Heres the first 4 hours of a potential design:

http://www.vacuumstudios.com/painter1.jpg

http://www.vacuumstudios.com/painter2.jpg

The friction belt is driven directly by the stepper which has 10x microstepping driver

the airbrush is $10.00 from harborfreight!

http://www.vacuumstudios.com/painter3.jpg


http://www.vacuumstudios.com/painter4.jpg

any solidworks users want the files? i would love if someone would do some mods.

this is the last digital image. i have all these parts allready, thats what inspired the particular design choices. i will take apicture of them and make a parts list.

http://www.vacuumstudios.com/painter5.jpg

incredible drawings. solidworks?

yep. solidworks. started at 2:30am, this far along at 6:30am

get some rest. you look tired.

yes! Incredible! ...have been following this and related threads....

if anyone ever needs a very fast CAD\3D Designer....


anyone with experience using friction belts got any sugestions for the design of the "tensioners"? i just have 2 nylon bushing holding the belt against the pulley now.. i think they may need to go closer together? should i put one on a spring-loaded cam?

One thing I like is that if i update these photos, they update here also, so i can just change these to new ones when i do another few hours modeling. gonna put in all the bolt holes, and bolts and screws, "limit switch" whiskers, "optical" home sensor & tags. etc. i am also going to design a pneumatic manifold and a 3 color printhead.

Any good ideas?

..just a thought...
you could incorporate a toothed pulley on the motor and use a matching open-ended toothed belt to make a more positive drive.
the nylon bushings could be on bearings to "roll" with the belt.

the belt is a belt with a bunch of tiny teeth, and so is the pulley. its from an old dotmatrix printer.

heres all my parts i have. all i need is to cut the acrylic, assemble the thing, and get 2 more xylotex drivers.

the small stepper has gears and will push the lever on the airbrush via the bike brake cable.

the tablet PC can run turboCNC, or other control software. the LCD will be used to configure the PIC microchip for the multicolor printhead when i do that.

http://www.vacuumstudios.com/painterparts.jpg

parts list, cost, and sources list forthcomming, as well as construction pictures.

i have been waiting to have a project that i allready have all the parts for. finally! i am still collecting parts for a big "production" machine.

Hey vacpress I was just wondering if you need a home? I have always wanted a son:banana: :D !! Now of course you would need to bring all your drawing tools with you cause I think I see some potential in you but it just needs to be developed....uh on my projects!:D

Great drawings btw. I am curious as to what the use for this machine would be? Sign painting or ????

Mike

the use of this machine is to massively waste the time of its creator!

what else?

its also the first step towards a www.pixation.com type product. i am thinking about using oil pant with little hoops running into it, and air jets blowing it off. the fun part is going to be programming the color-mixng algorithms.

my drawing tools are nuthin-just pens, markers, and computers.

its my machine tools you want.. check my user gallery. :)

Here is a partial parts list and sources file.


http://www.vacuumstudios.com/partslist.rtf

I would move the Paint gun motor up, and use larger idler wheels with a bearing (or even just use two roller skate bearings side by side) where the nylon bushings are.

This thing will just paint a line acording to a gcode at this point. One color, correct? (as drawn at least).

Eric

move the motor up? further from the plane the belt is mounted to? i have the belt on .25" pads on top of a extrusion it rests on. the .5" nylon "bushings" are actually bearings on SS rod. is there an advantage to be had by going to rollerblade bearings? slightly less drag?

yes. as drawn, the thing only paints a line. it can also control the thickness by varying the actuation of the trigger. if i switch to a different style of paint delivery, perhaps like www.pixation.com with a driven wire or a plastic hoop, or something, then i will have much finer control.

one thing i am invisioning that isnt a pixation rip, is just putting 4 of those cheapo paint guns in a star pattern with their nozzles all pointed at the same place. fueled with CMYK process color semi-opaque paints, this may allow for a decent color pallette..

I have just started reading about opensource RIP drivers, perhaps there will be an off-the-shelf solution. If not, it can just run G codes. I bet i could figure out a way to do 4-color through g code and DXF import into mach2... No idea for sure though..

