I'm with Peter on the precision/accuracy/repeatability (PAR) issues. I'd really like to see PAR at 5 "tenths" or less (5 ten-thousandths, something I picked up a while back from an old machinist. It's .0005", much less than Peter's goal), which should be achievable by those of us with more mechanical experience and/or money to spend. 1 mil (0.001) should be the minimum acceptable PAR for anyone building the machine.
The steppers should really be driven with a chopper drive. 75-100oz-in of torque would be good, especially if folks are using cheap threaded rod, which is notoriously inaccuracte. Microstepping is an approach I don't think we can afford to ignore. Yes, direct drive will be slow. DC servos are faster, but good encoders that will allow us the PAR we need will make the project cost prohibitive for some, and I'd like us to avoid that if possible.
Again in agreement with Peter, I think an X-Y table will make the system hard to build and sloppy. A gantry is much simpler to put together, and has much better weight distribution on the x-axis. It also allows the workpiece to sit on a flat, unmoving plate, easing the need for fixturing and simplifying alignment of the 0,0,0 point.
Something like this:
These types of machines lend themselves to easy construction from t-track (the 8020 stuff I mentioned a few posts up) or plate construction. The twin leadscrews for the major axis can be slaved together with chain or a toothed belt and driven from a central motor. I think that once the parts are assembled, a machine like that can probably be put together in a matter of hours by someone with a modest amount of mechanical skill.
I've built X-Y tables before, 3 of them. A gantry is simpler to build, more stable, and easier to maintain and clean.
As always, my highly opinionated $0.02
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
Certainly, if this is to be a bench tool....a moving table requires more room than a gantry system.
I like the overhead approach.
Also just thinking....lasers a very popular in a lot of tools these days.· (really more of a novelty marketing ploy).
But how about using a photocell and a laser to aid in alignment of the table -- basically, if you had the gantry/table centered, you could shine the laser thru some portion of the table and if it's in alignment, you'd have the photocell, opposite the hole, picking up the laser light.
Or perhaps using to lasers to act as a visual crosshair (so you'd know where the drill is supposed to go)....all these require aligning themselves of course and are definately icing on the cake once things are running!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ ·
Steve http://ca.geocities.com/steve.brady@rogers.com/index.html "Inside each and every one of us is our one, true authentic swing. Something we was born with. Something that's ours and ours alone. Something that can't be learned... something that's got to be remembered."
As cool as laser alignment would be, it's difficult. Inexpensive laser beams have width and divergence, both things that we would want to avoid in an alignment system. I think we can get a whole lot more bang for the buck and a lot more durability out of 3 limit switches conveniently placed at various points on the carriages. Now a laser set up as a visual indicator might be fun. Hmmm....
And I misspoke, it's easier to build a gantry with a single leadscrew down the center with two carriages on eeither side than placing screws at the carriage locations.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
A simple laser through a hole into a photocell does not work very well. If you put a 4 way optical splitter in the path after the hole with 4 as identical as possible pick-ups, and then move the table so as to balance the detected light among all 4, will work for lower degrees of accuracy.
I mounted a pin hole camera on one of my units, and that was adequate for aligning to pad centers. I added a video cross hair generated by an SX; that works great.
As for limit switches, by themselves they are woefully inadequate for positioning the reference. They will have up to 10 or even 20 thou of slop. You need to add a rotary index to the lead screw as do the commercial units. They have at a minimum one "reference" pulse per rotation. The limit switch is then used (is adequate for) to determine the reference turn on the screw, and the encoder's reference pulse gives you the count start position.
In using stepper motors one uses the limit switches to get "close" (within one turn) of the reference position, and then a rotary mechanical stop (not a jamnut) to bottom out the stepper. This works very reliably to the precision of a single step.
Not wanting to pour cold water on an interesting challenge, but all this is not exactly trivial to implement when one expects the precision Dave is hoping for. In fact Dave, I think 5 tenths of a thou is well beyond most hobbyists' limit unless they have a pretty competent machine shop at their disposal.
Peter,
That is why I suggested that the initial machine be geared towards drilling. It's not that hard to chemically etch a PCB, but drilling all the friggin' holes it what drives me insane. For drilling I think 0.005 is good enough.
Bean.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"SX-Video·Module" Now available from Parallax for only $28.95
Not wanting to pour cold water on an interesting challenge, but all this is not exactly trivial to implement when one expects the precision Dave is hoping for. In fact Dave, I think 5 tenths of a thou is well beyond most hobbyists' limit unless they have a pretty competent machine shop at their disposal.
Peter-
I know, 0.0005 is pretty darn small, but having played with this a few times (never quite making it to the full blown CNC stage) I think it is within the reach of determined amateurs building small scale machines. Using commercial extrusions and a machinist's square, I put together a 14" square base that was flat to no less than 0.001" (corner to corner, laterally it was closer to 0.0007")over a weekend a few years ago. It wasn't much more difficult than spending half an hour verifying my power miter saw's alignment and making a few test cuts. I don't thing that doing half a mil better is insurmountable, even for a garage hobbyist like me. It just takes persistance.
I'm willing to take a chance in trying it. I think the base is the easier of the two parts I need ot hit my 5 tenth goal..the tougher one will be the PAR related to the positioners and the actual tool spindle. Achieving one tenth (0.0001") of TIR will be a real task I think. I have a few ideas I'm looking into, but I can't say for sure if any will be even remotely correct.
As an aside, this is precisely why I tried to distance my personal goals for the project from the community goals. One mil is still something I consider to be a minimum number for PAR for the average implementation, desapite my own lofty goals.
10 mils (100 tenths, ten thou, 0.010", .254mm, etc) is an unacceptable error when milling closely spaced pads for something like an SX52 (11.8mil wide pads on 21.6 mil centers, worst case). Even if the boards are pre-etched and we're just drillingholes, if we have folks using 25 mil vias, that's still a 40% error. I may be the only one here doing 7mil lines/spaces and 15 mil vias, but being 10 mils off for someone doing 25 mil lines/spaces can be nearly as bad as it would be for me. a 25-50% hit in line width can be very bad.
Obviously, the minimum spacing of board features will be governed by the PAR of the machine, but I don't think 1 mil is an outrageous goal for the average hobbyist, is it?
I think this is one of the specifications that will be most difficult to nail down, since it will require an actual impelmentation to measure and verify.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
No, I think 1 thou is well beyond all but the most persistent and finniky hobbyist. Most would not have the measurement techniques nor facilities to even determine how square, level or long their pieces are. You can only do so much with a vernier caliper, even a digital one. Tools for measuring distances over 6 inches get expensive very rapidly.
