I think you may have overlooked your best alternative for a cutting head motor.
Routers are extremely high speed and built to withstand lateral pressure.· They are also available in a wide variety of sizes and configuratons.· Some of the cheap ones don't last but a few hours, but a good professional one will run for years.· Excellent bearings, good cooling, and well balanced 20,000RPM motor.
Generally, they tend to use collets to securely hold the cutting tool as 'drill chucks' tend to slip under pressure.
Nuts and Volts recently had an all plastic X-Y-Z table that is appropriate and easily built with a bandsaw or tablesaw.· The only special machining was to create a thread cutter tool and to fit the ends of screws to stepper motor shafts.
I believe you can do all that for under $500USD.· And you have a table that can switch to other devices as you wish.· At times you may want to have the table graphically draw where it is going.· This is a great error detection method as it does not involve actual material.· So you would change from router to a marker or pencil as required.
For slow speeds, you can switch to an electric drill with a light dimmer as a speed controller.
I think building the jig/rig for under $500 is easier accomplished if you don't include the price of the drill/dremel.
I think most ppl will lean more towards a metal device as opposed to plastic (but this would be discussed).
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ ·
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."
John C Kleinbauer's Brute was designed specifically as a circuitboard milling machine. (crankorgan.com).
Lot's of the comments above sound like they are reinventing the wheel. Study John's ideas and experiences. Developing an effective CNC system involves lots of decisions, many which come with experience as to what works and what doesn't. John has documented his experiences in his rather rough tone. Its worth reading.
He also has a bulletin board where CNC users exchange ideas.
One thing I haven't thought about yet is runout. How much is there on the shaft of a Rotozip? Dremel?
Small PCB drills (0.015) want to be spun aroudn 20,000RPM. 0.001 (1 mil) of runout can destroy them very quickly, inside of a few revolutions. Industrial PCB drilling machines operate at 20-30k rpm with runouts in the neighborhood of 0.0001" (one tenth for the machinist types).
I'm not sure what the plan is for drill bits or engraving bits, but with the number of each available in the surplus market, I'd think we'd want to focus on them.
My $0.0125 (depreciated)
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
Dave Andreae said...
Tech support would be happy to answer any questions that you may have provided that is with in our knowledge base
Thanks Dave. Parallax being supportive, now there is a surprise. Note Emoticon. I would hate for that to be taken in the wrong way.
Bean said...
Maybe we should start with a machine to just drill holes first.
Another great idea. Software wise that is a perfect stepping stone to the trace milling.
If we can't get it to drill holes there is no point in working on the milling side.
Plus the machine would be of great use while the milling aspects are being developed.
That should be part of the development plan.
Forrest said...
Here's another article on a low-cost, home-built NC Drill machine for drilling PC Boards www.circuitcellar.com/fi2003/F2019.htm
What a wealth of information on that page. That is the machine I saw in my "minds eye" when I made my first post.
Great link.
bobledoux said...
John C Kleinbauer's Brute was designed specifically as a circuit board milling machine. (crankorgan.com).
I understand what you are saying about reinventing the wheel, but I am not comfortable buying plans site unseen.
So, looks like things we be coming together on the 18th.
We should spend a couple/few weeks just talking about our plan of attack.
This project could easily spin into the abyss if we have a weak plan.
Also the serial protocol should be designed so that you can use any of the Parallax ICs. (stamp, SX, etc.)
Can't wait to get all theses ideas in one place.
So far it's pretty exciting, lets keep it that way.
I have the Brute plans sold by Crankorgan.com. The plans are good and include details of a low cost controller. He encourages users to run their machines under DOS using a parallel port and low cost or free software.
His Brute has ways constructed from commercial building window channel that can often be picked up for free. The biggest cost is locating suitable stepper motors.
Its possible to build a brute for $200 with some scrounging.
He is a big supporter of Dremel power tools and argues that most other small routers have too much bearing play for precision work. Also note that a PC board cutter is a small tool. This permits small, light and inexpensive construction.
There are lots of ways to go wrong when armchair designing a CNC machine. I suggest contributers read available information before suggesting ideas that are "less than optimal."
All suggestions have Merit...it's up to the moderators/leads to make the decisions.
Do you want a "scavenged" machine that not everyone will be able to find pieces for by scavenging? (as in some of our international brotheren)
Or make something that will work with common enough materials that it would work for beginners (relatively) anywhere?
