Here is a quick outline for a proposed "hardware" simulator for cnc. it is going to facilitate the software development of the raster image print drivers. i am looking for comment on whats missing. this program is going to recieve step\dir pulses from the software being written to drive the thing. with luck, that software will end up living in a microprocessor, and only a front end driver will be on the PC..

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

this link will take you to an interactive 3D model of the painter. you need the "edrawings" viewer software.

http://www.vacuumstudios.com/painter.htm <---500k so it takes a while to load. also, there are 3 parts that are unexplicably floating by one of the wheels. i would right click em and hide them.

comments appreciated... espeically criticism

Nice presentation, Vac!

Why thanks Bloy,

Heres a few more pics of current development. The gantry moves pretty nicely on the toothed belt and ball bearings. With everything clamped together for fitting, there still dosent seem to be much backlash. This is a good sign, because the distance from the ganty to the ground might limit this prototypes usefullnes as a pen plotter device... we shall see in a few days.

I found that limit switches were easy to hold on with large document clips. Nice and adjustable untill i find the right positioning.

This picture shows how i mounted the motor to a piece of large angle aluminum, and then used the motor mounting bolts sticking through to hold the ball bearings. The gnatry does a nice job of tracking this way. I have to goto school, but when i get back i look forward to the first motorized testing, as i try and figure out the best place to position the motor. Im sorta guessing with the toothed belt centered to distribute the force is going to be best.

MUAHAHA More Building

This is a side of the machine. next up is wiring and testing the assembly. i allready know it needs more metal. it is sorta flimsy. i didnt have enough acrylic to make the sides as bulk as i wanted, either. the next machine(!) will be much bigger, and probably be made of mostly this wierd steel channel i found at home depot for $13 per 10 foot piece. basically long steel C chanel. cheap

This is another shot from the side. top down. simple motor to stepper direct drive.. a next machine will have at least 4:1 reduction on the drive train. i think .02in/step is WAY to big. probably 1:6 to 1:10 reduction will be appropriate. i am hoping to find some surplus gearhead steppers, to avoid more machining on this guy.. tomorrow i will probably build a spring loaded marker holder.

The whole thing, so far. still needs lots of work.. at least i saw the gantry move tonite. even with a crappy unipolar driver at 10v and no clamping diodes got the belt drive axis going very fast. with a xylotex and 24v on it, i am sure it will be able to hit 200-250ipm ?

it was doing at least 50 to 80 tonight..

One last image. this one is from the bottom. i used 1" wide L-brackets to connect the rails to the sides. the rails are NOT very ridgid, they are made of 1x.25" aluminum bar with .5" 1/16wall aluminum C-channel riveted to it. i am going to add a considerable amount of metal to this to make it more ridgid and heavier, so its more stable. i am worried i might have to decelerate the gantry, and i am not sure if thats possible with the software available.

By decelerate the gantry you mean starting and stopping gently to avoid to much wobble?. This sequence could be part of the "move down command' and you will end up with a bit of a margin.
I got around to reading the pixation site and as I suspected they are using a very 'unsquare' resolution to give the effect of continuous tone output. They give the X resolution at 1500 dpi and Y at 5 or 10 each application of color must only cover the canvas 2/3% needing 150 overlapping pixels to get a saturated color. You will probably want to do the same thing.

240 ipm will give you 20 square feet per hour at 5dpi if you haven't worked it out. So your resolution proably is not going to be very high and if you are lucky lack of accurracy or rigidity might not be as big a problem as you think.

Nice job Vac,
is that your kitchen your using as your shop?

You are using a 10.5 inch dia wheel driven directly by a stepper motor microstepping X8 right?
Now Pi x 10.5 /1600 is .0206 inch/step, right?
When you change the speed of a stepper you are only changing the the length of the pause between the steps so whether you input G00 or G01 F1 that .0206 step will fly at the same speed, is that OK?
I'm sure it's all going to bounce around and that gantry looks rather light!