I do share your views though in needing accuracies approaching your numbers for a serious product, and anything much worse than your goals would be an interesting exercise, but of marginal real value. Of course every one will have their own expectations, and hence different views on precision needs will prevail.
For myself, I would be disappointed with precisions of 5 thou, satisfied with 2 or 3 thou, and elated with 1 thou. I can't imagine getting to 5 tenths.
As far as the spindle is concerned, I am less worried about that. I would pursue an avenue of using NOT a brushed motor; I can't stand the racket. As very little cutting power is required, I would look to invent a spindle driven by something brushless and ratioed up with several pulleys and belts (O rings?). It would be nice to be able to get variable speed to 50K RPM, but bearings might become a serious issue.
I believe dentist drills are air driven, and come in a throw-away module including the bearings and turbine, and run up to 200K, all for not much money. This could be an interesting avenue to investigate for the high speeds/low loads required for mechanical PCB dry "etching".
Please use something else other then a dentists' drill. Or supply a large stereo with the final product....I don't want to cry and whimper everytime I try to drill some boards!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ ·
Steve http://ca.geocities.com/steve.brady@rogers.com/index.html "Inside each and every one of us is our one, true authentic swing. Something we was born with. Something that's ours and ours alone. Something that can't be learned... something that's got to be remembered."
I think 5 thou/mil should be easy, especially if we can find a source of positioning actuators that aren't hardware store threaded rod. I've done enough with that stuff to hate it with a passion. It's not round or straight, and the threads are rarely even close to smooth.
I'd much rather spend time trying to dig up a source of 3/8 or 1/2" Acme thread rods and anti-backlash nuts. Same goes for positioning slides...you can get enough THK and OKI slides to do a small gantry mill (12"x12"x2" or so) on eBay for under $200, sometimes less if you're lucky, and I have yet to see one that is more than 0.001" over it's length, they're usually less.
I'm with Steve on the dental drill idea. There are much better ways of getting a high speed spindle going, and probably with more precision. ABEC-7 skateboard bearings (available for a few dollars each online) have a TIR of no more than 2.5 uM (micro meters, 0.0025mm, 0.0000918"), more than sufficient to support a very high speed spindle. A 2 stage belt drive (Kevlar, rubber, o-rings, whatever) with a normal DC motor motor (8k-20k, 20k-40krpm) could do the RPM conversion pretty cheaply, and I think with a 1/4" spindle the moving mass woud be low enough that things would spin up and down pretty fast. A few pieces of extruded angle 4-6" long could support the whole shebang with good PAR as well. Hmm...
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
Paul Baker said...
I've stayed out of this thread primarily because the scope of the project is a bit daunting
You have hit upon one of the great challenges of this project.
Scope is the key word here.
Your average "Joe" is not going to build an industrial milling machine with a .0005 tolerance.
It would be unkind to lead them to believe that they could. There is a good reason that a "cheap"
CNC machine cost $10.000. That money is for man hours / development not material.
The average hobbyist PCB board has a space between traces in the order of 1/32 to an 1/8 inch.
The overall goal here is to design something that everyone can build.
And that means Simple. Which in turn means not incredibly accurate.
The design goal is to create a solid design (software and hardware) that can be
build upon for those with the wherewithal to take it to the next level.
Our immediate goal should be to create an elegant design that can drill lots of
holes with a "design eye" towards a respectable milling machine.
LostboY said...
I still would love to help any way that I can....
I'll be sure to take you up on that offer. [noparse]:)[/noparse]
steve_b said...
I am hoping that the group is quite public about their progress and that I get to learn along!
That is definitely my intention.
steve_b said...
In my dreams (told you everyone has one) I would hope that there could be a learning mode for this setup.
Now that is why it is a community project. I did not think of that.
That is more into the phase three of the project thought.
pjv said...
In my humble opinion, for a seriously functional machine, the following are some minimum requirements.........
Those are some humbling facts. Thanks for that.
Especially the notes about stepper voltage.
pjv said...
DO make the X and Y motions relative to the base,
Are you saying travel the head across the table? A oppose to the table traveling under a stationary head.
Dave Paton said...
I think an X-Y table will make the system hard to build and sloppy. A gantry is much simpler to put together
That would seem to be another recommendation for a stationary table.
steve_b said...
Certainly, if this is to be a bench tool....a moving table requires more room than a gantry system.
OK, OK, I get it.
Bean said...
That is why I suggested that the initial machine be geared towards drilling
I'm with you on that one.
I would hate to design something that only the elite hobbyist could build.
As for etching, I still can not get that resist pen idea out of my head.
Lets see, what did we learn today.
A moving table is probably out of the question.
Our accuracy needs to be in the single thousands of an inch. Much less than 1 thousands seems out of reach.
We need a simple solution to positioning. I'm starting to wonder if we can solely rely on the stepper motor.
We need to start designing the mechanical layout of the machine.
Lots of good stuff there that will help to put some repeatable precision into the design, and not too expensive either.
With that product as a reference point, I think quite a number of the mechanically inclined hobbyists will be able to come close to the (my) 2 or 3 thou goal, and some of them with better tools will get to 1 thou. (I still have trouble believing it can get better than that, but never mind)
I agree with you on the spindle other than bearings at over 20K. I have a commercial 4 headed PCB drill with tiny direct drive 3 phase motors running at up to 60K, and you should see the trouble I have in getting replacement bearings that will last. Roller blade parts won't cut it for long. I'm contemplating a conversion to air bearings.
Since I have some spare spindle motors, I will probably adapt one of them into my system.
So now the hard part is to secure an inexpensive source for lead screws and anti-slop nuts. I wonder if someone knows of a good low cost source for recirculating ball nuts and matching lead screws. For me the diameter should be minimum of 3/8 or 1/2. The thread pitch should be 0.2 inches as that will nicely match a 200 step motor giving one thou per full step. Anyone requiring better than that can microstep.
The 8020 stuff is great for structure, but their linear motion components have to much stiction for out uses. (UHMW + anodized Aluminum = not smooth, even with teflon spray). The extrusions and brackets themselves are really handy for this kind of thing. 8020 guarantees them to have straightness better than 0.0125" per foot and 0.120 per 20' section, but I've found them to be much much better than that.
Air bearings would be nice, but their manufacture for a hobbist is probably out of the question. ABEC-7 or -9 skate bearings have amazing runout, and if kept clean and oiled regularly, they should last plenty long at 30k rpm. If any one plans to use this stuff for production, they'll need a 'real' spindle built to DIN-6 bearing standards or something similarly expensive. I think with a 3 stage belt system we could knock out a basic 20-40k spindle desing in a weekend of testing.