Vacpress,

Thanks for posting on this forum. Your ideas are always refreshing.

A couple of thoughts, maybe dumb but that's never stopped me before.

1. More your airbrush out from under the gantry. It will be hard to add more paint to the hopper with it tucked under the drive mechanism. Alternately, you'll need a quick disconnect for air brush removal for cleaning and such.

2. For your friction drive, I wonder if you could just mount the belt flat on the bar and have the stepper motor spring loaded so there is a small amount of tension against the drive roller.

3. Instead of the airbrush, you could mount a swivel knife and this device could do double duty as a large scale vinyl cutter.


Good luck.

Trent

congratulation with the solidworks drawings.
How far you are with the electronics?
Whene you want you can make colorpictures to with the one spraygun.
You can first paint the first color.
Thene clean the spraygun and paint the second color .
You can do this as much colors you have.
Finaly you have paint the black color to make contours to all the other colors.
This is the same principe like painting with the airbruch by hand.
There you can also make color paints by working color after color.
Whene my milling cnc is finished completly i will start with a project like this.
I just have to make a vacuumtabel to my milling tabel.

buscht-i like the swivel knife idea... it changes some stuff about the carriage. would have to add 4 bearings to hold the thing down.. no biggie..

publitime- thats to much hassle. it would be better to have 4 nozzles. and the electronics as described are Off the shelf. you could build controller boards, but when there are xylotex, why?

so far it is designed just to use 1 color and act like a plotter. to make it full color would require only like $40 more cost... its making a RIP driver that is a challenge then.

Im going to say it again Vac, nice job!! You have so much to contribute and your open-source philosophy is very heart warming. We need people like you to run the big corporations out there whose guiding principle is self-interest and profit. How fast do you hope to have it going with the steppers?
How are you going to change the colors for such complex pictures as on pixation?
-Tei

vac

did you have build the controllerboard yet?
Whit 4 colors it's not that simple to get a fullcolorpicture.
Whene you see a photo you have split in photoshop to cmyk the you can see that
you will have places who are dark colored and other places who are colored light.
So the spryguns have to be controlled permanently and with a very fast speed.

teilhardo- yah. Ef the man! that bastard.


as for how this thing would make pictures. with 1 cololr its fairly simple - fill in the grid. with 4 colors its much more complex. it becomes: fill in the grid by spraying the right CMYK colors.

To do this, a few pieces of software need to be adapted\written. One piece has to breakdown a bitmaped raster image, convert it into values such as x,y,color. Then it needs to transmit that information to the "print engine"

the print engine needs to be able to:

1) read the x,y value of each pixel and locate the apropriate print nozzle over than location.
2) decide what color to spray out to achieve the desired color. so, yes, lots of software and possibly 1 piece of hardware needs to be developed for this...

the easiest way to do this is probably with a "post processor" style arrangment. basically you would

1)load your fullcolor, highquality image into the processor engine.
2)the engine will break the image into a grid, and assign a color value for each coordinate location.
3)the engine will control the print mechanism as it plots these color values.

this is why the wirejet uses those steppers and wires - to maintain an accurately metered, yet very minute amount of paint. The wirejet has 3 nozzles that direct the each paint drop to the work material. someone should work on this.. i have found other companies use the sme concept as wirejet, so that means it must be valid to dopt their idea, as it may be more universal than they make it sound on their site. I checked out their pattents, and the drawing of their nozzle array could prove usefull.


if someone would begin to do real research on the software, while i constuct this hardwae, that would be really cool.. A place to start is finding out exactly what a "RIP engine" does.

Best Regards

something else worth persuing is the same wheeled ganty idea for VERY large vinyl sign making. Using my design as a starting point, a 24foot wide gantry could be constructed... hmmmm..... tarps for sign material... hmmm...

worth mentioning - such a device can also cut nylon fabric very accurately by holding a soldering iron type heated cutting element.

if i remember correctly, there is some smple math to keep a knife's cuttong edge perpindicular to your cutting direction - i saw it over on luberths site a year or so ago...