Dremels are pretty common but a pricey item....some parts you'll have to bite the bullet in order to get enough quality that you get the production you want out of it!
Having something with an expandable work area is a must as not everyone uses the same size board.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ ·
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'm going to have to take at least some issue with the Dremel comment WRT runout. The one I have on my desk is brand new (opened the box yesterday) and has 0.007" of TIR. For a 15 mil drill, that's nearly 50% of it's diameter, and sure to destroy it. I think we're going to need a MUCH more precise spindle mechanism to do PCB drilling (ABEC-7 bearings come to mind). I have no issue with using a dremel or other thing as the power for such a spindle, but the more I get into it, the less I want to chuck my bits into the end of a rotary tool.
Of course, I don't know if anyone else shares my needs to drill 15 mil holes from time to time, so this may be more moot than I think, but whereever possible, I think we should aim for as much precision/accuracy/repeatability as we can afford. It will pay off in the end.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
I question whether a single design, or even a series of designs will meet readers' objectives.
At one extreme may be the builder who expects to build SMT designs using 8 mil traces. Such a machine will cost big bucks.
At the other extreme is the builder seeking simple through hole construction on 0.1 inch standards using 60 mil traces. Such a machine exists in The BRUTE.
The first machine may require expensive ball screws and ways. The second machine can be constructed with industrial threaded rod and plastic bearings and screws.
Its a matter of tradeoffs. How do you get a variety of reader expectations into a design? I read the various CNC bulletin boards and find the whole process is a bee swarm.
In terms of the Dremel: Its small and reasonabily precise, for the money. Most other tools are much larger which requires greater structural integrity and mass in the Z axis. This translates into bigger machines. Bigger machines may be difficult to fit on a small work bench. Most routers can't match the runout of the Dremel just because of the larger scale. A "precise spindle mechanism" is another way of saying an "expensive spindle mechanism."
What kind of budget are you willing to spend to get a machine? At what point does ExpressPCB or iron on etching sheets swing the fiscal argument the other way?
I'm not here to be a naysayer. There are good reasons there are no "cheap" CNC machines. Should Parallax be budgeting expensive staff time on a complex project without first evaluating the potential for success? Maybe, with the construction techniques they are using on their robots, they can extend into a basic CNC process. I don't know. If successful, it could be a killer application.
Post Edited (bobledoux) : 6/14/2005 10:56:03 PM GMT
I agree. I don't think a Dremel will fit everyone's use, not will my idea of a 20k RPM stable spindle with a one-tenth (0.0001) runout. I also don't think everyone will need a 12" Z-axis like I'm planning.
I have a hunch the project will need ot be broken up into modular blocks like this:
Control software (specific to the drive hardware?)
Stepper/servo drive hardware (USB would be nice here...FTDI chips perhaps?)
X-Y mechanism (cheap leadscrews? Acme threads? Linear bearings?)
Z-axis mechanism (travel? actuator??
Cutting system (dremel, rotozip, spindle?)
Discuss
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
I've been talking to RotoZip about the specs on the RZ01.· To quote "We do not really recommend the RotoZip for a CNC machine".· The conversation is on-going and they will be getting back to me.
I think we should define the performance characteristics of what we are trying to build.· Personally, I would be mesmerized by a machine that could make lands .012 to .015 wide, make square or rectanglular pads .065 x .065 or .050 x .100 and can drill a .030 or .035 hole in the center of the pads.· We aren't making an LPKF - we want to make something the average person can afford and use for prototyping, so let's not set our sights too high.· I could live with the Dremel or the RotoZip, both of which I have and use extensively.· After I've prototyped, I can go to ExpressPCB for more sophisticated boards.·
I think we should define the performance characteristics of what we are trying to build.
Sid
That should begin shortly.
My hope is that each milling machine built will be different from the others.
As long as the "IC" and the PC software can communicate (given user adjustable parameters)
there should be lots of room for modifications.
I think just having the interface and design software done would be the biggest part. Because everbody can build their own x,y,z table to their own degree of precision and their own style. The software wont care if your doing a small circurt board or a large block of steel. the precision comes from the table, ball screws and back lash nuts.
So i think the ideal way to proceed with this project would be to divide up the software, and make it the best as possible and with all the feartures we would want. Although we would at least decide as a group on the stepper/ servo motor interfaces. And it would be intresting to see what everyone comes up with for the spindle motor and table.