Leadscrews can be had on eBay, especially in smaller sizes, for pretty cheap. Acme threads seem to prevail. I think we should make the thread and motor steps/TPI adjustable in software, since I doubt I'll be able to secure 3 identically threaded screws for my machine. That will also allow people to use a finer step on, say, the Z axis if they want better control for copper milling.
One thou/1mil +/-1 (or 2) seems like a good goal for everyone. I think with care, even someone with a hacksaw, a file, and a good square (coupled with patience) could make that happen.
Jack-
The design of a gantry is pretty simple, and rather mature in the hobby CNC world. I suggest we define the workpiece size, perhaps 12"x12". It's easy to CAD up something from that. Peter has the right idea matching the motors to the leadscrew thread pitch. Microstepping can make that even more precise.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
Oh, clearly I agree with you that 8020's slide "bearings" are totally inadequate.
That said, this afternoon I whipped up a (conceptual) design to complement their structure with a recirculating ball slide that I think I can make with my CNC shop. So I am getting some samples of the various 8020 profiles to garner some idea of their sturdiness and hence suitability for a 24 inch table.
That then triggered me into thinking about how to make the lead screws, and their recirculating ball nuts. I have some ideas, so I'll have one of our lads CAD both recirculating concepts up (actually we use a solid modeling package called Solid Edge) and then I can actually get parts animation from that. Then of course the critical thing will be to see as to the suitability of our shop to make the recirc slides, the recirc nuts as well as the screws.
The critical thing for me now is to get those profile samples in.
Could be a lot of fun, especially if it worked well.
As I was re-reading a number of the posts to this topic, it struck me that I appear to be hogging the spotlight and totally ignoring the input of many of the other contributors. I apologize for that. I is not my intention to be inconsiderate of others, and I value all the collective experience on this subject.
You have started a very interesting and challenging topic that I am very passionate about, and I would be pleased to offer my experience and knowledge in whatever form you choose to use it.
Sounds good Peter. Being rather 3D cad impaired, I'd love to see some animations. They're hard to do with form C and a pencil, and should help get the point across to folks.
The 8020 guys are in Indiana...the profiles should be to you Monday if you're close, or next Friday if you're not, based on my experience with Canadian shipping to the various locales.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
I've been lurking on this tread...but when I saw David McNab's posting to the Homebrew_PCB yahoo group of his recently completed $US200 PCB mill (http://www.freenet.org.nz/cnc/), I thought I would pass the link on.
Although it may not meet the resolution and repeatability previously discussed, it is interesting to see what can be done.
Bean (Hitt Consulting) said...
Peter,
That is why I suggested that the initial machine be geared towards drilling. It's not that hard to chemically etch a PCB, but drilling all the friggin' holes it what drives me insane. For drilling I think 0.005 is good enough.
Bean.
Drilling is the only thing that stopped me from making my own pcb. etching and getting the pattern on the board are not so hard.
Driling 30 or 40 mini holes stopped me dead in my pcb making tracks. I still have the etch chems, and the boards sitting here. I ended up using wire wrap terminals. Took a while, but works great.
It's not a bad link. David has done a quick-and-dirty version of what I think we're trying to acomplish, and seems to have done it pretty well. His advertised resolution of 1 mil is encouraging to me, especially given the doubt imposed over my personal goals
If he can get to 1 mil with threaded rod and an MDF frame, we can surely make it to 0.0005 with extruded channel and Acme or ball leadscrews, no?
Jack-
I fear I may be guilty of the same inadvertant domination of the conversations. I don't want ot squash anyone here either. Mea culpa if I have.
How's the modelling and extrusion sourcing going? It's been almost a week since the last substantive burst of conversation in the thread, and I'm itching for more
Let's get to it!
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
Well, I for one have been working on this project for the past while now. Following Dave's advice on extruded aluminum (8020.net) I agree that a good precision machine could be made using that material as a substantial building block, so I have ordered some extrusions and decided to design one. In fact, several.
My proposal is as follows:
I will design the mechanical motion portion of a family of quasi-precision machines (.002 to .003 target) in a form so that hobbyists can readily build them. The design and bill of materials will be available for hobbyists to build on a non-commercial basis, and of course everyone is free to modify the design to suit their own needs.
In order to get the desired precision repeatability, here will be the requirement for some precision parts such as recirculating slide bearings and lead screws with recirculating nuts. Such materials are not readily available at low prices, so I have decided to make or source them in bulk, and of course the design will be based on those specific components.
The design will be modular for size and application (not just a drill or router), and the whole bill of materials (mechanical motion only) will be available for purchase from Parallax as individual pieces as well as complete bolt-together kits. This latter approach will facilitate the desired precisions, and is expected to cost no more than the builder sourcing his own pieces. Depending on size and features selected, we anticipate the price for a mechanical motion kit to range from a base cost of $50 plus $50 per each axis for smaller units, to a base cost of $100 plus $100 per axis for the larger or top end. It all depends on the number of axes and options selected.
For the time being, you still will have to source your own motors, electronics and software, but these may become available at a later date; the moving mechanical platforms are the critical items for now. The approach is that the critical portions of mechanics would be available, and the machine could be sized and expanded for hobby uses.
Having Parallax agree to stock these items is of great benefit to all, it permits the bulk fabrication at reasonable cost from a trusted and professional supplier.
We want to do a good job on this design and it will still take a while to complete the CAD drawings (hopefully animated 3D). As soon as these are ready they will be posted on the Parallax web site so folks can see the individual pieces or complete kits they may want to order.
It would be useful for Parallax to receive feedback as to the interest in this concept, and to solicit applications for this general purpose multi axis motion machine.
Beyond the obvious circuit board drill and router, how about...................
Adding a cylinder with powered plunger to squeeze icing onto a cake?
Winding transformer coils?
Dispensing glue?
Expose sensitized PC board with a moving optical fiber?
A small surface-mount pick-and-place machine?
Making a movie by "flying" a camera among a model layout?
The list truly is endless.......
I suggest as interest in this subject grows, to move this thread to a separate section on the Parallax site.
We still need electronics and software to be designed, so have at it!
Peter is correct -·we are·committed·to supporting the product line by being a source, offering partial and complete kits, and ensuring viability and availability of the idea started on this forum. All of this, of course, is only if you want us to do that considering the concept is one of a "community mill".
