[High Seas]

Greetings - Here is a new thread to assemble the "stuff" we've got going for the OpenSource DIY CNC - mother -of -all things.

The name need a little work - but we'll get plenty of help I'm sure!

Started this thread to give us a place to start the voulnteer list. Collected the following names from the other thread - and everyone have a review and validate or add your name, any thing you need to say, maybe even indicate any preferences as to which team(s) you'd like to be involved in

Please make sure you've had a read of the documents that follow:
the Roadmap;
the intro to IPT;
the Draft Charter; the Mission, Vision, Goals (draft),
and
Timeline.

These can be considered "DRAFT" and offer a baseline for discussion. That said, the sooner we settle on them, the sooner we get our project completed.

The thread pminmo started is comming along and represents the type of communications I believe we'll see.

[High Seas]

Please Review the following: This is the Charter - It what we'll be getting on about:

[High Seas]

This is a copy of previous posting (if you've read it before it shold be the same)

[High Seas]

Here is a sample of the proposed Mision, Vision and Goal statement for the effort. Teams will need to review, validate, develop, ratify, theirs and this set as well. Offered for our discussion and consideration.

[High Seas]

We need some sort of schedule to get things going and organized - here is a target we can all take a shot at:

Week 1
Within each Element:
Team members volunteer and choose the element(s) they wish to work in
Teams develop the organizational style they wish to operate in (flat, tree, nuclear, etc)
Teams develop a means for self assessment Team members identify the components and processes associated with their element "virtual brainstorming"
Team members develop a means of cataloging and recording their activities and findings
Team members identify their interfaces and constraints with adjacent elements
Teams allocate their resources (membership) to the tasks
Teams initiate a "test" communication with their adjacent element
Teams identify and correct any problems in the process

Week 2
Teams begin communication with each other
Identifing which team is the fixed constraint - and which team will record the information
Teams continue communications until the interface is fully documented for each Size and Style as identified in the
"Roadmap"

Week 3
Teams continue communication with each other
Teams continue cataloging and recording their activities
Teams begin developing alternatives and means to present their results
Teams begin integrating their results into alternative design "schemes"

Week 4
Teams begin assembling their final products
Teams/team members begin developing a draft set of mini-plans using their products

Week 5
Final preparation of product(s)

[High Seas]

I reviewed the posts from start to finish over the weeked and was very happy to see that we're getting new volunteers all along. I compiled a list (as of 04/18/04) and I may have a few off but ask for you to review and make your volunteer statenments on the following thread:


Volunteers to date - these have sotta clearly oput up their hands - but beofe getting too far down the track:

BENHILL
berin
cadesignr
CamIam
chuckknigh
Clyde
Hack
High Seas
lsfoils
teilhardo

Potential Volunteers - Have expressed interest - but not quite there in their posts - I read quickly though:

arvidb
dart70ca
Ferenczyg
HiString
JEFFY
ML1
mvaughn
samualt
vacpress

High Posters - You guys read a lot and post well too - any thoughts as to joining? Feel free to raise you hand.

Al_The_Man
MrChips
turmite
wbjzone
ynneb

Review the previous documents then raise you hand - The timeline, MVG, and charter are all in draft and subject to revision - I think we're pretty much settled on the structure (9 teams) though - any discussion though we ought to clear up before we go to far down the track.

Cheers - Jim

[Ferenczyg]

Count on me for the orange-cyan blocks of the cpt charter.


Fer

[CamIam]

i think i would be most helpful in the mechanicals, frame design style linear ways team. Kind of a nuts and bolts guy i guess.

[BENHILL]

Well I design electronics, I can wire a mains plug too...lol. I'm disabled, so I don't do running!

I can check ccts for those who may need that. Seems pointless to design a driver when others have such pro work almost finished.

Tell me what you want, & I'll see what I can do

[Ferenczyg]

Originally posted by BENHILL

Seems pointless to design a driver when others have such pro work almost finished.

Agree 100%. Maybe good work can be providing a good one side pcb, easy to make and with components easy to locate.

Tell me what you want, & I'll see what I can do

Same for me, pls.

Fer

[berin]

We should get a show of hands on how to design the PCB electronics. A: what kind step or servo? Uni-polar or bipolar. My money is on step-unipolar just because its cheap and easy. You don’t have to worry about run away servos not to mention it doesn’t require a complex controller unlike the bipolar motors. The next thing is how to drive it? Straight phase control, or do we want to add a bit of logic? If we wanted to add a bit of logic. Adding 2 chip’s an xor and a flip-flop can make a nice 1-2-3-4 stepper controller. Even better it can add stepping control and direction from a source of just 2 pins. One for step one for direction. Plus it can be set for step or half step. The easiest would be by far 5408 or what every they are called series. But your limited by what you can do for power. So high power would require PnP power transistors.
Lastly do we want to brake up the step source and driver circuit. Or make one big all in one board? Broken up has an easy fix and repair design.

