather some thoughts on where on a CNC Mill you would mount proximity switches, some say "HOME" should be in the centre of the X&Y Axis, some say at the extreme + end of each Axis, i would like to hear your thoughts.
Ray
I have my XY homes at the extreme of the table. You have to tell your controller which direction to go in the setup for home. If your position is on the wrong side of home if you have it in the middle, the controller will go off in the wrong direction until it hits the end of travel and not find the home switch!
On the other hand, work coordinates can be any place.
I have my XY homes set to the 0,0 on the X,Y coordinates and top of travel on the Z axis. It is just a personal preference, as I suspect a lot of set-ups are.
When I was considering CNC'ing my little Clarke CMD10 (which never happened) I thought about having a toggle switch on the centre of each of the X and Y, with spring loaded flaps to operate them. The idea was that you then had some input to say which part of the table you were on and could home accordingly to the centre.
Many great inventors start off with what other people rate as stupid ideas. My end point seems to be the stupid idea.:stupid:
Hi, where would your table be when at "0" ?Quote:
Originally Posted by cropwell
As I look at the table, 0,0 is nearest me and to the left, just as though it were a graph on a piece of paper. I have a gantry machine BTW.
I can't see how it matters in reality, the home position is just a datum point that is repeatable. If you want to use the centre as zero then just apply an offset in your CAM software.
My home position is head all the way up with the spindle centre over the back right hand corner of the envelope.
I have Home Offsets programmed into Mach3 which then set my zero at X & Y centres of movement with my standard edge finder touching the table surface (with it's tool offset selected), this way when you Ref All Home Mach3 goes and finds the home switches then applies your Home Offsets.
Then you have multiple Work Offsets which allow stored settings for all 3 axes for clamps, vices, fixtures, standard workpieces etc.
Then you have Tool Offsets which allow stored settings for the lengths of the various standard tools you use.
Once you've got your head around it all set-ups become far faster and simpler, Ref All Home the machine, select the tool of your choice and the work offset of your choice and your zero can be where you want it in little more time than it takes to fit the tool and position a vice on the bed,
- Nick
I found the huge disadvantage to using G code was you threw the original drawing away. I changed to cutting from the .dxf which meant I could see the original shape on screen. I then made circles into datum points. Circles that have depth in the Z axis cut, those without depth merely exist.
If starting from a blank piece of metal, I print the .dxf then stick it on with Pritt. I like Pritt because it dissolves in soluble oil.
Left mouse moves the table to a circle centre XY when you click on it, right mouse sets the XY position to a circle centre XY when you click on it.
Robin,
That sounds like an interesting thread you're responding to, which one is it?
- Nick
Sorry Nick, it is this thread, I just couldn't be bothered to add the logical progression that makes sense of it. I have tried but nobody here programs computers so they cannot escape the limitations of using MACH and G-Code.Quote:
Originally Posted by magicniner
Robin,Quote:
Originally Posted by Robin Hewitt
I think it's worth expanding your reply. I cut my teeth on Z80 Assembler and progressed to PLC Ladder Logic & Statement List, VB, SQL, Crystal Reports, etc. and you've lost me.
Any "Limitations" of Mach3 and G-Code (and I've not found them yet) are far outweighed by the capability provided by an affordable CAD/CAM system coupled with Mach3 -
-
http://i135.photobucket.com/albums/q...ps2af5f84b.jpg[/URL]
-
http://i135.photobucket.com/albums/q...psa585817f.jpg[/URL]
-
http://i135.photobucket.com/albums/q...psab81830a.jpg[/URL]
?????
The stamp does this in hot steel -
-
http://i135.photobucket.com/albums/q...ps62609573.jpg
-
It's about 8mm high ;-)
Quote:
Originally Posted by magicniner
You say "etc." before you say "Visual Studio". Windows is event driven and will never guarantee to return control to you in time to step the motors. The moment you give in and add a separate processor to sort out the timing problems MACH becomes a totally redundant unnecessary burden and you may as well run Win8, check your e-mails and have a Shoot'em'up game running in the background to while away the time while it cuts.
I don't think I want to criticize MACH here too much, look what happened to the bloke in the news today, the one who accepted a Gaza appeal to be published in the Jewish Chronicle.
That was my biggest hurdle when moving from DOS programming to Windows, instead of main() running in a continuous loop, it just had to sit there until there was an event.Quote:
Originally Posted by Robin Hewitt
Will Windows interrupts be too slow ?
Windows will not let you take over a timer interrupt, you can hook into a 1kHz system tick but that isn't fast enough. This came to a head when Windows could no longer guarantee to service a serial interrupt in time to stop an over-write with only one level of buffer, they needed 16.
Brilliantly put whilst totally avoiding the perfect opportunity to identify your "personally programmed" alternative where you haven't given up hardware control to Windows. You are signally failing to make a point here.Quote:
Originally Posted by Robin Hewitt
I'm really intrigued, what system are you using? How much percentage wise and specifically which bits are your own code?
If it turns out that this is an OS bias debate then I'll laugh my socks off as all that really matters is how quick and easy a system is to get running reliably, what the software support is like and what results it produces ;-)
Not sure I understood too much of that. The system I am using is Win7, it is 100% my own code and it certainly does not run reliably :hysterical:
Robin, this DOS based any use ?
http://www.dakeng.com/turbo.html
I'm not Robin, but Turbocnc is my controller of choice. I have been using it since about 2000. I used it for quite a while before I registered it. It is fully functional when you download it. Registration gives you access to the source code so you can make your own changes. It is written in pascal and compilers are available as freeware. Doesn't have a lot of bells and whistles, but it gets the job done and as far as I am concerned, it does it quite well.
Unless you take over windows.!!. . . . Tell Art fennerty it can't be done.! . . . . .The Boffin's said Mach3 could never be done but Hey presto.. .:cheerful:Quote:
Originally Posted by Robin Hewitt
Thanks, handy to know it works okay, I think I still have Borland Turbo Pascal for DOS so might still be useful :glee: and I have a 5 1/4 floppy drive for the media but don't know if it will work with Windows 7.Quote:
Originally Posted by C_Bubba
Yep, Mach3 own high priority driver takes care of that. I'm not sure it can work with Win7 and 8. Although I've read some people use it under Win7.
As for the coordinates of the machine, my personal opinion is that the near left corner relative to the operator's most convenient position should be assigned X0/Y0. It does not matter which axis is the longest one. Thus you maintain correspondence to the CAD and CAM software. Setting work zero is another thing and depends on where you fix the material on the table. For long machining sequences I zero the machine by means of home switches (Mach3) and then write down the real (machine) coordinates of the work zero. Just in case something goes wrong I am able to reposition with acceptable accuracy.
The current version requires full blown boraland 7 (which can be had at <http://vetusware.com/download/Pascal%207.01/?id=3724>. I don't remember the reason Turbopascal won't work, but that is what the authors told me. V7 and windows7 won't cooperate either, so you will need to run that in a virtual box.
There are also more recent compilers out there that should work but I haven't tried them yet.