I had a chance to use EMC2 to cut parts on my mill. Aside from still having problems getting the idea of touching off vs. homing through my thick skull, it was a good session. After a fair bit of experimentation, I get the idea of touching off vs. homing, I got semi-automatic toolchanges working with my CAD/CAM software (Vector CAD/CAM), and I even made some parts while I was at it (motor mount blocks for a friend's autonomous quadracopter). It was a good evening.
Tonight I've got a list of "wannas" loaded onto the flash drive I'm using to move files on and off of the mill computer. It's a shame, really, that I have to go that route. EMC2 runs on Ubuntu Linux, which is fully networked. But I can't get a cable out to my shop, and I can't afford wireless hardware at this point. Ah well. At some point it'll happen.
It's odd, along with all the items on my list like "install a spindle encoder so I can do CNC threading" and "finish designing and building the touch probe" I have one overwhelmingly important item: Lights! I have no lights in my shop anywhere near my mill. This is extremely hard on the eyes, and at times I've resorted to using flashlights. I need a lamp on my mill. BAD. Maybe I'll pick one up tonight when I swing by the hardware store.
In the meanwhile, the supplies for my rokkaku kite came in! I'm planning to build it using Gary Engval's plans. I ordered fiberglass spars rather than carbon fiber, more for economic reasons than anything else. But I can always replace them with carbon fiber spars at a later date. The sail will be bright green ripstop with matching green Dacron for the reinforcement patches. The webbing will all be black (I'm a traditionalist, and the spars are black anyway). I can't wait.
There are a number of flying spots on the Big Island I really haven't been able to fly with my Flowforms. Most of the spots on the Kona side are because of wind availability (little of it, and usually of the thermal variety). Most of the spots on the Hamakua side are because there are so many darned trees! Don't get me wrong. I love the trees here. But when you're flying on a beach with only thirty feet of sand between the water and the trees, and the trees are eighty feet tall, that's really not much space. The Flowforms have about a 35-55 degree flight angle, making narrow launch spot flying tough. The okkaku will have roughly a 75-85 degree flight angle, making such flying much much easier.
Another added benefit is that with higher line angle, you can lift the same weight with less line pull. Look at it this way: At a 30 degree line angle, you need four pounds of line pull to lift two pounds of payload (sin 30 = 0.500). At 90 degrees, two pounds of line pull will lift the same two pounds of payload. Not that you get kites flying 90 degrees straight up like an elevator, but kites like the rokkaku and delta get angles that are pretty darned close!
The disadvantage is that a rokkaku has a frame. Flowforms don't. I can stick two Flowforms, two line spools, a KAP rig, camera, transmitter, gloves, spare batteries, etc. into a backpack. The shortest spar on the rokkaku will be three feet long. It's not a backpack kind of kite. But that's ok! With a 32" kite bag and a spare hand, it gives me flight characteristics I haven't had before, and a big pretty green kite to go play with.
Hmmm... Since I've got lots of black Delrin in the shop these days, that might be a fun test for the EMC2 controlled mill: Chuck the Delrin in a drill chuck, use the lathe toolbar on the mill, and have it make nice, semi-spherical end caps for the spars!
Maybe these hobbies play better together than I thought...
Tom
Monday, October 29, 2007
Wednesday, October 24, 2007
More fun with EMC2
So I ran that toolpath the other night. "I'll just run it in the air with no tool," I told myself. Sure... After I saw everything was running smoothly, of course, I loaded a 3/16" end mill, grabbed some aluminum scrap, touched off, and let it rip. Holy cow... I have to say, I'm impressed. The pulse stream coming from EMC2 is cleaner than anything that's ever driven my mill. No missed steps, no locked axes, no nothing.
But the resulting cavity was the wrong size. I more or less expected this. It's something you face every time you change software on a CNC mill. My mill gets 8000 steps per inch (0.000125" resolution), and EMC2's setup files defaulted to 4000. That's first on the list to fix.
I started pawing through the ample documentation (a huge change from the older EMC), and my jaw just about dropped. The change I needed to make was easy. So's adding home switches, which my controller is already set up for, though I haven't added the hardware itself to the mill. So's adding servo enable/disable, which my mill is completely set up for. Three easy changes that make it just like it was under TurboCNC.
Now for the fun part: You can give the mill a pre-defined location for tool changes. So you can use standard G-codes for tool changes, even on a mill that doesn't have an automatic tool changer on it. It will apparently move to the change location, pop up a window telling you which tool to change to, and wait for you to finish before starting the spindle back up. WOW! This I gotta try.