[Ferenczyg]

My first bet is for stepper+unipolar, because the cheaper electronics. But in a second look the only economic difference between a newbie unipolar board and a newbie bipolar board are twice power semiconductors for the bipolar, 4 for unipolar against 8 for bipolar, per axe. Below 10$ of difference per axe probably. Bipolar of course will be a little more complex due the H bridges etc. but can be useful for two types of motors, bipolars and bipolar wired unipolars.

For the electronics i thing that step/dir is better than phase control. Step/dir is the usual way of work for the proggies out there (turbocnc has phase control but this will limit the useful life of the board I think). For the logic I bet for take profit of the IC's already designed: allegro 5804 or ST297 for unipolar and ST297-298 pair for bipolar. All of them are cheap and easy to locate almost all over the world (sorry if not, but in my country is difficult to locate a lot of components, so my feel is that if any is here, is almost everywhere ).

Fer

[berin]

There really isn't much of a difference between using phase control and step/dir. The big differnce is in the fact that, there are fewer programs which support phase control. Thats really about it, aside from the fact that your also cut down on other control and sensor features.

An other thing I'm looking at is I'd put out a request to you high seas. I think we need a concept web put up on the site somewheres. So people can vote on ideas, giving us all a better direction of where were going to go. The center of the web is cheap easy cnc DIY. The first branches would easly be design style, and motor as well as driver. The more techy stuff would come there after. I figur a design kind of like the basketball playoffs layout.

[High Seas]

First - thanks to the "new volunteers" and it is helpful in having an idea whare you'd like to "fit" into the plan/program. I've been in and out on this a bit and haven't done a reply - waiting for more to raise their hands - I guess.

Thanks to berin, Fer, BenHill, CadIAm, for raising you hands!!!

An other thing I'm looking at is I'd put out a request to you high seas. I think we need a concept web put up on the site somewheres. So people can vote on ideas, giving us all a better direction of where were going to go.

berin -- I like that idea - sort of a "brainstorming sticky?" Could you take that action and send a note to CNCAdmin detailing how and where? If you could - pass alon that now would be a good time to build the new Forum he and I discussed. We'd need some sort of name - Ideas anyone - simple is easiest! The brainstroming sticky could be at the top of the forum and the 9 elements beneath. Give Paul the list of names too from my previous post and we'll be getting started!
As an alternative, Paul if you're lurking about - could that happen now? Names are several posts upthread.

For all:
The concept is; the 9 elements have write access from the list of names -the participants in this experiment. All cnczone members have write access to the "brainstorming sticky"( or what/how you think we ought to call it.) with the 9 elements on the forum as independant threads - then the participants can focus their inputs in the right thread and keep it a bit organized!


Ok - more ideas and volunteers! Keep 'em comming - so far I've seen no negative feedbacks on the timetable/timing, guess thats good?

cheers - Jim

[berin]

I hate to say it but I dont know how to? I've looked all over the cnczone site but can't find any thing related to cncadmin. Mybe I'm just blind, then again I do ware glasses .
Also I had a few more IDEAS for the software guys. Now we all know dos is the way to go on this system becuase windows doesn't do direct access to the par port well at all. I wonder how hard it would be to create a type of NODE software that runs in dos with network support. Basicly you create a dos program to control the cnc hardware via phase or step/dir. Now basicly all the node system is, is just a cache system. It only does what its told, but this opens up an entire free range for upgrades and improvments. you could use windows or linux or what ever on a nice high power system where your doing all of your modeling ect.... Now you can control the node with any type of advanced cnc control LIKE LOOK AHEAD. Just think of the possiblites. Maybe turbocnc can get some updates????
I can give a great idea of the types of computers which are now considerd throw aways. But woud serve a great as a cnc controler. (I know because I use one of them with great results) almost any 386-486 laptop or better yet the "fujitsu stylistic 500" computer. I have a bunch of these, just because there great! they can use a pcmcia CF card as a boot device so every thing is solid state. they have a built in BW 8" screen with a hard cover on it (designed for stylis pen). An other good example are all of the thin clients you find on ebay that have parallel ports. most go for $25.

[pminmo]

Berin,
While the use of old PC's at first sounds like a great low cost controller, in practicality I quesion if it is. Too many different mother boards, clock rates, peripheral chip implementations of the parallel port, to many BIOS differences.

The Windows issue needs to be addressed, using the parallel port for real time control is the problem.

To me utilizing a COTS micro board is a possible solution. It would act as a cache, convert the serial stream, do the computation on rate and issue a pulsed parallel command to appropriate drivers. Then depending on desired implementation the user could select a chopped board, unipolar....

If the CNC software packages were modified to use the serial port, it would become less of an issue. At 128K baud a serial stream could effective keep up with three steppers synchronously at 10K steps per second, faster than most want to be stepped at. An 8 bit protocol could be developed such that each byte could represent a command in the first two bits with the following 6 being info:

Bit 0 1

0 0 Step command
0 1 Spindle coolant on/off
1 0 Rate
1 1 OPEN

For example

| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |

0 0 DX SX DY SY DZ SZ

S = 0 no step, S = 1 Step in the direction specified by D

USB would be even better, but the implementation is more technically involved and less off the shelf hardware.