Parallax's access to quality CNC means they might be able to develop"ways" designs that are beyond the scope of the hobbyist. Making precise X, Y, and Z axes, at a reasonable cost, is key to a project. (Hobby builders often employ extruded aluminum for ways and plastics for bearings, or round bar/tube ways with "skateboard bearings" for the carriers. )
A small machine for 4" by 6" printed circuitboards could be light and inexpensive, employing something like a Dremel tool. But if you want to mill aluminum parts, the machine construction is raised to a whole different level. A Dremel tool won't "cut it."
The Crankorgan Brute plans are beginning to be dated, as they depend on 74xx series chips for the servomotor controllers and use DOS machines for control. But the plans were specifically directed to circuitboard interests.
One issue with Windows machines concerns lost steps. The CNC software must make allowance for Windows interagation of open applications. Occasionally, a step can be missed, putting the axes out of sync with the program. Single applications running under DOS avoid some of these problems.
Measured my X / Y axis tonight and have .012 and .028 slop,·will need to get this taken care of. What do you have in mind for stepper control? After looking around on Google there are a few companies selling controls. I assume that the SX will toggle the driver, control forward and reverse and control travel limits correct?
I have been patiently waiting for the 18th to roll around.
KenM said...
If the above can be resolved by Saturday June 18, PL and I will get started to organize the project.
I'm not quite sure what that means but I have not heard from anyone yet.
My original idea was for the community to get involved with this project. I never expected Parallax to be involved.
I sure they are much too busy developing ingenious products.
I am still willing to moderate / manage incoming suggestions.
Design and write the PC interface for the mill.
And whatever else I can help with.
But I can not do it alone.
We still need an ASM / SX/B programmer.
Someone with sound mechanical design skills.
And lots of people to provide input and ideas.
If this idea dies it will be from lack of support from the community.
I think there were only 12 people that voted in the poll.
There is no reason we can't do this project in this thread as long as I am able to modify the first post with updates.
KenM said...
I posted a thread over a week ago suggesting two people oversee the project, and included a "survey." There were zero (0) replies to the survey
I think they have been posting support in this thread.
KenM said...
So at this point I am not sure if the project is worth pursuing as there was no response with regards to who will oversee the project.
I guess I am confused about that.
I thought we were measuring community interests? I am ready to go.
Just waiting for a community member or two to volunteer their time.
So, I ask again, what is it going to be people.
There is always another software project on my list.
Should I cross out this one?
I have a list of starter suggestions for this project.
Everyone that wants me to start posting say "I".
Another Aye Jack. We need a fearless leader to whip the troops into shape here
As I mentioned before, I have no problem doing layout for the board (Eagle or Protel), and initial prototyping if I can source the parts cheap enough. I came up as a EE, but I spent many of my college off hours working at the campus machine shop, so I'm not afraid to get my hands dirty at a mill.
I'm with Bean on serial comms. USB might seem like a good idea, but the way Windows handles timing for virtual com ports I'm afraid it might miss a step or overshoot something unless that's all taken care of in the controller.
The guys over at 8020.net have a eBay garage sale on blemished extrusions for some really cheap prices. The dimensional tolerance is always the same as normal production parts, they're just scratched or marked some way. I plan on using some for my machine. Strong, light, and easy to work with. (search for 8020 aluminum)
Are encoders for servo style feedback an option (mmm...servos....), or do we want to go with the microstep-and-hope positioning method? Both work, we just need ot define our accuracy, precision, and repeatability requirements.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
OK, lets roll......
Seems we have the beginnings of a team.
Jack - PC Software, protocol development.
Bean - IC code, protocol development.
Dave - Prototyping, circuit layout.
The Software.
Communication with the controller should be via Serial to reduce cost.
More on serial protocol below.
Software should run on all flavors of Windows from 95 up.
Graphic display of current position.
Ability to perform dry runs of a mill job.
Should be able to read at least one standard CNC file format.
User selectable serial communication speeds.
Ability to calibrate the stepper motors.
This list is endless, we will just take it as it comes.
The Controller.
Should be designed as one unit, i.e. a black box.
Designed as simple (low cost) as possible with the ability for
others to build upon the design.
The IC.
Unless speed issues are encountered it would be nice if the
controller could be build with any of the Parallax micro controllers.
With that in mind the IC code should be written in SX/B.
Assembly language is out of reach of most hobbyists.
The project should be developed with an SX and leave the porting
of the SX/B code to a Basic Stamp user.