The xylotex can be cheaper unipolar drivers - but then the thing will go slower


pulitime-did you read my .rtf?

ok. here is a quickish 4-color printhead concept. the PCB is a db25 breakout board i figured could do double as a wire distro point on the printhead carriage. not modeled yet are the steppers or solenoids that control the brake cable, also not modeled. Maybe soon. Anyone got a model of a Xylotex Board? or do i have to do everything? :)


http://www.vacuumstudios.com/4colorsml.jpg

all these transparent parts are acrylic. i am making them different colors for clarity. acrylic is nice because when machined, it can attain tight tolerances.


http://www.vacuumstudios.com/4colorsml2.jpg

next up : a swivel knife assembly, and a hot knife assembly. for cutting paper & vinyl, and synthetic cloths respectively.

Vac! You astound me with your renderings...and the quickness!
They sure look like they will work, but then I only dreamed of these type of machines after visiting pixilation....
How do you do it? Is the control and machanics basically a glorified printer-operating with much higher volts and amps?
...and yes, I've been dreaming of toolchangers, and laser digitizers, and variable speed spindles, and....etc..etc...etc...:)

P.S. the pics are still huge....are you using low res files? they sure don't appear to be.

I used to a printer tech so I've got a few thoughts.
Don't try and angle your nozzles to try and hit the same spot. The dot shape from each nozzle will be differrent and one color will bleed out from under another. For example a green dot made from cyan and yellow will be cyan on one side green and the middle and green on the other. Space the nozzles apart a interger number of pixles apart and handle the offset in software.

You might want to check out Ghostscript, www.ghostscript.com its a open source postscript interperter (RIP) which as well as all the standard printer drivers can also output simple bit map files with the required pixel on/off type of format.

tachus-good point - the angle affecting dot shape. i am probably not going to build anything like that 4 color printhead. i am allready looking at a system that will allow much smaller amounts of paint to be delivered to much smaller spaces. possible a thin plastic or metal 3" ring being driven through a pool of paint, and past some very small nozzles. with 3-4 nozzles, i can make much smaller circles.

if i try to use HP RTL, how do i take the info from a generic driver and store it, and then print it? can i emualte all that in software?

i am a fairhand at "PicBasic" style microcontroller programming..

for now i am going to just build the roving gantry, as i have all the parts on my table.. just gotta take the time to do it right or it will go in circles! any slop in build on this guy will show after a foot or so of movement.

Vac, Just saw from you post in another thread that you are already looking a ghostscript, here is a link which might be helpfull if you havn't aready found something similar http://www.ghostscript.com/doc/gnu/7.05/Devices.htm#Uniprint
It looks like you could make a custom upp file and have it output a sun raster file which could then have the head movement commands spliced into it by a simple script.

tachus- yes i am discussing this with someone in anothe thread - the script would read the raster iamge pixel by pixel, line by ine, and move the printhead to these locations.. any elaboration on this script would be lovely :)

what is this process called? where can i find an example? who else has done this? how hard is it?

the uniprint document looks very promising. one question.. if i define all theis information for the machine in this driver, what actually takes the pixel\colro data, and gets the mechanical system to spit it out? i like how you can set offsets between print nozzles with unidriver.. thats something we will need.....
i am gonna start building the gantry in a few hours. I may try my "chepa linear bearing" idea since i have really never used my lathe for much other than making model components, and making motor couplers. printing the templates out of solidworks is so obnoxious when it takes 3 pages of 81/2x11 all cutup nd taped together.

Vac, you mentioned using PIC basic...have you settled on a particular PIC? Just curious. I'd lend a hand with the PIC code, but my experience is strictly PIC assembly, and probably more limited than yours. But I'd be happy to lend a coding hand, if need be.

I know people keep saying this, but those drawings are incredible! You're pretty handy with the 3D.

kcoaks- ive used PBP to make robots and stuff with PIC16f84A s


and the drawings: thanks! i love compliments, as its something I hav tried very hard to get good at..

Vacpress I can see already this adoption thing I proposed is not gonna work!:D These folks are not gonna let you go.

I have a couple of commets and only hope to help what little I can on this open source thing.