Tachus-yes - as it is now with my pulse generator running the print head, the thing shudders quite a bit when it stops or changes direction. the ganty at this point needs some more metal. probably i am going to go get some 1x1x1/8" steel square tube to replace the aluminum rails, and another piece of .5 plexi to make the sides much sturdier. that, and 4 more L brackets will make the thing 50% heavier, and like 200% stronger. shoulda waited for steel in the first place. got impatient.

cutfinger- yes -that is right. .02 roughly. Any yes, it is rateher light! it does bounce around at this point! fortunately, it is only about 60% constructed, and i dont have enough microstep drivers yet - i am going to purchase xylotexs soon. as for the Gcode speed issues, i dont know what you mean, clarify? Also, i am going to definiately gear down the steppers, im waiting to find some surplus gearhead steppers. the print head, on the other hand is fine as is, with deceleration, multistepping, and the added weight of "painthead" it will hopefully be stable enough for testing software. if not, i will run it very slow...

Teilhardo - yah, thats my kitchen! its my latest in a long line of kitchen work areas. kitchens here in chicago seem to be designed for 8 person families... so i have always put my tools and\or music gear in the kitchen.. since i moved here.. its alwas the only free space... my current kitchen is like 25x20 feet, so its perfect.

thanks for input! i will post more photos soon - tonite i am going to build a simple pen\vinyl cutter holder, and hopefully get to the hardware store to get some steel.

Steppers rotate like a second hand on a clock, the type that stops momentarily every second. That is why they are called step motors, they are not smooth. Check it out on what you have, no matter how much microstepping you apply to a direct driven 10.5inch wheel it will be run rough on the circumference.
Why use a stepper to drive the 10.5 wheel? Think about a syncronous gear motor, their speed is very accurate because they lock onto the mains frequency and they are stepless. They can be switched on and off by a relays from a command in the programming. Even a Servo motor from a direct drive record deck or VCR could be used.
Your presentation is just super but your designing skills are a few paces behind - more thought and less photography is recommended (grinning sarcastically)
You did mention that you appreciated criticism!

Cutfinger,

A printer is a start/stop type of device. The head moves precisely to a location and then lays down a drop of paint/ink/dye. There is a reason steppers are used in printers.

cutfinger -if your argument had ANY merit, i would take the criticism fine. however, you are pretty much wrong. Using an "syncronous gear motor" may mean the thing only moves in one direction. Steppers do create "steps" however, due to interpolation by the controller, lines come out nice n smooth, otherwise all those thousands of different plotter systems built from 1970 till now wouldnt work. they all use steppers.. some use steppers and servos, however, the large main roller, in my fair experience is always on a stepper. the servo is often used on the printhead carriage to make the thing move faster.

There is so much wrong with your argument - are you trying to be nasty for a reason? because your logic is faulty.

i appreciate criticism that isnt rediculous and random and negative! there is NO printer that i know of that uses an "AC Synchronous gear motor" or whatever. thats just silly.

:)

hope you appreciate criticism.. besides, if you followed the thread, you would know this is a prototype device, that gearing will likely be introduced to "up" the resolution(but not the accuracy) of the main drive, and that the painting device is a very coarse airgun. the added.. uh.. accuracy, of using a AC Synchronous gear motor would be wasted on the reproduction mechanism.

if your interested in the project, read ALL of the posts, please, before recomending something silly, while doing so with a tone of resentment\sarcasm\whatever..

thank you, drive through

Great job Vacpress.
I cannot wait to see this painter to do a Van Gough on your kitchen floor.

And cmon vacpress, it take 2 to a fight and I guess cutfinger nailed you hard with Your presentation is just super but your designing skills are a few paces behind

Not all the people can have the 'right' image of you, the image that you have of yourself. Maybe some person will feel threatened by or jelous of your work, wich is quite possible since you have great skills.

Dont take it too seriously, from this forum you are one of the most innovative members and you are showing it.
Keep up the great work.