I think the combination of people on this forum and their skills would make this possible. Specifically, Peter (pjv) can turn·these ideas into a reality. And·his facility·is in Canada, where manufacturing costs are a tad lower than the U.S. Having an engineer do double duty as a manufacturer is a real benefit, too. That could save us all a lot of headaches and ensure the highest quality. This is almost the way it has to be for the core assemblies, else the concept·is less·viable due to added expense of communication between separate design engineers and manufacturers (and minimum purchases, sales people involvment, additional drawings and specifications that have to be created, etc.). Initially this·arrangement would not be viable without trusted parties,·distribution channels, and know-how of the forum users.
This also·opens up opportunities for Parallax to sell control electronics designed and built by our customers.
Parallax would aim to keep the costs·as low as possible with the "community" concept·remaining·in tact.·I told Peter I think we could support this with minimal overhead on top of his assemblies. The intent is to remain open-source, well-supported, and offer initially only the core assembly required (motor mount, drive system, etc.) for this micro-motion machine. In time it would grow.
Like you, we're interested in machine control and making things·like PCBs and aluminum/plastic parts. If our interest were not involved then we would not be involved as a business.
So, don't be quiet and tell us if you think this would work.
Ken, that sounds like a wonderful idea. I for one would greatly welcome a single source for a majority of my supplies, tho I admit to having greater aspirations for the P, A, and R of my mill than the group at large (which is well documented in previous posts).
Peter, if you ask, the nice folks at 8020 will kit things up for a paltry sum, assuming you pay for materials and cutting/machining. That may help both you and Ken in this, not only by having the machining done at the factory, but by being able to stock a smaller numbers of SKUs for the project, helping reduce the inventory management load on Parallax. It migh tnot be as cheap as doing it all locally to you, but that's a
I think the easiest way is to head for a more granular level of modularity, something like this:
Small, medium, and large gantry bases with Y axis (10"x10", 16"x16", 24"x24", DIY worksurface)
Small, medium, and large gantry arms (3, 6, and 9-12" YZ axis max)
Matching X axis for gantry arms (10, 16, 24")
Several spindle options (rotozip/dremel/etc holder, bearing support for HS drills, blank plate for DIY)
Modular motor drivers (single axis bridge, chopper and linear drivers)
Controller "brain" (the Sx board we started talking about on page 1, with or without software?)
"Backplane" housing for modular components (edge cards seem like a bad idea...perhaps ribbon cables or something?)
Builders would be required to come up with their own motors and feedback devices, power supply, spindle, and cutters. A modular controller concept allows for segmented building of a machine as well as open expansion of the controller and a lower repair/replacement threshold when something goes wrong.
I think two options should be explored for the mechanics. Ballscrews have phenomenal accuracy, especially C3 or better ground ballscrews, but are terrifyingly expensive for some. Acme leadscrews and spring loaded anti-backlash nuts provide a level of accuracy that can match the 0.002" goal for a lot less money, not to mention being a lot easier to order (McMaster-Carr and Grainger both stock them in a variety of sizes). I think ballscrews should be offered as an extra-cost option, but may not be completely necessary for the majority just looking to drill holes for their DIPs and leaded resistors.
my $0.02
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
Wow. This project is exceeding all my expectations.
Very exciting.
As for people afraid of hogging the thread. Have at it.
That is why we are here. Nothing will kill a thread faster than
no posting for a few days.
Peter (pjv) said...
My proposal is as follows:
You have really stepped up to the plate Peter.
obviously the mechanical design is very important to this project.
I can't wait to see your draft.
Peter (pjv) said...
Having Parallax agree to stock these items is of great benefit to all
Parallax to boot, your a genius.
Peter (pjv) said...
Expose sensitized PC board with a moving optical fiber?
Ha!, I love it.
Ken Gracey said...
All of this, of course, is only if you want us to do that considering the concept is one of a "community mill".
Absolutely. Definitely, yes.
I feel strongly that if we were to make a basic kit available that
we should see many ingenious applications from the community.
So, looking to the future.
Do we know what kind of motors we will be using for this project
so that Bean and myself can start fleshing out the software.
I'm starting to think that the PC->PIC protocol should be so generic that
it can be incorporated into any "user" motor control.
I received my extrusions yesterday, and they are indeed very robust. I'm busy concocting a recirculating ball slide bearing at this moment; the lead screws are next, and they will take quite a bit of effort. I'm also contemplating a belt driven motive mechanism for those who can tolerate a little less precision but who want fast positioning speed.
As far as choice of drive motors, I expect to be able to accommodate several sizes of "standard flat plate 4 corner mounting screw" style stepper motors. The shaft size I expect is 1/4 inchs, but I still need to confirm that.
In due course I will also offer the matching SX variable voltage (learned this one the hard way) stepper drivers.
For PC interfacing, it's probably a good idea to use industry standard "G codes" for the motion commands, so that those with industry standard CAM (CAD) packages can directly output compatible motion code. I'll need to determine which of the codes would apply.
umm...G-Codes, huh.· I could probably help a bit there, if you want.
My day-job is all about CNC firmware and part programs.· I have off and on considered doing a simple G-Code interpreter in a micro "just for fun" and possibly as a intermediate hobby 'bot controller.
Daniel
PS.· I would also suggest using the machining standards for "drip feeding" part programs, something else I've a spot of experience in.
Once the machine parameters are a bit further along, the applicable G codes to choose should be quite simple for us to determine as we also use them continuously.
But for those who are not in the know, please could you take the time to make a descriptive post to bring others up to speed as to their application?
Please elaborate on "drip feeding"; you mean a move at a time?
May I strongly suggest making provisions for the standard NEMA 17, 23, 34 and maybe 42 frame mounts? Steppers and servos are easy to find in those sizes, and are well suited to the project, as there are a wide range of torques and step angles available across those 3(4?) sizes. Also, instead of a linear variable voltage drive, a chopper drive should be a must for driving steppers. The LMD182xx series H-bridges work pretty well from what I've heard, and could easily be adapted to do what we want. It's like the difference between bitbanged PWM and the hardware PWM. Variable voltage and chopper stwpper drives both work, one's just so much nicer. With the high excitation voltage but limited/chopped current choppers do better at maximizing dynamic torque, which will be important for folks doing milling in addition to drilling.
My $0.02, as always.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
Re: voltage drive, I had expected to use a chopper, NOT linear. Whatever folks like, giving the highest torque vs low dissipation at stand-still. At this point I'm not (yet) bothering with the electronics or software as I have my hands full with the mechanical.
Dave, could you please look up and then post your proposed mounting dimensions for the 4 recommended sizes? Also please include shaft diameters if they apply to those standards.
Drip Feeding, also goes by a number of other names, is basically using some sort of flow control to stream a·large part program into a controller's small memory.· In this context, an SX would likely have a small memory as compared to the probable large number of move segments that a typical PCB part program would have.