Keep the above statement in mind when developing the SX/B code.
The Protocol.
The sky is the limit here. This is really an important aspect of the
project and needs to be worked hard before any coding begins.
Any change to the protocol propagates throughout the software and SX/B code.
We should make an attempt to avoid any serial hardware to reduce cost.
The software and controller should be in constant communication.
The controller should acknowledge a received command. Then acknowledge its completion
to request the next command. This should avoid the need to implement a hardware flow control.
Commands should be tokenized to reduce serial bandwidth.
The Mill.
An XY table with a Z head seems to be the preferred design.
A desktop version would be nice. But I see no reason the machine could not be
scaled up if needed.
The mechanical design should only implement the movement of the XY and Z planes and
mounting of the stepper motors. This will allow for a builder to use belts or screws for
the actual mechanical movement. Does that make sense?
I do not see any way to avoid using stepper motors. I'm not sure how we could design for
both steppers and servos.
As for milling traces, I see no reason we could not place a resist pen on the Z head
and have the machine draw the traces. then place a drill bit in the head to drill the holes.
Then use an etching solution to complete the board. This approach would reduce cost and
make a desktop version a little more practical.
Hmm, I like that the more I think of it.
Notes.
There has been some concern about accuracy of a home made milling machine.
I think we need to keep this in perspective. This machine is for prototyping circuit boards.
I have seen the resolution of a tenth of a thousands of an inch mentioned.
I have build circuit boards by hand that were off by many thousands of an inch and
it went together just fine.
We are not grinding diamonds with this thing. I feel that a accuracy of a less than
a millimeter would be fine. Any thoughts?
Ive stayed out of this thread primarily because the scope of the project is a bit daunting. I think the mechanical aspect of it is what frightens me most, primarily through lack of experience. My biggest concern is the potential need for milled components to connect the stepper to the table etc, and the difficulty of an individual to find a cheap source for such compenents. I haven't noticed discussion of this point, but perhaps Im making a big ballyhoo about nothing.
What actually prompted my chiming in is about the resolution. I think to provide the highest value for the project the milling machine should be able to mill layouts using surface mount parts. At a bare minimum a resolution less than .65 mm should be obtainable, and .5 mm to enable all but the .3mm surface mount parts to be utilized. Obviously since I havn't volunteered for the project, you should take my advice with a grain of salt, but hopefully you will consider my opinion.
*CAD drawing of parts (if necessary)
*direct access to a fab lab (for two weeks more anyway)- as well as a father-in-law who builds precision parts for a living (and he owes me)
*Code (probably not as proficient as Bean, but still willing to help!)
I have limited experience in all sections but near none in the programming end (C/BAsic..etc..).
I am hoping that the group is quite public about their progress and that I get to learn along!
I just started my current position last year and inherited some old milling machines and workshop setups. I haven't played with it much (need a project) but hope that I can one day!
Anyhow, everyone will have a dream for this project...it's up to the moderator to weigh what's obtainable and practical over what's wishful and difficult.
Babysteps guys....work on the controls first and drilling holes....then up to the next part.
In my dreams (told you everyone has one) I would hope that there could be a learning mode for this setup.
Basically, if I were to take an existing board, could I guide the jig and mark where holes are. Then have these points saved in memory and then tell the PC to clone me a second board etc.... (this might be easier with drilled boards but gets harder when you are milling trace paths.)
Just a thought!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ ·
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 I have already built two XY positioners, perhaps I can add some of my experiences for you to consider. Much was learned the hard way.
One machine was for my (manual commercial) printed circuit drill, and the other was for moving an assembled board around under a microscope inspection station.
In my humble opinion, for a seriously functional machine, the following are some minimum requirements.........
Unless you have huge pads for components and vias, the drill needs to position to much better than 10 thou or often (on an etched board) the pad will be severed from the trace by not being centered on the via.
If you are planning to have the machine perform "etch milling", as is I believe the main purpose, you must have far less than 5 thou backlash/slop (try for 2 thou if you can get it) otherwise you will get very ragged edges, and produce unsatisfactory results except on the coarsest layouts.
Threaded rod "lead screws" are generally workable for positioning, (depending on their quality), BUT you will need to make anti-backlash nuts for them. Furthermore, with stepper motors, the motion with this type of screw will be agonizingly slow, which may be OK for the milling, but will drive you insane for the drilling.
The stepper will need to be sizable as there will be a fair bit of drag in the system, and you CANNOT afford to lose even a single step.