Konstantin.

konstantin- why thank you, very much.

i dont take it seriously. its fun to be contentious. im sure he is some machineist who knows he could build a better device of type {X} than me, so he thinks "this guy is just flash and bang - look at all those useless perdy colors..." and as for my design idea, it was based on what i had in my junk pile. i actually could have built this with no CAD. i thought people here might want it. i read like 2000 archive posts here over the course of a month - the biggest help are long threads with lots of build pictures. so, i am doing my part. its almsot like free plans.

i agree, to an extent. i have seriouse trouble finishing the projects, but i have finished many... i am a very sloppy builder. thats half my desire for having a good CNC machine in the near future, much less hand work.. much less, since i can do the cad stuff easily, much easier than seting up manual cuts on a table saw or whatever.

my image of myself isnt very conceited, i dont think. i am putting all these iamges here because i think they can help people. i hope some of these 3000 lurkers are reading along, planning their own device.

i think the reason i was defensive was that this guys criticism was obviousely aimed at getting a reaction, not at helping with the design. the main reason i put the desing here is because i wanted help, which i am getting alot og. the users tachus32, publitime, and chagrin's help with software and ideas is very helpfull, and is keeping me interested.

robert

one thing to add on the free plans thing: if someone spent like 2 hours, gathering all the photos and images, and the important parts from the text, this would be the first 1/2 of a fairly complete how-to, for semi-experienced or better DIYer hobbiest types...

one exciting thing i want to do in a future machine, is try and make some large 3d structures using the layer method like SLA and FDM machines... for my medium i might use that expanding insulation foam, or hot glue sticks.. i think that it could build things like 5'x4'x5' 3d shapes that no cnc could cut. the low resoultion would be part of the appeal. adding a z axis that is like 4' would be easy because there is no tool\work piece contact, so it can be pretty light duty, and given the nature of the device even with .05 or so of slop could still work, depending on the material used. it would be fairly easy to do all this:

ultimate goal: DIY hobbiest rapid prototyping equipment.


its possible. i mean. we could do it with a glue gun and our current wood machines. just add some sort of cole-air gun or something, or just wait between applications of hot glue - make a new hot glue tip that is much smaller. this is still gonna make a faiirly limited number of shapes - to really increase the machines usefullness, add a second nozzle that spits out some material that can wash away or be broken out - a filler, so cavities are possible.

what material is better than hot glue?

Hey Vacpress,

I think what you have looks really good so far. Sheez, it's only a prototype. I think that what you want is critism of the design and to extend that to the person that designed it is getting out of line.
I had a thought. You were saying that you were getting some vibration or oscillation on stopping the movement. It seems like you are spanning a long distance between the wheels. Maybe try a truss between the two sets of wheels. One vertical and one horizontal. It is the lightest way I know of spanning a large distance.
Don't mind my crude CAD.

Chris

crude cad is wonderfull!

yes - a truss idea is good. especially for a cheap, lightweight machine.. i am using all this lame aluminum chanel i got on super-sla eat a sears hardware sale. like. 75% off their normal prices. i was going to build a enclosure box out of it, with 1/4" clear acrylic walls, for my next machine, but instead built this.

A truss, or a bunch of steel is where im headed. i like a truss cuz i have the parts allready - just more lame aluminum channel. the thing will weight alot more with a computer, a power supply, etc. on it...
oh - one problem with the horixontal truss is that i wanted to drop my tools down through the opening - the gap between the spars... i can just make it like 2 bridges next to each other though. i like that.

Originally posted by vacpress
what material is better than hot glue?

Would silicon sealant do any better? No heat required. Couple a stepper to the plunger and you have some crude control over the amount deposited.

Problem: how to prevent the sealant in nozzle from hardening during non-injecting moves.