It also should soon have support for a tool height setter. This is basically a touch probe mounted to the mill table that faces up. When a new tool is loaded, it lowers to the tool setter until it touches, saves the Z height as an offset, and then goes back to work. Changing drill bits? No problem. The new length will be picked up. That's NICE.
I can also apparently just finish and plug in the digitizing probe I've had in the works for years. Digitizing a pre-existing part has apparently been in there for ages. Great!
The one that looks the toughest to set up is the spindle encoder, which would let me do CNC threading. I've seen a video of an EMC2-controlled lathe doing some really coarse threading, but not much on how to do it. (In case you're new to machining, "coarse threading" is typically harder than "fine threading" because you're cutting deeper, removing more material, and have to do all this with a slower spindle speed.) I have an indexer on my spindle, but no real rotary encoder. Since I built the indexer myself using a discrete IR interrupter, I can easily change out the optical plate it senses for one with more positions. So even if I can't use my setup as-is, I can certainly change things to suit. It'll happen!
One neat feature I saw was that it's got a feature for touching off on your workpiece. "Touching off" refers to finding one corner of the material and setting that one corner as your (0,0,0) coordinate point. On a manual mill this is typically done with an edge finder. An edge finder is a tool of a given diameter (mine's 0.200" in diameter) that reacts when it touches a workpiece. I've got an old-school edge finder that de-centers itself. But you can get electronic edge finders that light up when they touch off. In any case, the trick is to then compensate for the radius of the edge finder so you know where true zero is. In EMC2, apparently you can tell it the geometry of your edge finder, and when you're touching off you simply jog in an axis until it touches, click the "Touch Off" button for that axis, and it does the radius compensation for you. Slick!
Two features EMC2 has that TurboCNC didn't (at least the version I was using) are constant velocity contouring and cutter compensation.
CVC basically means long strings of short moves (which you get when you're cutting some contoured surface) are treated as a single move, and they're cut at one constant velocity.
In a CNC controller that doesn't support CVC, each move would be treated as a discrete entity, complete with acceleration from a dead stop, deceleration to a dead stop at the end, etc. So a contoured cut could take hours! I made some parts for work on an older system, and what should've been a two hour part took almost all day. It was a pain. I need to make spares of those parts, so I'll get to put CVC to the test in the very near future.
CC lets you change your mind about what tool you use without having to re-code the whole program. For example, if your controller knows you're using a 0.250" diameter mill, and that the material is off to the left, it'll displace the toolpath 0.125" to the right so that the cutting edge of the tool is in the right place. With CC, if you change to a 0.500" diameter tool and let the controller know, it'll displace the toolpath 0.250" to the right so that the cutting edge of the tool is still in the right place.
So far I haven't had much need for CC, but I'd like to see what I can do with it now that I know I have it. My CAM software supports CC, so it'll be neat to try.
Enough rambling for now. I need to start making chips!
Tom
But the resulting cavity was the wrong size. I more or less expected this. It's something you face every time you change software on a CNC mill. My mill gets 8000 steps per inch (0.000125" resolution), and EMC2's setup files defaulted to 4000. That's first on the list to fix.
I started pawing through the ample documentation (a huge change from the older EMC), and my jaw just about dropped. The change I needed to make was easy. So's adding home switches, which my controller is already set up for, though I haven't added the hardware itself to the mill. So's adding servo enable/disable, which my mill is completely set up for. Three easy changes that make it just like it was under TurboCNC.
Now for the fun part: You can give the mill a pre-defined location for tool changes. So you can use standard G-codes for tool changes, even on a mill that doesn't have an automatic tool changer on it. It will apparently move to the change location, pop up a window telling you which tool to change to, and wait for you to finish before starting the spindle back up. WOW! This I gotta try.
It also should soon have support for a tool height setter. This is basically a touch probe mounted to the mill table that faces up. When a new tool is loaded, it lowers to the tool setter until it touches, saves the Z height as an offset, and then goes back to work. Changing drill bits? No problem. The new length will be picked up. That's NICE.
I can also apparently just finish and plug in the digitizing probe I've had in the works for years. Digitizing a pre-existing part has apparently been in there for ages. Great!
The one that looks the toughest to set up is the spindle encoder, which would let me do CNC threading. I've seen a video of an EMC2-controlled lathe doing some really coarse threading, but not much on how to do it. (In case you're new to machining, "coarse threading" is typically harder than "fine threading" because you're cutting deeper, removing more material, and have to do all this with a slower spindle speed.) I have an indexer on my spindle, but no real rotary encoder. Since I built the indexer myself using a discrete IR interrupter, I can easily change out the optical plate it senses for one with more positions. So even if I can't use my setup as-is, I can certainly change things to suit. It'll happen!