1. It looks to me that the paint guns you have drawn are in reality jamb guns or touch-up guns instead of airbrushes.

2. While thinking about your project I envisioned a huge, yet lightweight frame that could be mounted vertically and programmed with the layout of the front of a house and viola or whatever that word is.....a cnc house painting machine.

I agree with everyone else here that for a young person that has tons of talent you also seem to have a bigger heart.

Mike

<smile>
i think im just an MIT wannabe... im in art school. it somtimes leaves me craving math..

the paintguns are in fact "detail" guns, or touch up guns. on the package it said "detial airgun for..." in some bad asian translation...

I am going to move to something much more like pixation is doing when i move from a pen for testing upto a "printhead" i am actually about to go walk into the shop to start building this, after i print a few cut templates out of solidworks. sucks, i dont have a bandsaw here. im gonna try and do it with a scroll saw, if that dosent workout, i am just going to machine some motor couplers and maybe give the v-bearing idea a shot. (see holy grail thread).

its odd so many of the "gurus" here have remained silent on this. i was hoping for some really rigorus criticism and saugestions..


thanks for the input so far though! i will photograph and post my evenings progress. i only hve like 3 or 4 hours before i gotta go see my girlfriend.. darn. :)

I am so jealous. If only I could draw like Vac in Solidworks.

Keep those concepts rolling in, its soooo inspiring.

I still want to see a CD coater launcher design from you Mr Vac.
That neighbours cat would make a great target.

I am currently building a 1/2 scale model of foamcore and balsa, since i have decided to wait and use a bandsaw to do the main body cuts isntead of my crappy scrollsaw. Heres the first picture of several i have prepared this evening.

I also posted about 15 messages in various yahoo groups looking for a programmer to help me with the "raster to step\dir" problem.


http://www.interinar.com/BSD-01.htm @$26 this driver may work great for the 3 steppers that need microstepping.

solidworks is easy to learn if your a graphics freak like i am.. the first time i used it was like a revelation - something like the first time i played with legos...

to really learn solidworks takes a year though.. my first projects didnt look like this.. if anyones really curious, i can start a thread "the 2 year development of solidworks mastery" or something.. show some images from my first projects... its inspirational for me at least to see people go from suck - to success..

vac

I think HyperXpress will be intressting to RIP an image and thene import it into a CAD program.

publitime - checkout the "free rip" program on the pixelation site. it may provide a key also..

and searching on the perl GD libraries is also very usefull, and someone recomended reading something called the "comp.graphics faq" so im gonna hunt that down for bedtime reading.

gotta do this homework... whats HyperXpress?

I FINALLY found this thread again! hope I can add to the game.

I recently wrote a VB program converting a BMP into gcode for 3d engraving on our bridgeports here at work. Your coordinate system does not need to be complicated since your going to run through your entire image sequentially. i.e do complete first row - return home - goto next row - do complete row - return home. Yes you can add complexity if you want to print bi-directionally - but lets walk first - run later;-)

Keeping this Raster frame of mind eliminates the x,y issues, but keeps you color mixing at the print head. Since all graphic images are broken down into pixel grids all you need is to take the color values at each location and time the 'spray' based upon its value. The value is relative to the sprayer and will need to be fine tuned during assembly.

My 3d engraver used the color density to determine depth on the z. Your case would have to send all 3 parameters separately for each color. Very do-able in my mind.

jeff

can you give littel more info about how you see the controller for the printer?
What electrical components are needed to run.
With what program will you run the whole thing

jeffterm - thats what i am comming to realize. i am also building a prototype. i was going to use acrylic, but have been forced to compromise the design because acrylic costs to much. so sad, but the 4 pieces i need will aparently cost me $40. i am far to poor to spend $40 on acrylic. The guy who I normally see wasnt at the plastic shop today. SOO, i gotta either find a cheaper source for some 1/2" acrylic or, change to MDF or something.

the software you describe sounds interesting. i have decided basically to try and get a 48" wide gantry up by the weekend and on wheels. i have a tiny computer im about to put an array of DOS CNC software on to test the thing. I just need to order 2 more xylotex drivers, but have some 1-pulse/step unipolars i made that will allow me to get the thing moving as long as i use unipolar motors. im gonna try and make it have auniversal nema mounting scheme tho..

jeff- for the latest software scoop goto:

www.luberth.com and checkout cstep - thats gonna be for the vector mode. AND
www.pixation.com goto the downloads site and checkout the "Free RIP" program. that may be for the image prep. Or ghostscript. Then it needs to be decoded and printed: custom perl script, maybe even in an IC, with a windows style driver that sends the program and parameters to the board.