P/S; Keep the pictures coming on your prototype. It's a source of inspiration :banana:

Hey Vac,
Your artistic skills are amazing. Did I mention before that you are the fastest "engineer"/"designer" I have ever seen on my adventures on the web?
hey cutfinger,
come on pal, a suggestion would have been just fine, why take away from his skills. We're all learning here, we're all the students in the classroom witnessing show and tell. Let him tell the story first, give SUGGESTIONS, not critisisms after he's educated the rest of us...
Peace:)
Tei

abasir-if i remember corrct, silicon sealant takes 12-24 hours to dry? and the opening of the nozzle for such a device would be very small. this thing has to basically make "3d pixels"


i will keep the pics comming, and video when its moving. hopefully the second version will be a usefull tool. i want it to be able to cut vinyl signs, make very large pen-plots in multiple colors, and maybe(just maybe) do raster graphics in paint or ink..thats so hard though.

a third version migh be a rediculous large-scale "3d printer"

Hey vac,
a technical question...
Have you figured out how your going to make a nozzle that makes all the right color combinations?
Are you going to have to design your own software?
-Tei

tei,

most of the info is at: http://www.cnczone.com/forums/showthread.php?s=&threadid=3714&perpage=15&pagenumber=3

there are several posts with summaries about what will be used:

briefly i am going to build a 4 colro print head eventually, at first i will probably trry and do black and white... grey... i need to work on getting a VERY small, reproducable dot of paint.. that may mean lots of eperimenttion.

the software will probably use a mix of opensource image rasterization\vectorization software, perl scripting, and perl or other stepper control software.

the HPGL software is standard cnc plotter SW. currently, i like the lookc of Cstep, although my first use of it left me with questions. it seems unfinished, which luberth repeatadly mentions on his site...

Originally posted by vacpress
silicon sealant takes 12-24 hours to dry?

There are some that are fast curing (minutes). In addition, you'll be placing minute amount ('pixel') so they should hardens faster (?).

BTW, what sort of 'pixel' size are we talking here?

abasir- for a first device? very large? the main part would be getting a reliable mix of hard material, filler, and ways to dispense them, then it could be made better.

making it better would involve decreasing the size of "3d pixels" and improving the machines accuracy. also, the software to take a 3d model and slice it into layers needs to be developed. however, this software cant be that hard to write - in fact, with java, i think it may be possible to do it in a few long programming\research sessions.

in a week or so when this machine works, i will try it with this device and some insulation-canister foam. i just want to make a sphere, layer by layer. i will have to find out about how long it takes to harden.. what we need is something like a superglue foam material.. for a filler, starch and water? what hardens very fast, yet is water soluable? im sure something about "starch" is promissing. maybe a tiny little heating element that dries or cures each "3d pixel" could be made... with nichrome.. micro heating elements are def. possible.

what is the correct term for the "3d pixels" im sure there is a terminology.

"3d pixel" is voxel; the three dimensional equivalent of a pixel. A pixel is a "picture element," and a voxel is a "volume element." A voxel may also be defined as the minimum volume that a rapid prototyping system can fabricate.

More glossary at
http://home.att.net/~castleisland/glos.htm

abasir - great!

voxel is the coolest word i have learned all day! creating voxels sounds magical.. i think i may go create some right now(heads for bathroom..) wait. just kidding. har.


any desire out there to create a redamentary diy 3d printer? maybe just go with abs plastic beads, or hot glue, and either no filler, and a long cooling cycle, or some sort of meltable, yet water soluble filler... like... i know there is that wierd molding material... moulage? is moulage an option?

Hey Vacpress, this is the most funniest project I've seen. And a main reason to kill the neighbours golf equipment ("I really neeeed those wheels..."). ;)
How are you going to control the colour output? I think I havent read that or maybe I just missed it. The precision is not really an issue because the paint will spray out and not dot-a-drop as an inkjet printer, so how about making for a example a 4 holes nossle with black and RGB output?
I believe the mixing of the liquids will be a problem, mixing colour and adding colour as layers are different things, so or are you going to mix or spray one colour at a time?
Your project is somehow similar to a project me and a photographer collegue spawned after a couple of friday beers once (a couple of uge ones), we came up with the idea of ripping a plotter in pieces and make it 10 times bigger and replace the pen with a spraygun nossle to paint on cotton fabric, but it wasn't more than an idea...