Drive the steppers with LOTS of voltage (3 or 4 times rated) if you want to maximize stepping speed, BUT THEN while the step rate slows down and stops, BE SURE to bring the voltage down to a very low level. Some (10% of rated) residual current is required when stopped to prevent pull-out.
Depending on the table design, you will need to consider to screw-drive both sides (synchronized of course) of the X table otherwise some off-balance friction is likely to make it jam.
DO NOT have the Y motion ride on top of the X motion. It compounds the slop and really is very unsatisfactory for anything but a toy.
DO make the X and Y motions relative to the base, similar to popular printers. This much improves the robustness of the system.
There are lots more points to consider, but I hope these few help get you into the right direction.
Comments
Routers are extremely high speed and built to withstand lateral pressure.· They are also available in a wide variety of sizes and configuratons.· Some of the cheap ones don't last but a few hours, but a good professional one will run for years.· Excellent bearings, good cooling, and well balanced 20,000RPM motor.
Generally, they tend to use collets to securely hold the cutting tool as 'drill chucks' tend to slip under pressure.
Nuts and Volts recently had an all plastic X-Y-Z table that is appropriate and easily built with a bandsaw or tablesaw.· The only special machining was to create a thread cutter tool and to fit the ends of screws to stepper motor shafts.
I believe you can do all that for under $500USD.· And you have a table that can switch to other devices as you wish.· At times you may want to have the table graphically draw where it is going.· This is a great error detection method as it does not involve actual material.· So you would change from router to a marker or pencil as required.
For slow speeds, you can switch to an electric drill with a light dimmer as a speed controller.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
G. Herzog in Taiwan
I think most ppl will lean more towards a metal device as opposed to plastic (but this would be discussed).
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·
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."
Lot's of the comments above sound like they are reinventing the wheel. Study John's ideas and experiences. Developing an effective CNC system involves lots of decisions, many which come with experience as to what works and what doesn't. John has documented his experiences in his rather rough tone. Its worth reading.
He also has a bulletin board where CNC users exchange ideas.
http://www.rcgroups.com/forums/ also has a discussion forum on CNC by hobbyists.
Post Edited (bobledoux) : 6/13/2005 1:24:13 PM GMT
Small PCB drills (0.015) want to be spun aroudn 20,000RPM. 0.001 (1 mil) of runout can destroy them very quickly, inside of a few revolutions. Industrial PCB drilling machines operate at 20-30k rpm with runouts in the neighborhood of 0.0001" (one tenth for the machinist types).
I'm not sure what the plan is for drill bits or engraving bits, but with the number of each available in the surplus market, I'd think we'd want to focus on them.
My $0.0125 (depreciated)
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
Another great idea. Software wise that is a perfect stepping stone to the trace milling.
If we can't get it to drill holes there is no point in working on the milling side.
Plus the machine would be of great use while the milling aspects are being developed.
That should be part of the development plan.
What a wealth of information on that page. That is the machine I saw in my "minds eye" when I made my first post.
Great link.
I understand what you are saying about reinventing the wheel, but I am not comfortable buying plans site unseen.
So, looks like things we be coming together on the 18th.
We should spend a couple/few weeks just talking about our plan of attack.
This project could easily spin into the abyss if we have a weak plan.
Also the serial protocol should be designed so that you can use any of the Parallax ICs. (stamp, SX, etc.)
Can't wait to get all theses ideas in one place.
So far it's pretty exciting, lets keep it that way.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jack
Post Edited (PLJack) : 6/13/2005 8:18:01 PM GMT
Thought those throwing in on this project might enjoy the book - it's not a technical as I'd like, but interesting none the less.
Ryan
That look like a great book.
I'm liking the phrase "desktop milling machine".
I think we will use that.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jack
His Brute has ways constructed from commercial building window channel that can often be picked up for free. The biggest cost is locating suitable stepper motors.
Its possible to build a brute for $200 with some scrounging.
He is a big supporter of Dremel power tools and argues that most other small routers have too much bearing play for precision work. Also note that a PC board cutter is a small tool. This permits small, light and inexpensive construction.
There are lots of ways to go wrong when armchair designing a CNC machine. I suggest contributers read available information before suggesting ideas that are "less than optimal."
Do you want a "scavenged" machine that not everyone will be able to find pieces for by scavenging? (as in some of our international brotheren)
Or make something that will work with common enough materials that it would work for beginners (relatively) anywhere?
