ok, now this thread is sucking me in....
was surfing the web last night and looked at the basics of dancad and i'm thinking this would be easy with the propeller..
pc puts out direction and pulse for each stepper thru paraell port, ya write a object and pass it the direction pin ,pulse pin and 4 pins to the uln2803... then start 3 cogs with it so each gets that paticular pulse and controls its own stepper axis.... what am i missing here ? seems as if you wouldnt need a seperate controller board for each stepper motor, and as long as you werent tring to cut heavy you should be able to use printer steppers, geared down of course..hardest part is making the table... and i just happen to have some 1" x 24" slide table slides
dan
KK, Expresspcb can generate a .dxf file.· The first problem is - the .dxf can only be created from the silkscreen layer.· So, if you draw your etch paths with the silkscreen layer you can generate a .dxf file whoch can be converted to gcode using the ACE converter.· The second problem is - the .dxf file will contain all the moves you made but not in the same order that you drew them.· In other words, if you had 100 steps, the first 25 might be in the proper sequence, then it might jump to step 35, then step 65, then go back and get the missing steps.· I have tried several times to sort out a .dxf file into the proper order but it is next to impossible.· I have no problem converting gcode to a DATA list the Stamp or Propeller can read, if I just had a good gcode to start with.· The third problem -· drawing etch paths with a silkscreen layer is much harder than using a top or bottom layer - silkscreen lines can be difficult to edit.· I've tlked to Express about the way the .dxf are generated but they say that is the way the program works and there is nothing they can do about it.· They also said they had no plans to allow creation of .dxf files from the top or bottom layers.· So that is the reason I developed my own system, which may be slow, but it works perfectly.
Dan, the PC has nothing to do with the actual operation of the mill itself, at least in my case,· I use the PC to generate the DATA files I use and the actual program for the Stamp or Propeller.· The program and all the DATA is loaded into my Stamp which runs the whole thing.· You only want to run one stepper at a time.· The number of pulses required to move a table
1 mil is pre-calculated and programmed into the Stamp.· You tell it how far you want to go - it calculates the numbers of pulses it takes to get there.· In the case of a standard stepper - 200 pulses per revolution -· it takes 1015 pulses to move my Y table 100 mils.· I have it geared ddown 2:1 to give me greater precision so it actually takes 2030 pulses to move the Y table 100 mils.· I have an old Gecko microstepper on the X table and it takes 5156 pulses to move that table 100 mils.· Greater precision than the Y table, but it travels much slower.
Incidentally, I did away with the floating point math.· I found I could get exactly the same number of pulses using a simple formula.· In the case of the Y table, pulses = (travel in mils·* 2030)/100.· I think the whole program only uses 1 cog, but if I use the vga_hires_text object instead of the vga_text I'm using now it it will use two more cogs.
Making a mill computer controlled is not an easy task - far from it.· It has taken me a long time to progress to the point where I am now, but it was worth it.· Everytime I need a little PC board for something, I just go out to the shop and make it.· A 3 x 1.5 single sided board costs me about 50 cents - I have saved hundreds of dollars making little boards for various projects.
Dan, I guess you are suggesting the use of the propeller as a means of converting step and direction signals into phase signals for a simple bridge. You could do this or you could use turboCNC in phase output mode. Personally though I would go for some chopper drives otherwise you have to run the motors at rated voltages which are generally low, this means slow, or you end up adding power resistors which is OK but for the money something like the xylotex is probably better.
Sid, the source code for aceconverter is available, I suspect a man of your calibre could modify it to allow correct ordering of the moves or perhaps use it as a way of automating some of your data file production.
Graham, I'm going to play around with the .dxf files in Express.· If I can figure out how many moves I can make before it gets out of sequence, I could make several .dxf files and then just put them all tegether in my Excel template.· We'll see.
Express can produce a bitmap of the top layer pads and silkscreen components but I don't use it very much.· I usually use Screen Shot.· In fact, when I'm making a double sided board, I take a screen shot of the second side, open it with Picture Publisher and flip the drawing horizontally.· Now everything is reversed and·I use that as a guide when I'm drawing the etch paths - makes things a little easier.