Keep it up!
Sven

Sorry,
I did not pay too much attention, I was thinking it was to be a little like a line marking machine used on highways.
If a large printer is the idea, why not build it like a printer? It would be a lot easier to draw the fabric or paper under the the printhead than drag a very fragile looking contraption over the material.
There must be reason for using 10.5 inch stroller wheels with plastic bearing rather than something smaller with more precision.

cutfinger- yes - the reason is that i have them. wheels get expensive quick.. $60 to $100 is reasonable to expect for decent wheels.. this device is 99% my junk box. :)

so, its got lots of flaws..

and besides, its way more fun to make robots than printers.

It would be a lot easier to draw the fabric or paper under the the printhead than drag a very fragile looking contraption over the material.


And besides that it will be kind of tough to get the mini train landscape into the printer... ;)

Cheers,
Sven

Sven
Printers use Cyan, Yellow, Magenta and optionally black and the color you see is whats left after white light has passed through the ink reflected off the white paper and through the ink again. The ink absorbs some colors but lets others pass. It's the light passing though the ink rather than reflecting off the surface it which gives the color, If you want proof of this try printing on black paper!. So perhaps counterintuitively its doesn't matter which order the colors are applied or if they are mixed - good thing too or printers would have to be a whole lot more complicated.

Vac, I gather that when you were testing you just connected up you pulse generator and went form zero to full speed. What was the frequency of you shudder? At the gantry resonant frequency or the Belt / Printhead one?

tachus - the printhead runs very smooth when i ramp it up, and miss the one area of resonant frequencies. if i stop it suddenly, however, it shudders, alot. but, it isnt done. by the end of the weekend, it should be much, much more stable. all i really need now is to do some smallish stuff, add some steel, mount some ballbearings on the wheel drive axels,.

i havent seen the gantry axis move yet. the axels are not done. another sad thing is that it wont quite rotate on its own axis, i dont think... it binds, but that is just what i observed from pushing the thing around... oh well.. i bet it will workout fine...

Printers use Cyan, Yellow, Magenta and optionally black and the color you see is...

Well, that is very up to your printer manufacturer. My HP printer uses CMYK (new Hp's uses 8 colours), my Epson runs on eight different pigments (CMY, red, blue, gloss pigment and two blacks) and my old HP runs on RGB. And for sure, the plotter I use has RGB and black.
This is pretty off topic, but putting ink into a paper as a printer or adding a layer of paint with a spraygun is not the same thing. First of all a printer is putting very small drops of colour into the paper. Small enough to not interfer with its neighbouring colour drop and it's hot enough to dry so fast that it doesn't mix with next drop out.
The spraypaint layer is covering the surface and if it's thin it will let the light be filtered in the same way as the inked paper. Maybe an extrem example, but adding a blue layer on top of a silver surface will not give te same result as mixing silver and blue as a single layer, right?.. ;)
I agree that printing on a black paper isn't the best way to get good colours. But I spraypaint my almost black motorcycle parts in any colour, and you cannot tell were I put the yellow scratch filler.

The colour problem, if a green area is supposed to meet a red area and the colours are sprayed at the same time and the spraygun doesn't change colour fast enough there will be a zone between the areas that will have a third colour, brownish. If the paint layer is thick enough and the colours can dry between the layups the mixed zone should be smaller, as long as the gantry is spraying and not "laying" or dotting the paint out there will be a zone but hopefully smaller and the colours should fade into each other instead of making a blended third not-so-wanted alternative. That is really happening when a printer cartridge is worn out.

Sorry Vacpress for the off topic, back to your gantry. :rainfro:

Cheers,
Sven

Yes I guess the trick is how thick you put the ink on.
I'm taking a guess but I imagine most light colored paint has a white pigment in it (titianium dioxide?). The trick is probably to get the paint shop to mix you some process color paint without any white pigment in it, then it should be nice and transparent.

Vac you might want to allow for the posibility of a spare rail to run a counter weight on in your next machine. You would have a wire running from each end of the print head around a pulley to the counter weight. I have seen some quite small motors stop and start big heavy print heads suprisingly fast inducing wobbles in big heavy printers which is not a problem if your media is attached to frame but if it is not... .