This could be as simple as a buffered serial comms link.· The catch is, to a great degree, whether the motion engine brings the axes to a complete stop at the end of one move segment before it starts another.· It the moves are blended, then the buffer must be at least large enough to have several successive commands so that the blends can be accomplished.
Move segments to execute are taken from the buffer, and the buffer is not allowed to go empty.
Blended moves might be overkill for the intended audience of this mill.
pjv said...
But for those who are not in the know, please could you take the time to make a descriptive post to bring others up to speed as to their application?
Are you asking for a brief tutorial / whitepaper on the subset of G-Codes relevant to PCB milling/drilling? or about the larger arena of G-Code usage?· What level of audience should I aim for?
Do any of the Don Lancaster Flutterwhumper concepts (postscript processors)·have applicibility to this project?· what about HP/GL interpretation?· CAD drawing to G-Code? post-processing?
I'd want to limit the description to a narrow enough focus to not wax pedantic (a tendency I do admit).
Comments
The steppers should really be driven with a chopper drive. 75-100oz-in of torque would be good, especially if folks are using cheap threaded rod, which is notoriously inaccuracte. Microstepping is an approach I don't think we can afford to ignore. Yes, direct drive will be slow. DC servos are faster, but good encoders that will allow us the PAR we need will make the project cost prohibitive for some, and I'd like us to avoid that if possible.
Again in agreement with Peter, I think an X-Y table will make the system hard to build and sloppy. A gantry is much simpler to put together, and has much better weight distribution on the x-axis. It also allows the workpiece to sit on a flat, unmoving plate, easing the need for fixturing and simplifying alignment of the 0,0,0 point.
Something like this:
These types of machines lend themselves to easy construction from t-track (the 8020 stuff I mentioned a few posts up) or plate construction. The twin leadscrews for the major axis can be slaved together with chain or a toothed belt and driven from a central motor. I think that once the parts are assembled, a machine like that can probably be put together in a matter of hours by someone with a modest amount of mechanical skill.
I've built X-Y tables before, 3 of them. A gantry is simpler to build, more stable, and easier to maintain and clean.
As always, my highly opinionated $0.02
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
I like the overhead approach.
Also just thinking....lasers a very popular in a lot of tools these days.· (really more of a novelty marketing ploy).
But how about using a photocell and a laser to aid in alignment of the table -- basically, if you had the gantry/table centered, you could shine the laser thru some portion of the table and if it's in alignment, you'd have the photocell, opposite the hole, picking up the laser light.
Or perhaps using to lasers to act as a visual crosshair (so you'd know where the drill is supposed to go)....all these require aligning themselves of course and are definately icing on the cake once things are running!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·
Steve
http://ca.geocities.com/steve.brady@rogers.com/index.html
"Inside each and every one of us is our one, true authentic swing. Something we was born with. Something that's ours and ours alone. Something that can't be learned... something that's got to be remembered."
Post Edited (steve_b) : 6/21/2005 11:22:56 PM GMT
And I misspoke, it's easier to build a gantry with a single leadscrew down the center with two carriages on eeither side than placing screws at the carriage locations.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
A simple laser through a hole into a photocell does not work very well. If you put a 4 way optical splitter in the path after the hole with 4 as identical as possible pick-ups, and then move the table so as to balance the detected light among all 4, will work for lower degrees of accuracy.
I mounted a pin hole camera on one of my units, and that was adequate for aligning to pad centers. I added a video cross hair generated by an SX; that works great.
As for limit switches, by themselves they are woefully inadequate for positioning the reference. They will have up to 10 or even 20 thou of slop. You need to add a rotary index to the lead screw as do the commercial units. They have at a minimum one "reference" pulse per rotation. The limit switch is then used (is adequate for) to determine the reference turn on the screw, and the encoder's reference pulse gives you the count start position.
In using stepper motors one uses the limit switches to get "close" (within one turn) of the reference position, and then a rotary mechanical stop (not a jamnut) to bottom out the stepper. This works very reliably to the precision of a single step.
Not wanting to pour cold water on an interesting challenge, but all this is not exactly trivial to implement when one expects the precision Dave is hoping for. In fact Dave, I think 5 tenths of a thou is well beyond most hobbyists' limit unless they have a pretty competent machine shop at their disposal.
Cheers,
Peter (pjv)
Post Edited (pjv) : 6/22/2005 4:29:20 AM GMT
That is why I suggested that the initial machine be geared towards drilling. It's not that hard to chemically etch a PCB, but drilling all the friggin' holes it what drives me insane. For drilling I think 0.005 is good enough.
Bean.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
"SX-Video·Module" Now available from Parallax for only $28.95
http://www.parallax.com/detail.asp?product_id=30012
Product web site: www.sxvm.com
"What's the difference between ignorance and apathy ?"
"I don't know, and I don't care."
·
I know, 0.0005 is pretty darn small, but having played with this a few times (never quite making it to the full blown CNC stage) I think it is within the reach of determined amateurs building small scale machines. Using commercial extrusions and a machinist's square, I put together a 14" square base that was flat to no less than 0.001" (corner to corner, laterally it was closer to 0.0007")over a weekend a few years ago. It wasn't much more difficult than spending half an hour verifying my power miter saw's alignment and making a few test cuts. I don't thing that doing half a mil better is insurmountable, even for a garage hobbyist like me. It just takes persistance.
I'm willing to take a chance in trying it. I think the base is the easier of the two parts I need ot hit my 5 tenth goal..the tougher one will be the PAR related to the positioners and the actual tool spindle. Achieving one tenth (0.0001") of TIR will be a real task I think. I have a few ideas I'm looking into, but I can't say for sure if any will be even remotely correct.
As an aside, this is precisely why I tried to distance my personal goals for the project from the community goals. One mil is still something I consider to be a minimum number for PAR for the average implementation, desapite my own lofty goals.
10 mils (100 tenths, ten thou, 0.010", .254mm, etc) is an unacceptable error when milling closely spaced pads for something like an SX52 (11.8mil wide pads on 21.6 mil centers, worst case). Even if the boards are pre-etched and we're just drillingholes, if we have folks using 25 mil vias, that's still a 40% error. I may be the only one here doing 7mil lines/spaces and 15 mil vias, but being 10 mils off for someone doing 25 mil lines/spaces can be nearly as bad as it would be for me. a 25-50% hit in line width can be very bad.
Obviously, the minimum spacing of board features will be governed by the PAR of the machine, but I don't think 1 mil is an outrageous goal for the average hobbyist, is it?