One neat feature I saw was that it's got a feature for touching off on your workpiece. "Touching off" refers to finding one corner of the material and setting that one corner as your (0,0,0) coordinate point. On a manual mill this is typically done with an edge finder. An edge finder is a tool of a given diameter (mine's 0.200" in diameter) that reacts when it touches a workpiece. I've got an old-school edge finder that de-centers itself. But you can get electronic edge finders that light up when they touch off. In any case, the trick is to then compensate for the radius of the edge finder so you know where true zero is. In EMC2, apparently you can tell it the geometry of your edge finder, and when you're touching off you simply jog in an axis until it touches, click the "Touch Off" button for that axis, and it does the radius compensation for you. Slick!
Two features EMC2 has that TurboCNC didn't (at least the version I was using) are constant velocity contouring and cutter compensation.
CVC basically means long strings of short moves (which you get when you're cutting some contoured surface) are treated as a single move, and they're cut at one constant velocity.
In a CNC controller that doesn't support CVC, each move would be treated as a discrete entity, complete with acceleration from a dead stop, deceleration to a dead stop at the end, etc. So a contoured cut could take hours! I made some parts for work on an older system, and what should've been a two hour part took almost all day. It was a pain. I need to make spares of those parts, so I'll get to put CVC to the test in the very near future.
CC lets you change your mind about what tool you use without having to re-code the whole program. For example, if your controller knows you're using a 0.250" diameter mill, and that the material is off to the left, it'll displace the toolpath 0.125" to the right so that the cutting edge of the tool is in the right place. With CC, if you change to a 0.500" diameter tool and let the controller know, it'll displace the toolpath 0.250" to the right so that the cutting edge of the tool is still in the right place.
So far I haven't had much need for CC, but I'd like to see what I can do with it now that I know I have it. My CAM software supports CC, so it'll be neat to try.
Enough rambling for now. I need to start making chips!
Tom
Monday, October 22, 2007
Boot to the Head
Sometimes it takes a serious kick in the pants or a boot to the head to get off top dead center. And sometimes the things that should be painful really aren't in the end.
I've got a small benchtop CNC mill. It's an older machine, and runs bi-level chopper drivers on its stepper motors rather than a microstepper, which is more common these days. Because of this it's suffered from a couple of quirks, one of which is resonance at some speeds, which results in lost steps, destroyed parts, etc. Also, because my shop tools are more or less being operated on the basis of "if you can't afford it you can't do it", I don't buy a lot of new hardware for my mill. The computer that drives it is a curbside pickup special on computer recycling day, and up 'till now I've been running TurboCNC on a DOS installation.
It has been painful!!!
Don't get me wrong. TurboCNC is a good product, but it's not much fun running DOS for something that needs files to be moved around. I do my CAD work on a Windows XP machine. I have to get those files onto the CNC machine. With DOS that pretty much says floppy or nothing. I managed to get a DOS setup working that talked to USB flash drives, but you had to reboot the DOS machine in order to disconnect and connect your drive. And since the CNC machine is the only machine in my house with a floppy drive, it has not been fun. It's been so much of a pain I more or less quit using my mill out of frustration at how hard it is to design a part, tweak it, test it, change it, etc. and only then finally make it. It's been more trouble than it's worth. But it's what I had.
Until recently, anyway. I suffered yet another hard drive crash (shops are mean to computers), and realized I no longer even had the installer for the version of TCNC I was using. I hadn't used my mill in over a year because of the difficulties already mentioned, so I figured it was time to start fishing for a new program.
(As a quick aside, while thinking about what to do I disassembled the dead drive, a 2GB Samsung. I got two really nice flat round plates I can use as index plates or robot wheels, I got a brushless DC motor, some bearings, and a killer pair of fridge magnets!)
After a little hemming and hawing, I stuck a used 13GB drive in the machine, grabbed a copy of the EMC2 Brain Dead Install CD, and gave it a whirl. I am impressed! I had a bad experience with EMC in the Y2K era, so I had a lot of misgivings going into it. I have to say, though, my fears were utterly unfounded. This thing is impressive!