As for the gantry axis not moving can't you just print on a downhill slope or something?

Idea for a print head.
Have you seen those pumps they use for blood, food, corrosive chemicals where rotating wheels squeaze the liquid though a flexable tube?
If you made one powered by a stepper each time you moved the stepper you would get a small amount of ink disspenced to the end of the tube where a air jet can blow it onto the page. The tube could be removed and replaced or cleaned after each run. If you used a pump with a circumferance of 5cm and a tube with a internal diameter of 3mm each 1.8 degree step would pump out about 5 microlitres.

I Realy dont have the right to talk here but may i sugest you look into the print heads that bakery shops use.

I'm not saying USE one just look at how there doing it.
The proses can be moded to your use verry easly.

Are you talking about air brushes used for cake decoration? Or some sort of pump to dispence dough or icing?
The idea of the Peristaltic Pump (found out the correct name) was as a alternative to the wire in the wire jet. The wire probably can meter out smaller amounts of paint but the pump should be easy to build, is pontentially less messy and clog free. - I'm just throwing it out as a Idea.

5 microlitres times a paint coverage of 7.5m^2 per litre (300 square feet per galon) is 37.5 mm^2 or a dot about 7mm diameter. About 3 or 4 dpi thats, a little low but it's damm close to what is required gearing down the stepper, using smaller tube or diluting the paint or similar techniques would make it feasible.

One possible advantage would be you might be able use the same jet for all four colors with 4 tubes going into one venturi.

The cartridges used for printing on cakes is the same as any other inkjet ink cartridge. The manuafcturers use clean cartridges and fill them with extremely fine food colorings. Even though the food colorings are very fine the cartridges are still prone to clogging.

I wonder if the Peristaltic Pump can be used for pumping precise amounts of air (like the Wirejet) instead of using it to pump the ink.

i just got home with 8' of 1x1x1/16 CRS steel square tube... hopefully this adds the stability this thing needs.

sven- those comments are totally on topic. in another thread, tachus & I discussed a similar idea- that we are not really mixing ink or paint, we are using halftone patterns and "pixels", i think, to create the color density, gradation, etc. the trick is going to be "tuning" the thing, once we get a basic software figured out. if the software knows how to at least do an off\on for each pixel, or pixel area, we can figure a way to make the colors right... i hope.. and if not - then it will jsut be a big quirky plotter.

tachus- the venturi and peristaltic pump is a good idea- and one i thought of also, but i am sorta leaning towards some sort of blade\wire\hoop\disc\etc that is run in a circle through the trough of paint - much like pixation.. i bet "wire metering" or whatever it would be called is used for other uses than the wirejet.

sdfine- i dont think the peristaltic pump can pump precise amounts of air, i dont know why though :) heres a link: http://www.animatedsoftware.com/pumpglos/peristal.htm
one good posibble thing about the peristaltic, it looks like it could move pretty viscous fluids...

cake painting machine are an interesting thing to bring up - it reminds me of the industrialized inkjet heads i was looking at before i really decided to build this thing. there are problems with inkjets though that the pixation machine overcomes... the inkjet heads seem to need a very accurate delivery

i am gonna spend 5 or 6 hours on the machine today. ill post some pics!

Good luck with the steel Vac
At your rate it looks like you might be able to finish the mechanics of it today. Then onto designing the print head...

teil- yeah. the printhead in my 3d drawings is not a final design at all.. for 1, i only have 1 paint gun..

i think i want to spend a day with the thing doing vector stuff using standard CNC software and a Pen or marker. I can get the thing "dialed-in" that way.

i need new wheels though... my wheels are too light, and too large. i need 6", hard rubber wheels, with heavy metal hubs...

vacpress




I think i will be simple to convert a photo to G CODE.
I tryes something last night.
The program i used is simular with a program for laserwork.
All you have to do is import a photo into the program.
The program will translate the photo to g code.
Thene the laserbeam will burn the photo into wood are an other material.
With the laser you will get a black and white picture .
So to make a black and white picture it will be easy.
I have just to put a spraygun in place of the laser and ready.
To do the same with multiple colors it can not be that hard.