You may have an easier solution - a simple G code scheme with a seperate command for each color, and a good post processor application that generates the code?

there is a GCODE in a chip opensource Atmel project. I bet that could be customized very easily to work specifically with the program you created. The program you made needs a few thing smost likely to be just right : output resolution, output size.

Tell me more!

SO. Axrylic:

$195 for a 4x8' sheet
roughly 7$ sq/ft

its wierd, last time I went there, the guy who i delt with was really cool, gave me lots of stuff for $10. this guy wanted me to pay $45 for a 1/3 sheet. sooo. i said no thanks.

vac

i can mail you very good cnc software that is running in DOS.
It will translate a hpgl to G Code.
It's very easy to use.
The software is from one of my friends who lives in germany so its German language.
But my wife did translated it to englisch.
It is 2,5D cnc software.

vac
I'm thinking to see if it is possible to rebuild a injector to a printhead.
I means the injectors you find in a gasoline engine to spray the gasoline in the engine.

thats an interesting idea. how many nozzles are in an injector?

OK, I'll see if I can keep up with you guys on this. If work does not get in the way.

How prototypical do you want to proceed with this just to get a proof of concept? We could make some quick decisions now that would 'get the thing to print' but in the long term might not be such a great concept.

On the other hand, I'm really sure we could spit something out that gets the thing to move & spit ink(paint) at a wall just to get the incentive to continue working on it.

Whose call? vac-this is your baby.

I see the logic as this: The image needs to be processed. Top down, and left to right. each pixel will be processed for color values and sequentially sent to the gantry. The gantry will take this information, break it down to timed pulses for solenoids controlling the air/paint. (deeper color - more/longer pulses) for each color. upon completing the cycle the gantry will step one pixel and start the next color pulse. When the end of the row is reached the whole gantry drops one pixel and starts again. Preferably in the opposite direction (but this adds more to the code than I ever thought it should - but I'm rusty)

Over simplified, I know but its been working for HP / Canon / Lexmark for years why change it? What we should not do is look into vector driving at this time, I dont think it is what you guys want. I might be wrong but, for multi-color prints aka pixelation.com the vector prints will give a more grafitti look & becomes more difficult to comile the image for process. -- somebody tell me if I'm wrong here?

I am really scouring the luberth site now, because I feel there are some great things here. If we can offload the stepper control onto a PIC, it can be programmed for simple commands: x-step-right, x-step-left, y-step-down, and all the required pulsing can be done here. This would incredably simplify the PC program and just require breaking down the graphic image. This does make the entire thing more proprietary- instead of g-code we'll be sending 'vac'-code.

publitime
- just a note on the automotive fuel injector: I played with some a long time ago out of a Datsun 'Z' car ;-) I never got the signal pulsing to work and was told that by just adding power to the injector, thinking it would 'just stay on', basically ruined the injector. Also, they operate at a relatively high pressure. 50-60psi if I remember correctly.

- for a chuckle - I was trying to build a snow machine with mine ;-)

jeffterm

A injector is working at a very high pressure that's right.
But you can rebuild the injector so that he will open at a much lower pressure.
The only thing that could be a problem is that the nozzel will be clogged because he is very very small.
Maybe paint won't come out of the nozzel because he is so smal.
We have to become a dot and not a splach of paint so the nozzel must be smal.
One of the plus point with a system like a injector is that you don't need a airflow true the nozzel.
The only compressed air you need with a injecto is to put pressure on the paint tank.
Maybe there are people here arround who will have a better suggestion.
So let us know all suggestions you have.

So, when does the building start?

Eric

http://www.cnczone.com/forums/showthread.php?s=&threadid=3796

My bad...I didn't see it. Cool.

Eric