The developers of EMC2 seem to have fixed just about every EMC bug that drove me nuts back in the day, and in the exchange I get constant velocity contouring (CVC), much faster travel, a really nice interface that displays your toolpath before you commit to running it, cutter diameter offset, cutter length compensation, and all sorts of other goodies I haven't even begun to find out about yet. It's COOL!
In a little over half an hour I went from a dead mill and a dead computer with no hard drive to a really slick mill that'll do 50 inches per minute, whiz the rotary stage like a top, cut deep and clean, and let me use networking, file sharing, USB flash drives, etc. I'm in heaven!
So what's a feller to do? Make parts, of course! I don't have anything lined up, but this is serving as a perfect opportunity to re-learn my CAD/CAM software. That, too, has changed since I last used my mill. I draw things in Rhino3D these days, and generate toolpaths in Vector CAD/CAM. I whipped together a quickie to make a turner's cube on the mill, and will give it a try tonight in a couple of different materials. It'll test circular interpolation, backlash compensation, and all sorts of other things that more complex parts will need. In the end I'd love to make some of these in brass, throw them in the mass finishing setup, polish them to a mirror finish, and get them plated so they don't oxidize. Should be pretty!
What's down the road? In the short term I'd like to finish out the deck fittings for an RC sailboat, make a new backpack winder for my kite with fold-down handles, make a new chassis for my line follower and downhill racer robots, make a Delrin case for the Orangutan and for the AVR Dragon, make some gear cutters out of some O1 tool steel I've had lying around forever, and see what I feel like doing from there.
MAN it's nice to have my mill back! I can't wait to start playing.
Tom
I've got a small benchtop CNC mill. It's an older machine, and runs bi-level chopper drivers on its stepper motors rather than a microstepper, which is more common these days. Because of this it's suffered from a couple of quirks, one of which is resonance at some speeds, which results in lost steps, destroyed parts, etc. Also, because my shop tools are more or less being operated on the basis of "if you can't afford it you can't do it", I don't buy a lot of new hardware for my mill. The computer that drives it is a curbside pickup special on computer recycling day, and up 'till now I've been running TurboCNC on a DOS installation.
It has been painful!!!
Don't get me wrong. TurboCNC is a good product, but it's not much fun running DOS for something that needs files to be moved around. I do my CAD work on a Windows XP machine. I have to get those files onto the CNC machine. With DOS that pretty much says floppy or nothing. I managed to get a DOS setup working that talked to USB flash drives, but you had to reboot the DOS machine in order to disconnect and connect your drive. And since the CNC machine is the only machine in my house with a floppy drive, it has not been fun. It's been so much of a pain I more or less quit using my mill out of frustration at how hard it is to design a part, tweak it, test it, change it, etc. and only then finally make it. It's been more trouble than it's worth. But it's what I had.
Until recently, anyway. I suffered yet another hard drive crash (shops are mean to computers), and realized I no longer even had the installer for the version of TCNC I was using. I hadn't used my mill in over a year because of the difficulties already mentioned, so I figured it was time to start fishing for a new program.
(As a quick aside, while thinking about what to do I disassembled the dead drive, a 2GB Samsung. I got two really nice flat round plates I can use as index plates or robot wheels, I got a brushless DC motor, some bearings, and a killer pair of fridge magnets!)
After a little hemming and hawing, I stuck a used 13GB drive in the machine, grabbed a copy of the EMC2 Brain Dead Install CD, and gave it a whirl. I am impressed! I had a bad experience with EMC in the Y2K era, so I had a lot of misgivings going into it. I have to say, though, my fears were utterly unfounded. This thing is impressive!
The developers of EMC2 seem to have fixed just about every EMC bug that drove me nuts back in the day, and in the exchange I get constant velocity contouring (CVC), much faster travel, a really nice interface that displays your toolpath before you commit to running it, cutter diameter offset, cutter length compensation, and all sorts of other goodies I haven't even begun to find out about yet. It's COOL!
In a little over half an hour I went from a dead mill and a dead computer with no hard drive to a really slick mill that'll do 50 inches per minute, whiz the rotary stage like a top, cut deep and clean, and let me use networking, file sharing, USB flash drives, etc. I'm in heaven!
So what's a feller to do? Make parts, of course! I don't have anything lined up, but this is serving as a perfect opportunity to re-learn my CAD/CAM software. That, too, has changed since I last used my mill. I draw things in Rhino3D these days, and generate toolpaths in Vector CAD/CAM. I whipped together a quickie to make a turner's cube on the mill, and will give it a try tonight in a couple of different materials. It'll test circular interpolation, backlash compensation, and all sorts of other things that more complex parts will need. In the end I'd love to make some of these in brass, throw them in the mass finishing setup, polish them to a mirror finish, and get them plated so they don't oxidize. Should be pretty!