Sid, can it output Gerber and aperture files etc? These should contain all the info you need, they should certainly contain x,y co-ordinates. Well worth a look at the format on some simple example pcbs.
Graham, Express can not output Gerber files.· It is a free, very proprietary program and only Express can read the files when they make a board for me.· You can purchase the Gerber files for $60 after you have a board made, but that would be pointless for me.· I have hopes I can do something productive with the .dxf files that are available to me.
I opened the dxf file for the holes as a text file, you find sections like this:
Circle
8
0
62
7
10
104.14
20
44.45
40
0.66
0
This was a metric file, I think 0.66 is the circle radius and 104.14 and 44.45 are co-ordinates. These could be extracted automatically, I realize this is only half the story.
Graham, I started playing around with a .dxf file and I remembered something else.· G codes start with the origin - 0, 0 - in the lower left hand corner.· Express, and most other programs I think, start with origin in the upper left hand corner.· So to wind up with a gcode that is the same as your drawing you have to set the origin to the lower left hand corner with the mouse, then do all your layout in the negative area.· I think tht is another reason I shied away from gcodes.
And there is something else, Graham.· Since you can only draw .dxf files in Express with the silkscreen layer, you have to ungroup all components that have silkscreen outlines and delete the yellow lines, otherwise the outlines will show up in the gcode.· Do you wonder that I stopped using gcodes?
Comments
was surfing the web last night and looked at the basics of dancad and i'm thinking this would be easy with the propeller..
pc puts out direction and pulse for each stepper thru paraell port, ya write a object and pass it the direction pin ,pulse pin and 4 pins to the uln2803... then start 3 cogs with it so each gets that paticular pulse and controls its own stepper axis.... what am i missing here ? seems as if you wouldnt need a seperate controller board for each stepper motor, and as long as you werent tring to cut heavy you should be able to use printer steppers, geared down of course..hardest part is making the table... and i just happen to have some 1" x 24" slide table slides
dan
Sid
1 mil is pre-calculated and programmed into the Stamp.· You tell it how far you want to go - it calculates the numbers of pulses it takes to get there.· In the case of a standard stepper - 200 pulses per revolution -· it takes 1015 pulses to move my Y table 100 mils.· I have it geared ddown 2:1 to give me greater precision so it actually takes 2030 pulses to move the Y table 100 mils.· I have an old Gecko microstepper on the X table and it takes 5156 pulses to move that table 100 mils.· Greater precision than the Y table, but it travels much slower.
Incidentally, I did away with the floating point math.· I found I could get exactly the same number of pulses using a simple formula.· In the case of the Y table, pulses = (travel in mils·* 2030)/100.· I think the whole program only uses 1 cog, but if I use the vga_hires_text object instead of the vga_text I'm using now it it will use two more cogs.
Making a mill computer controlled is not an easy task - far from it.· It has taken me a long time to progress to the point where I am now, but it was worth it.· Everytime I need a little PC board for something, I just go out to the shop and make it.· A 3 x 1.5 single sided board costs me about 50 cents - I have saved hundreds of dollars making little boards for various projects.
Sid
Sid, the source code for aceconverter is available, I suspect a man of your calibre could modify it to allow correct ordering of the moves or perhaps use it as a way of automating some of your data file production.
What other types of file will expresspcb output?
Graham
Express can produce a bitmap of the top layer pads and silkscreen components but I don't use it very much.· I usually use Screen Shot.· In fact, when I'm making a double sided board, I take a screen shot of the second side, open it with Picture Publisher and flip the drawing horizontally.· Now everything is reversed and·I use that as a guide when I'm drawing the etch paths - makes things a little easier.
Sid
Graham
Sid
I opened the dxf file for the holes as a text file, you find sections like this:
Circle
8
0
62
7
10
104.14
20
44.45
40
0.66
0
This was a metric file, I think 0.66 is the circle radius and 104.14 and 44.45 are co-ordinates. These could be extracted automatically, I realize this is only half the story.
Graham
Sid
Sid