I think this is one of the specifications that will be most difficult to nail down, since it will require an actual impelmentation to measure and verify.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
No, I think 1 thou is well beyond all but the most persistent and finniky hobbyist. Most would not have the measurement techniques nor facilities to even determine how square, level or long their pieces are. You can only do so much with a vernier caliper, even a digital one. Tools for measuring distances over 6 inches get expensive very rapidly.
I do share your views though in needing accuracies approaching your numbers for a serious product, and anything much worse than your goals would be an interesting exercise, but of marginal real value. Of course every one will have their own expectations, and hence different views on precision needs will prevail.
For myself, I would be disappointed with precisions of 5 thou, satisfied with 2 or 3 thou, and elated with 1 thou. I can't imagine getting to 5 tenths.
As far as the spindle is concerned, I am less worried about that. I would pursue an avenue of using NOT a brushed motor; I can't stand the racket. As very little cutting power is required, I would look to invent a spindle driven by something brushless and ratioed up with several pulleys and belts (O rings?). It would be nice to be able to get variable speed to 50K RPM, but bearings might become a serious issue.
I believe dentist drills are air driven, and come in a throw-away module including the bearings and turbine, and run up to 200K, all for not much money. This could be an interesting avenue to investigate for the high speeds/low loads required for mechanical PCB dry "etching".
Cheers,
Peter (pjv)
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·
Steve
http://ca.geocities.com/steve.brady@rogers.com/index.html
"Inside each and every one of us is our one, true authentic swing. Something we was born with. Something that's ours and ours alone. Something that can't be learned... something that's got to be remembered."
I think 5 thou/mil should be easy, especially if we can find a source of positioning actuators that aren't hardware store threaded rod. I've done enough with that stuff to hate it with a passion. It's not round or straight, and the threads are rarely even close to smooth.
I'd much rather spend time trying to dig up a source of 3/8 or 1/2" Acme thread rods and anti-backlash nuts. Same goes for positioning slides...you can get enough THK and OKI slides to do a small gantry mill (12"x12"x2" or so) on eBay for under $200, sometimes less if you're lucky, and I have yet to see one that is more than 0.001" over it's length, they're usually less.
I'm with Steve on the dental drill idea. There are much better ways of getting a high speed spindle going, and probably with more precision. ABEC-7 skateboard bearings (available for a few dollars each online) have a TIR of no more than 2.5 uM (micro meters, 0.0025mm, 0.0000918"), more than sufficient to support a very high speed spindle. A 2 stage belt drive (Kevlar, rubber, o-rings, whatever) with a normal DC motor motor (8k-20k, 20k-40krpm) could do the RPM conversion pretty cheaply, and I think with a 1/4" spindle the moving mass woud be low enough that things would spin up and down pretty fast. A few pieces of extruded angle 4-6" long could support the whole shebang with good PAR as well. Hmm...
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
So..., where do I start.
You have hit upon one of the great challenges of this project.
Scope is the key word here.
Your average "Joe" is not going to build an industrial milling machine with a .0005 tolerance.
It would be unkind to lead them to believe that they could. There is a good reason that a "cheap"
CNC machine cost $10.000. That money is for man hours / development not material.
The average hobbyist PCB board has a space between traces in the order of 1/32 to an 1/8 inch.
The overall goal here is to design something that everyone can build.
And that means Simple. Which in turn means not incredibly accurate.
The design goal is to create a solid design (software and hardware) that can be
build upon for those with the wherewithal to take it to the next level.
Our immediate goal should be to create an elegant design that can drill lots of
holes with a "design eye" towards a respectable milling machine.
I'll be sure to take you up on that offer. [noparse]:)[/noparse]
That is definitely my intention.
Now that is why it is a community project. I did not think of that.
That is more into the phase three of the project thought.
Those are some humbling facts. Thanks for that.
Especially the notes about stepper voltage.
Are you saying travel the head across the table? A oppose to the table traveling under a stationary head.
That would seem to be another recommendation for a stationary table.
OK, OK, I get it.
I'm with you on that one.
I would hate to design something that only the elite hobbyist could build.
As for etching, I still can not get that resist pen idea out of my head.
Lets see, what did we learn today.
A moving table is probably out of the question.
Our accuracy needs to be in the single thousands of an inch. Much less than 1 thousands seems out of reach.
We need a simple solution to positioning. I'm starting to wonder if we can solely rely on the stepper motor.
We need to start designing the mechanical layout of the machine.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jack
Thanks for the pointer to the 8020 site!!!!
Lots of good stuff there that will help to put some repeatable precision into the design, and not too expensive either.
With that product as a reference point, I think quite a number of the mechanically inclined hobbyists will be able to come close to the (my) 2 or 3 thou goal, and some of them with better tools will get to 1 thou. (I still have trouble believing it can get better than that, but never mind)
I agree with you on the spindle other than bearings at over 20K. I have a commercial 4 headed PCB drill with tiny direct drive 3 phase motors running at up to 60K, and you should see the trouble I have in getting replacement bearings that will last. Roller blade parts won't cut it for long. I'm contemplating a conversion to air bearings.
Since I have some spare spindle motors, I will probably adapt one of them into my system.
So now the hard part is to secure an inexpensive source for lead screws and anti-slop nuts. I wonder if someone knows of a good low cost source for recirculating ball nuts and matching lead screws. For me the diameter should be minimum of 3/8 or 1/2. The thread pitch should be 0.2 inches as that will nicely match a 200 step motor giving one thou per full step. Anyone requiring better than that can microstep.
Thanks to everyone for their views and input.
Cheers,
Peter (pjv)
The 8020 stuff is great for structure, but their linear motion components have to much stiction for out uses. (UHMW + anodized Aluminum = not smooth, even with teflon spray). The extrusions and brackets themselves are really handy for this kind of thing. 8020 guarantees them to have straightness better than 0.0125" per foot and 0.120 per 20' section, but I've found them to be much much better than that.
Air bearings would be nice, but their manufacture for a hobbist is probably out of the question. ABEC-7 or -9 skate bearings have amazing runout, and if kept clean and oiled regularly, they should last plenty long at 30k rpm. If any one plans to use this stuff for production, they'll need a 'real' spindle built to DIN-6 bearing standards or something similarly expensive. I think with a 3 stage belt system we could knock out a basic 20-40k spindle desing in a weekend of testing.
Leadscrews can be had on eBay, especially in smaller sizes, for pretty cheap. Acme threads seem to prevail. I think we should make the thread and motor steps/TPI adjustable in software, since I doubt I'll be able to secure 3 identically threaded screws for my machine. That will also allow people to use a finer step on, say, the Z axis if they want better control for copper milling.