What's down the road? In the short term I'd like to finish out the deck fittings for an RC sailboat, make a new backpack winder for my kite with fold-down handles, make a new chassis for my line follower and downhill racer robots, make a Delrin case for the Orangutan and for the AVR Dragon, make some gear cutters out of some O1 tool steel I've had lying around forever, and see what I feel like doing from there.
MAN it's nice to have my mill back! I can't wait to start playing.
Tom
Thursday, October 4, 2007
Taking Time
I'm back to making parts at work. The past two weeks I've been designing and making parts for a temporary instrument. It's only likely to be used a handful of times, but its purpose is to demonstrate whether a particular idea works. So the results from the experiment could steer the course for future instrument development. It's been a fun project.
Today I made a part for a completely different instrument, one of our permanent ones. Every time it's taken down and set back up, there's an alignment that needs to be done to better than 0.7 microns. (Yeah, better than the wavelength of red light...) Oddly enough, the instrument had no facility for making this adjustment! The "proper" method was to knock a mirror into position, then fine tune it with the tension on the clamping screws. Ergh...
We've been kicking ideas around for a while, and a couple of attempts have been made, none all that successful. But today I took the drawings for the latest design and parked myself out in the shop. It was a good day.
The part's basically a ring with three clear holes and three threaded holes in the front face. So far so good. Some time on the lathe making all the different profiles on the part and some drill work on the mill, and the basics were done. For the next trick, it required four holes spotted, drilled, and tapped, at a 38 degree angle from horizontal, all around the periphery of the part.
It's really tempting to take shortcuts. Really tempting. But this time I resisted. Rather than file the bevels on the edges of the part, I used the cross-slide and did it right. Rather than try to C-clamp the thing down for drilling, I made a fixture and held it down with a screw. Rather than try to fudge that 38 degree angle, I made a gauge block and used that to set the part up in all four orientations. And rather than trust to fate with a center drill, I actually stuck to the drawings I'd made and spotted a 5mm flat at the location for each hole.
While I was at it I added some witness lines on top of the part so you could tell what axes the push-push screws were going to move the mirror. In the end it wound up actually looking nice!
But the proof is in the pudding. That instrument is coming off tomorrow, so we should have some opportunities to test it in the not-so-distant future. I hope it works. This has been a major thorn in my side. I have one of the other failed attempts sitting on my desk at the moment. It's a removable kinematic mount design that should've worked, but didn't because I took a shortcut with the geometry and screwed it up. No more short cuts! Take the time and get it right!
Tom
Today I made a part for a completely different instrument, one of our permanent ones. Every time it's taken down and set back up, there's an alignment that needs to be done to better than 0.7 microns. (Yeah, better than the wavelength of red light...) Oddly enough, the instrument had no facility for making this adjustment! The "proper" method was to knock a mirror into position, then fine tune it with the tension on the clamping screws. Ergh...
We've been kicking ideas around for a while, and a couple of attempts have been made, none all that successful. But today I took the drawings for the latest design and parked myself out in the shop. It was a good day.
The part's basically a ring with three clear holes and three threaded holes in the front face. So far so good. Some time on the lathe making all the different profiles on the part and some drill work on the mill, and the basics were done. For the next trick, it required four holes spotted, drilled, and tapped, at a 38 degree angle from horizontal, all around the periphery of the part.
It's really tempting to take shortcuts. Really tempting. But this time I resisted. Rather than file the bevels on the edges of the part, I used the cross-slide and did it right. Rather than try to C-clamp the thing down for drilling, I made a fixture and held it down with a screw. Rather than try to fudge that 38 degree angle, I made a gauge block and used that to set the part up in all four orientations. And rather than trust to fate with a center drill, I actually stuck to the drawings I'd made and spotted a 5mm flat at the location for each hole.
While I was at it I added some witness lines on top of the part so you could tell what axes the push-push screws were going to move the mirror. In the end it wound up actually looking nice!
But the proof is in the pudding. That instrument is coming off tomorrow, so we should have some opportunities to test it in the not-so-distant future. I hope it works. This has been a major thorn in my side. I have one of the other failed attempts sitting on my desk at the moment. It's a removable kinematic mount design that should've worked, but didn't because I took a shortcut with the geometry and screwed it up. No more short cuts! Take the time and get it right!
Tom
Subscribe to:
Posts (Atom)