Possible low cost boards that could do the job:
http://www.futurlec.com/ATMegaDevBoard.shtml $30
http://www.futurlec.com/ATDevBoard.shtml $20
http://www.olimex.com/dev/avr-p40b-8535.html $14
http://www.olimex.com/dev/avr-p28b.html $14
http://www.olimex.com/dev/avr-mt.html $30
http://www.olimex.com/dev/avr-mt128.html $50

Phil

[ESjaavik]

Your thoughts are very much in line with what I think. The PC running windows could be good enough if the realtime task is left to its own processor. EMC is running a realtime functionality wedged in under Linux, and other do similar stuff with Windows, but I have always chosen to hand over realtime work to one or more separate processors. One thing to consider though is how the RTprocessor should handle the cases where it is starved for data from the non-RT part. Especially W2K is notorious for going out for lunch up to several seconds.

Also issuing step commands from the PC will not solve the RT problem. The step pulses needs to be generated by the RT processor. The PC have raw computing power, so is eminently suited for everything up to calculating the trajectories. In essence establishing where the axes should be at each (choose your time step). Then pass this over to the realtime part that can make sure the time steps are equal and that the steps issued between them are correctly distributed. Sometimes not evenly, like when ramping the speed. The RT processor needs to have a buffer of timestamped target positions so it can trundle on during short data starvations from above.

USB is no problem with this: http://www.ftdichip.com/
It should be a strap option as serial ports are now disappearing from PC's.

If making a board for this purpose, could it be made to either handle the steppers<->PC interface or a pendant depending on component population and program? I have played with the thought of making a handheld replacement for the now disappeared handwheels on my mill. I would be using AVR as I have emulator and other helpful tools for it.

[pminmo]

ESjaavik,
I think the personal ultimate intermediate box (for lack of a better term) would take usb in and take a G-Code stream. (More complexity) The output the three axis in sync with direction and step pulses. I say step and direction pulses as the application might be for unipolar or bipolar and full vs microsteps. Let a 5804, A3977, L297-298 circuit handle it from there. I can't see a microcontroller handling micro stepping, chopping for three full axsis as well as the communications and computation.

I agree that the PC host should not do the pulses. In my example above when a byte arrives the RT micro would basically output the direction lines(3) then do a pulse edge (3) for step, delay for a couple of microseconds, then return the step lines (3) to the starting state. With one byte of information the RT Micro then does the actual step. If the step bits were 0 it wouldn't issue and associated step pulse with that byte.

As far as what to do with a lul while Windows goes off. On the intermediate box it would cache a segment amount of data before it ever starts. Or the PC application would send steams with definite start and stops built in. i.e. do this segment disengage from the cut, pause load next segment. The intermedaite box would never start a segment until it has the full stream for the segment.

AVR's are GREAT!!

Phil

[High Seas]

I'll show my ignorance here - doesn't Mach2 set up its own "driver-thing" so it sort of bypasses the windows "thing" to work iout some of those windows hang-ups? (Note the high-tech references )

Wonder what the cost/performance trade-offs would be. Certainly there are heaps of PII/PIIIs out there that could run the dos based systems and a "card"/box/driver/cashe-thing that prevents the "drip-feed" loss of steps is a big plus too.

Certainly these are things a newbie would want to know as they line up the $$$ to spend.

You guys make me smarter all the time! Great inputs!
Cheers - Jim

[High Seas]

Our CncAdmin - - Paul is going to set up a seperate forum for our team's activity. With a separate Forum - we can have the 9 elements as individual threads, and we can post our efforts there in the appropriate thread - and move them around I suppose if needed.

The big plus is it gives write access to the team members only and read access to all. Sorta puts us in a fishbowl - the newest "reality-online-show?"


Title: DIY CNC Quick Start Team -- "Conferrence Room" - or something like that. We outta get a name - but thats not neearly as important to the tasks we've cut out for ourselves!!! (I'm sure we'll get lots of inputs
1 Sticky called "Brainstorming Space"

the 9 Threads -
Controllers (DIY designs, Others plug and Play syatems, etc)
Mechanical (Frame designs and Styles, Linear Ways, etc)
Motors (Servo, Unipolar/Bipolar Stepper)
Motion Transmission (Acme, Ball, Allthread Screws, Cogs, Belts, Wire)
Physical Attachments (fasteners, connectors, etc)
Power Supply (DIY and others)
Software
Tool (Spindles, Routers, hotwire, etc)
Organization (a catch all - - Graphics, display, Project Management, ect)

Here is a list of the Team Members as of 23/04/04 (you should all have read and write to all the above threads):
arvidb
BENHILL
berin
cadesignr
CamIam
chuckknigh
Clyde
dart70ca
Ferenczyg
Hack
High Seas
HiString
lsfoils
JEFFY
ML1
mvaughn
Pminmo
samualt
teilhard
vacpress

I may have "pushed" a few of the volunteers that seemed to be teetering on the edge of their seats - hope I didn't offend or missinterperet your "signals." Hope I didn't miss anyone that wants to participate - we can update if needed.
Start looking for the new forum at about 1800 GMT - 04/24/04. Pauls said he'd load it overnite - Thanks again to Paul the site admin!! :cheers: to Paul!
Cheers - Jim