Dremels are pretty common but a pricey item....some parts you'll have to bite the bullet in order to get enough quality that you get the production you want out of it!
Having something with an expandable work area is a must as not everyone uses the same size board.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·
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'm going to have to take at least some issue with the Dremel comment WRT runout. The one I have on my desk is brand new (opened the box yesterday) and has 0.007" of TIR. For a 15 mil drill, that's nearly 50% of it's diameter, and sure to destroy it. I think we're going to need a MUCH more precise spindle mechanism to do PCB drilling (ABEC-7 bearings come to mind). I have no issue with using a dremel or other thing as the power for such a spindle, but the more I get into it, the less I want to chuck my bits into the end of a rotary tool.
Of course, I don't know if anyone else shares my needs to drill 15 mil holes from time to time, so this may be more moot than I think, but whereever possible, I think we should aim for as much precision/accuracy/repeatability as we can afford. It will pay off in the end.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
At one extreme may be the builder who expects to build SMT designs using 8 mil traces. Such a machine will cost big bucks.
At the other extreme is the builder seeking simple through hole construction on 0.1 inch standards using 60 mil traces. Such a machine exists in The BRUTE.
The first machine may require expensive ball screws and ways. The second machine can be constructed with industrial threaded rod and plastic bearings and screws.
Its a matter of tradeoffs. How do you get a variety of reader expectations into a design? I read the various CNC bulletin boards and find the whole process is a bee swarm.
In terms of the Dremel: Its small and reasonabily precise, for the money. Most other tools are much larger which requires greater structural integrity and mass in the Z axis. This translates into bigger machines. Bigger machines may be difficult to fit on a small work bench. Most routers can't match the runout of the Dremel just because of the larger scale. A "precise spindle mechanism" is another way of saying an "expensive spindle mechanism."
What kind of budget are you willing to spend to get a machine? At what point does ExpressPCB or iron on etching sheets swing the fiscal argument the other way?
I'm not here to be a naysayer. There are good reasons there are no "cheap" CNC machines. Should Parallax be budgeting expensive staff time on a complex project without first evaluating the potential for success? Maybe, with the construction techniques they are using on their robots, they can extend into a basic CNC process. I don't know. If successful, it could be a killer application.
Post Edited (bobledoux) : 6/14/2005 10:56:03 PM GMT
I agree. I don't think a Dremel will fit everyone's use, not will my idea of a 20k RPM stable spindle with a one-tenth (0.0001) runout. I also don't think everyone will need a 12" Z-axis like I'm planning.
I have a hunch the project will need ot be broken up into modular blocks like this:
Control software (specific to the drive hardware?)
Stepper/servo drive hardware (USB would be nice here...FTDI chips perhaps?)
X-Y mechanism (cheap leadscrews? Acme threads? Linear bearings?)
Z-axis mechanism (travel? actuator??
Cutting system (dremel, rotozip, spindle?)
Discuss
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
I think we should define the performance characteristics of what we are trying to build.· Personally, I would be mesmerized by a machine that could make lands .012 to .015 wide, make square or rectanglular pads .065 x .065 or .050 x .100 and can drill a .030 or .035 hole in the center of the pads.· We aren't making an LPKF - we want to make something the average person can afford and use for prototyping, so let's not set our sights too high.· I could live with the Dremel or the RotoZip, both of which I have and use extensively.· After I've prototyped, I can go to ExpressPCB for more sophisticated boards.·
And them's my thoughts on that!
Sid
That should begin shortly.
My hope is that each milling machine built will be different from the others.
As long as the "IC" and the PC software can communicate (given user adjustable parameters)
there should be lots of room for modifications.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jack
So i think the ideal way to proceed with this project would be to divide up the software, and make it the best as possible and with all the feartures we would want. Although we would at least decide as a group on the stepper/ servo motor interfaces. And it would be intresting to see what everyone comes up with for the spindle motor and table.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
http://www.rcmodels.net/cnc/hobby/classic.html
http://www.rcmodels.net/cnc/system3/
Parallax's access to quality CNC means they might be able to develop"ways" designs that are beyond the scope of the hobbyist. Making precise X, Y, and Z axes, at a reasonable cost, is key to a project. (Hobby builders often employ extruded aluminum for ways and plastics for bearings, or round bar/tube ways with "skateboard bearings" for the carriers. )
A small machine for 4" by 6" printed circuitboards could be light and inexpensive, employing something like a Dremel tool. But if you want to mill aluminum parts, the machine construction is raised to a whole different level. A Dremel tool won't "cut it."