One thou/1mil +/-1 (or 2) seems like a good goal for everyone. I think with care, even someone with a hacksaw, a file, and a good square (coupled with patience) could make that happen.
Jack-
The design of a gantry is pretty simple, and rather mature in the hobby CNC world. I suggest we define the workpiece size, perhaps 12"x12". It's easy to CAD up something from that. Peter has the right idea matching the motors to the leadscrew thread pitch. Microstepping can make that even more precise.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
Oh, clearly I agree with you that 8020's slide "bearings" are totally inadequate.
That said, this afternoon I whipped up a (conceptual) design to complement their structure with a recirculating ball slide that I think I can make with my CNC shop. So I am getting some samples of the various 8020 profiles to garner some idea of their sturdiness and hence suitability for a 24 inch table.
That then triggered me into thinking about how to make the lead screws, and their recirculating ball nuts. I have some ideas, so I'll have one of our lads CAD both recirculating concepts up (actually we use a solid modeling package called Solid Edge) and then I can actually get parts animation from that. Then of course the critical thing will be to see as to the suitability of our shop to make the recirc slides, the recirc nuts as well as the screws.
The critical thing for me now is to get those profile samples in.
Could be a lot of fun, especially if it worked well.
Cheers,
Peter (pjv)
As I was re-reading a number of the posts to this topic, it struck me that I appear to be hogging the spotlight and totally ignoring the input of many of the other contributors. I apologize for that. I is not my intention to be inconsiderate of others, and I value all the collective experience on this subject.
You have started a very interesting and challenging topic that I am very passionate about, and I would be pleased to offer my experience and knowledge in whatever form you choose to use it.
Sincerely,
Peter (pjv)
The 8020 guys are in Indiana...the profiles should be to you Monday if you're close, or next Friday if you're not, based on my experience with Canadian shipping to the various locales.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
Although it may not meet the resolution and repeatability previously discussed, it is interesting to see what can be done.
Daniel
Drilling is the only thing that stopped me from making my own pcb. etching and getting the pattern on the board are not so hard.
Driling 30 or 40 mini holes stopped me dead in my pcb making tracks. I still have the etch chems, and the boards sitting here. I ended up using wire wrap terminals. Took a while, but works great.
It's not a bad link. David has done a quick-and-dirty version of what I think we're trying to acomplish, and seems to have done it pretty well. His advertised resolution of 1 mil is encouraging to me, especially given the doubt imposed over my personal goals
If he can get to 1 mil with threaded rod and an MDF frame, we can surely make it to 0.0005 with extruded channel and Acme or ball leadscrews, no?
Jack-
I fear I may be guilty of the same inadvertant domination of the conversations. I don't want ot squash anyone here either. Mea culpa if I have.
How's the modelling and extrusion sourcing going? It's been almost a week since the last substantive burst of conversation in the thread, and I'm itching for more
Let's get to it!
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
Steve
Well, I for one have been working on this project for the past while now. Following Dave's advice on extruded aluminum (8020.net) I agree that a good precision machine could be made using that material as a substantial building block, so I have ordered some extrusions and decided to design one. In fact, several.
My proposal is as follows:
I will design the mechanical motion portion of a family of quasi-precision machines (.002 to .003 target) in a form so that hobbyists can readily build them. The design and bill of materials will be available for hobbyists to build on a non-commercial basis, and of course everyone is free to modify the design to suit their own needs.
In order to get the desired precision repeatability, here will be the requirement for some precision parts such as recirculating slide bearings and lead screws with recirculating nuts. Such materials are not readily available at low prices, so I have decided to make or source them in bulk, and of course the design will be based on those specific components.
The design will be modular for size and application (not just a drill or router), and the whole bill of materials (mechanical motion only) will be available for purchase from Parallax as individual pieces as well as complete bolt-together kits. This latter approach will facilitate the desired precisions, and is expected to cost no more than the builder sourcing his own pieces. Depending on size and features selected, we anticipate the price for a mechanical motion kit to range from a base cost of $50 plus $50 per each axis for smaller units, to a base cost of $100 plus $100 per axis for the larger or top end. It all depends on the number of axes and options selected.
For the time being, you still will have to source your own motors, electronics and software, but these may become available at a later date; the moving mechanical platforms are the critical items for now. The approach is that the critical portions of mechanics would be available, and the machine could be sized and expanded for hobby uses.
Having Parallax agree to stock these items is of great benefit to all, it permits the bulk fabrication at reasonable cost from a trusted and professional supplier.
We want to do a good job on this design and it will still take a while to complete the CAD drawings (hopefully animated 3D). As soon as these are ready they will be posted on the Parallax web site so folks can see the individual pieces or complete kits they may want to order.
It would be useful for Parallax to receive feedback as to the interest in this concept, and to solicit applications for this general purpose multi axis motion machine.
Beyond the obvious circuit board drill and router, how about...................
Adding a cylinder with powered plunger to squeeze icing onto a cake?
Winding transformer coils?
Dispensing glue?
Expose sensitized PC board with a moving optical fiber?
A small surface-mount pick-and-place machine?
Making a movie by "flying" a camera among a model layout?
The list truly is endless.......
I suggest as interest in this subject grows, to move this thread to a separate section on the Parallax site.
We still need electronics and software to be designed, so have at it!
Sincerely,
Peter (pjv)
Peter is correct -·we are·committed·to supporting the product line by being a source, offering partial and complete kits, and ensuring viability and availability of the idea started on this forum. All of this, of course, is only if you want us to do that considering the concept is one of a "community mill".
I think the combination of people on this forum and their skills would make this possible. Specifically, Peter (pjv) can turn·these ideas into a reality. And·his facility·is in Canada, where manufacturing costs are a tad lower than the U.S. Having an engineer do double duty as a manufacturer is a real benefit, too. That could save us all a lot of headaches and ensure the highest quality. This is almost the way it has to be for the core assemblies, else the concept·is less·viable due to added expense of communication between separate design engineers and manufacturers (and minimum purchases, sales people involvment, additional drawings and specifications that have to be created, etc.). Initially this·arrangement would not be viable without trusted parties,·distribution channels, and know-how of the forum users.
This also·opens up opportunities for Parallax to sell control electronics designed and built by our customers.
Parallax would aim to keep the costs·as low as possible with the "community" concept·remaining·in tact.·I told Peter I think we could support this with minimal overhead on top of his assemblies. The intent is to remain open-source, well-supported, and offer initially only the core assembly required (motor mount, drive system, etc.) for this micro-motion machine. In time it would grow.
Like you, we're interested in machine control and making things·like PCBs and aluminum/plastic parts. If our interest were not involved then we would not be involved as a business.