The Crankorgan Brute plans are beginning to be dated, as they depend on 74xx series chips for the servomotor controllers and use DOS machines for control. But the plans were specifically directed to circuitboard interests.
One issue with Windows machines concerns lost steps. The CNC software must make allowance for Windows interagation of open applications. Occasionally, a step can be missed, putting the axes out of sync with the program. Single applications running under DOS avoid some of these problems.
Measured my X / Y axis tonight and have .012 and .028 slop,·will need to get this taken care of. What do you have in mind for stepper control? After looking around on Google there are a few companies selling controls. I assume that the SX will toggle the driver, control forward and reverse and control travel limits correct?
I posted a thread over a week ago suggesting two people oversee the project, and included a "survey." There were zero (0) replies to the survey.
So at this point I am not sure if the project is worth pursuing as there was no response with regards to who will oversee the project.
Ken
My original idea was for the community to get involved with this project. I never expected Parallax to be involved.
I sure they are much too busy developing ingenious products.
I am still willing to moderate / manage incoming suggestions.
Design and write the PC interface for the mill.
And whatever else I can help with.
But I can not do it alone.
We still need an ASM / SX/B programmer.
Someone with sound mechanical design skills.
And lots of people to provide input and ideas.
If this idea dies it will be from lack of support from the community.
I think there were only 12 people that voted in the poll.
There is no reason we can't do this project in this thread as long as I am able to modify the first post with updates.
I think they have been posting support in this thread.
I guess I am confused about that.
I thought we were measuring community interests? I am ready to go.
Just waiting for a community member or two to volunteer their time.
So, I ask again, what is it going to be people.
There is always another software project on my list.
Should I cross out this one?
I have a list of starter suggestions for this project.
Everyone that wants me to start posting say "I".
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jack
Again, I will write the code for the SX to control the motors. (I assume steppers, but whatever we decide).
But I have zero mechanical ability [noparse];)[/noparse]
I will propose using serial communications from the PC to the SX controller.
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."
·
As I mentioned before, I have no problem doing layout for the board (Eagle or Protel), and initial prototyping if I can source the parts cheap enough. I came up as a EE, but I spent many of my college off hours working at the campus machine shop, so I'm not afraid to get my hands dirty at a mill.
I'm with Bean on serial comms. USB might seem like a good idea, but the way Windows handles timing for virtual com ports I'm afraid it might miss a step or overshoot something unless that's all taken care of in the controller.
The guys over at 8020.net have a eBay garage sale on blemished extrusions for some really cheap prices. The dimensional tolerance is always the same as normal production parts, they're just scratched or marked some way. I plan on using some for my machine. Strong, light, and easy to work with. (search for 8020 aluminum)
Are encoders for servo style feedback an option (mmm...servos....), or do we want to go with the microstep-and-hope positioning method? Both work, we just need ot define our accuracy, precision, and repeatability requirements.
-dave
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
This is not a sig. This is a duck. Quack.
Seems we have the beginnings of a team.
Jack - PC Software, protocol development.
Bean - IC code, protocol development.
Dave - Prototyping, circuit layout.
The Software.
Communication with the controller should be via Serial to reduce cost.
More on serial protocol below.
Software should run on all flavors of Windows from 95 up.
Graphic display of current position.
Ability to perform dry runs of a mill job.
Should be able to read at least one standard CNC file format.
User selectable serial communication speeds.
Ability to calibrate the stepper motors.
This list is endless, we will just take it as it comes.
The Controller.
Should be designed as one unit, i.e. a black box.
Designed as simple (low cost) as possible with the ability for
others to build upon the design.
The IC.
Unless speed issues are encountered it would be nice if the
controller could be build with any of the Parallax micro controllers.
With that in mind the IC code should be written in SX/B.
Assembly language is out of reach of most hobbyists.
The project should be developed with an SX and leave the porting
of the SX/B code to a Basic Stamp user.
Keep the above statement in mind when developing the SX/B code.
The Protocol.
The sky is the limit here. This is really an important aspect of the
project and needs to be worked hard before any coding begins.
Any change to the protocol propagates throughout the software and SX/B code.
We should make an attempt to avoid any serial hardware to reduce cost.
The software and controller should be in constant communication.
The controller should acknowledge a received command. Then acknowledge its completion
to request the next command. This should avoid the need to implement a hardware flow control.
Commands should be tokenized to reduce serial bandwidth.
The Mill.
An XY table with a Z head seems to be the preferred design.
A desktop version would be nice. But I see no reason the machine could not be
scaled up if needed.
The mechanical design should only implement the movement of the XY and Z planes and
mounting of the stepper motors. This will allow for a builder to use belts or screws for
the actual mechanical movement. Does that make sense?
I do not see any way to avoid using stepper motors. I'm not sure how we could design for
both steppers and servos.
As for milling traces, I see no reason we could not place a resist pen on the Z head
and have the machine draw the traces. then place a drill bit in the head to drill the holes.
Then use an etching solution to complete the board. This approach would reduce cost and
make a desktop version a little more practical.
Hmm, I like that the more I think of it.
Notes.
There has been some concern about accuracy of a home made milling machine.
I think we need to keep this in perspective. This machine is for prototyping circuit boards.
I have seen the resolution of a tenth of a thousands of an inch mentioned.
I have build circuit boards by hand that were off by many thousands of an inch and
it went together just fine.
We are not grinding diamonds with this thing. I feel that a accuracy of a less than
a millimeter would be fine. Any thoughts?
Anyway, just some food for thought.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jack
What actually prompted my chiming in is about the resolution. I think to provide the highest value for the project the milling machine should be able to mill layouts using surface mount parts. At a bare minimum a resolution less than .65 mm should be obtainable, and .5 mm to enable all but the .3mm surface mount parts to be utilized. Obviously since I havn't volunteered for the project, you should take my advice with a grain of salt, but hopefully you will consider my opinion.
*CAD drawing of parts (if necessary)
*direct access to a fab lab (for two weeks more anyway)- as well as a father-in-law who builds precision parts for a living (and he owes me)
*Code (probably not as proficient as Bean, but still willing to help!)
Ryan
I am hoping that the group is quite public about their progress and that I get to learn along!
I just started my current position last year and inherited some old milling machines and workshop setups. I haven't played with it much (need a project) but hope that I can one day!
Anyhow, everyone will have a dream for this project...it's up to the moderator to weigh what's obtainable and practical over what's wishful and difficult.
Babysteps guys....work on the controls first and drilling holes....then up to the next part.
In my dreams (told you everyone has one) I would hope that there could be a learning mode for this setup.
Basically, if I were to take an existing board, could I guide the jig and mark where holes are. Then have these points saved in memory and then tell the PC to clone me a second board etc.... (this might be easier with drilled boards but gets harder when you are milling trace paths.)
Just a thought!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
·
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 I have already built two XY positioners, perhaps I can add some of my experiences for you to consider. Much was learned the hard way.
One machine was for my (manual commercial) printed circuit drill, and the other was for moving an assembled board around under a microscope inspection station.
In my humble opinion, for a seriously functional machine, the following are some minimum requirements.........
Unless you have huge pads for components and vias, the drill needs to position to much better than 10 thou or often (on an etched board) the pad will be severed from the trace by not being centered on the via.
If you are planning to have the machine perform "etch milling", as is I believe the main purpose, you must have far less than 5 thou backlash/slop (try for 2 thou if you can get it) otherwise you will get very ragged edges, and produce unsatisfactory results except on the coarsest layouts.
Threaded rod "lead screws" are generally workable for positioning, (depending on their quality), BUT you will need to make anti-backlash nuts for them. Furthermore, with stepper motors, the motion with this type of screw will be agonizingly slow, which may be OK for the milling, but will drive you insane for the drilling.
The stepper will need to be sizable as there will be a fair bit of drag in the system, and you CANNOT afford to lose even a single step.
Drive the steppers with LOTS of voltage (3 or 4 times rated) if you want to maximize stepping speed, BUT THEN while the step rate slows down and stops, BE SURE to bring the voltage down to a very low level. Some (10% of rated) residual current is required when stopped to prevent pull-out.
Depending on the table design, you will need to consider to screw-drive both sides (synchronized of course) of the X table otherwise some off-balance friction is likely to make it jam.
DO NOT have the Y motion ride on top of the X motion. It compounds the slop and really is very unsatisfactory for anything but a toy.
DO make the X and Y motions relative to the base, similar to popular printers. This much improves the robustness of the system.
There are lots more points to consider, but I hope these few help get you into the right direction.
Cheers,
Peter (pjv)