So, don't be quiet and tell us if you think this would work.
Sincerely,
Ken Gracey
Parallax, Inc.
P.S. We recognize Jack for launching the concept.
Peter, if you ask, the nice folks at 8020 will kit things up for a paltry sum, assuming you pay for materials and cutting/machining. That may help both you and Ken in this, not only by having the machining done at the factory, but by being able to stock a smaller numbers of SKUs for the project, helping reduce the inventory management load on Parallax. It migh tnot be as cheap as doing it all locally to you, but that's a
I think the easiest way is to head for a more granular level of modularity, something like this:
- Small, medium, and large gantry bases with Y axis (10"x10", 16"x16", 24"x24", DIY worksurface)
- Small, medium, and large gantry arms (3, 6, and 9-12" YZ axis max)
- Matching X axis for gantry arms (10, 16, 24")
- Several spindle options (rotozip/dremel/etc holder, bearing support for HS drills, blank plate for DIY)
- Modular motor drivers (single axis bridge, chopper and linear drivers)
- Modular feedback interfaces (optical encoders, limit switches, etc)
- Controller "brain" (the Sx board we started talking about on page 1, with or without software?)
- "Backplane" housing for modular components (edge cards seem like a bad idea...perhaps ribbon cables or something?)
Builders would be required to come up with their own motors and feedback devices, power supply, spindle, and cutters. A modular controller concept allows for segmented building of a machine as well as open expansion of the controller and a lower repair/replacement threshold when something goes wrong.I think two options should be explored for the mechanics. Ballscrews have phenomenal accuracy, especially C3 or better ground ballscrews, but are terrifyingly expensive for some. Acme leadscrews and spring loaded anti-backlash nuts provide a level of accuracy that can match the 0.002" goal for a lot less money, not to mention being a lot easier to order (McMaster-Carr and Grainger both stock them in a variety of sizes). I think ballscrews should be offered as an extra-cost option, but may not be completely necessary for the majority just looking to drill holes for their DIPs and leaded resistors.
my $0.02
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
Wow. This project is exceeding all my expectations.
Very exciting.
As for people afraid of hogging the thread. Have at it.
That is why we are here. Nothing will kill a thread faster than
no posting for a few days.
You have really stepped up to the plate Peter.
obviously the mechanical design is very important to this project.
I can't wait to see your draft.
Parallax to boot, your a genius.
Ha!, I love it.
Absolutely. Definitely, yes.
I feel strongly that if we were to make a basic kit available that
we should see many ingenious applications from the community.
So, looking to the future.
Do we know what kind of motors we will be using for this project
so that Bean and myself can start fleshing out the software.
I'm starting to think that the PC->PIC protocol should be so generic that
it can be incorporated into any "user" motor control.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jack
Glad you are back and continue to be involved.
I received my extrusions yesterday, and they are indeed very robust. I'm busy concocting a recirculating ball slide bearing at this moment; the lead screws are next, and they will take quite a bit of effort. I'm also contemplating a belt driven motive mechanism for those who can tolerate a little less precision but who want fast positioning speed.
As far as choice of drive motors, I expect to be able to accommodate several sizes of "standard flat plate 4 corner mounting screw" style stepper motors. The shaft size I expect is 1/4 inchs, but I still need to confirm that.
In due course I will also offer the matching SX variable voltage (learned this one the hard way) stepper drivers.
For PC interfacing, it's probably a good idea to use industry standard "G codes" for the motion commands, so that those with industry standard CAM (CAD) packages can directly output compatible motion code. I'll need to determine which of the codes would apply.
Lots to do, chat later,
Peter (pjv)
umm...G-Codes, huh.· I could probably help a bit there, if you want.
My day-job is all about CNC firmware and part programs.· I have off and on considered doing a simple G-Code interpreter in a micro "just for fun" and possibly as a intermediate hobby 'bot controller.
Daniel
PS.· I would also suggest using the machining standards for "drip feeding" part programs, something else I've a spot of experience in.
ddm
Post Edited (daniel) : 6/29/2005 9:56:12 PM GMT
Once the machine parameters are a bit further along, the applicable G codes to choose should be quite simple for us to determine as we also use them continuously.
But for those who are not in the know, please could you take the time to make a descriptive post to bring others up to speed as to their application?
Please elaborate on "drip feeding"; you mean a move at a time?
Cheers,
Peter (pjv)
May I strongly suggest making provisions for the standard NEMA 17, 23, 34 and maybe 42 frame mounts? Steppers and servos are easy to find in those sizes, and are well suited to the project, as there are a wide range of torques and step angles available across those 3(4?) sizes. Also, instead of a linear variable voltage drive, a chopper drive should be a must for driving steppers. The LMD182xx series H-bridges work pretty well from what I've heard, and could easily be adapted to do what we want. It's like the difference between bitbanged PWM and the hardware PWM. Variable voltage and chopper stwpper drives both work, one's just so much nicer. With the high excitation voltage but limited/chopped current choppers do better at maximizing dynamic torque, which will be important for folks doing milling in addition to drilling.
My $0.02, as always.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
Re: voltage drive, I had expected to use a chopper, NOT linear. Whatever folks like, giving the highest torque vs low dissipation at stand-still. At this point I'm not (yet) bothering with the electronics or software as I have my hands full with the mechanical.
Dave, could you please look up and then post your proposed mounting dimensions for the 4 recommended sizes? Also please include shaft diameters if they apply to those standards.
Cheers,
Peter (pjv)
This could be as simple as a buffered serial comms link.· The catch is, to a great degree, whether the motion engine brings the axes to a complete stop at the end of one move segment before it starts another.· It the moves are blended, then the buffer must be at least large enough to have several successive commands so that the blends can be accomplished.
Move segments to execute are taken from the buffer, and the buffer is not allowed to go empty.
Blended moves might be overkill for the intended audience of this mill.
Are you asking for a brief tutorial / whitepaper on the subset of G-Codes relevant to PCB milling/drilling? or about the larger arena of G-Code usage?· What level of audience should I aim for?
Do any of the Don Lancaster Flutterwhumper concepts (postscript processors)·have applicibility to this project?· what about HP/GL interpretation?· CAD drawing to G-Code? post-processing?
I'd want to limit the description to a narrow enough focus to not wax pedantic (a tendency I do admit).
Daniel
Post Edited (daniel) : 6/30/2005 2:37:25 AM GMT
My take on that would be for the basics. To those already familiar with G codes it would be less interesting; to others a bit of new understanding.
A short explanation of CAD vs CAM would probably also be appreciated.
Thanks,
Peter (pjv)