re Peristaltic pump - i see great minds think alike ha ha
The problem comes down to how to meter out very small amounts of a viscous fluid continuously with a large variation in flow rate without sticking it though any sort of small hole - there aren't two many options other than the 'fluid film' option which is probably the best i can't really think of any others apart from the pump.
I didn't read the pixelation patents but i did notice on one of the pictures of the wire jet the was a area to the side of the print with bands of paint where "the excess paint is cleaned off" does this mean the wire is rewound? air pressure is increased and the wire cycled to remove "cling ons" or perhaps another air jet to clean paint off that is dripping down off something?

Getting the color output right was the reason that i thought ghostscript was a good idea, it has all the mechanisims built in. If you get paints that are close to cmyk the biggest issue will be getting the transfer curve for each color right but there are some simple ways to do this with special test prints.

tachus
right the wire on the wirejet must be rewound after each printed line.
But in one of the pattents you can olso see a construction with a closed wire.
That wire is weld with a laser.
In this construction you don`t have to rewound the wire.
For remoeving paint they are useing scrappers.
Isn`t it possible to RIP the picture with a RIP software like the one we use on a large format printer and thene translate that file to G CODE.
Good RIP`s are flexisign and postershop .

Ghostscript is a RIP and free! It's every bit a good as the commercial ones, in fact many commercial RIPs are licensed versions of GhostScript.

Rewinding the wire after each line Ouch. That makes things more complicated than they need to be. Now you have 2 spools times 4 that need driving. Can anybody think of a source for premade wire loops without noticeable joins?

While thinking about cake decorating... for future discusssion...

How about an overhanging gantry machine that decorates cakes, a cake plotter. Vacpress, You were talking about 3D designs. Here's the perfect medium, cake icing. It wouldn't be photo quality but could lay down decorations and line-art.

Publitime,

Maybe I'm missing something but GCode doesn't seem necessary. The problem is not the path of the head. It's how to control each pixel. The path of the head would be the same for every print job.

As far as a RIP is concerned, my problem is understanding what's the output of the RIP and how to use that information to control the machine. Does anyone have any idea what the output of a RIP is and how to use the information? If I knew how to access the output of the RIP I would be able to write a control program to interface with it.

Sdfine

Has anyone looked at Pixation's simpler BladeJet technology. I know it uses a fine jeweler's saw blade to transfer the paint in front of the air jet. In one patent he references a Paasche airbrush that moves a reciprocating needle from the paint to the air jet. Does anyone have any more info on this?

i havent really read all the pixation patents, beause i find them unplesant to look at.. i think they just run their wire in a loop, past some scrapers. the scrapers make sure the paint is even on the wire. the bladejet must refer to the technology that allows the machine to paint on the ground, as opposed to the wirejet on the wall..

i get the impression that a blob of paint probably forms on the wiper, which would help to keep the wire evenly coated. i also think that the large spring-arrangement they use to tension their wire could be removed if a simple "hoop" was used" like 1 2" metal hoop. this would essentially be the same thing - there could be some small wipers that keep it from getting messy, and it would only need 1 small stepper and simple, cheap driver.

airbrush parts could prove handy. i have seen cheap asian nozzles going for $2 each.. so a small manifold to hld 4 airbrush nozzles and adjust their angles positioned behind a very thin metal loop that runs through the paint could work. the nozzles just need to be adjusted to make the most desireable dot of paint.

sdfine- if you goto the ghostsript page, you will find that it can utput lots of files. i think there is a "sun binary format" or something like that. tachus mentioned it earlier in the thread. the sun format is just an x,y location, and a color value.

Hi, I am insterested in this thread some time ago. I think the first time I see the pixation machines was in an subject in yahoo group
DYI CNC.

I have tested some things.
1) Ghoscript aproach. I think is the best so is the most compliacted
you need a bunch of noozles in order to get C M Y K, similar to inkjet printers.
Commercial Noozles are 0.3 mm in diameter, in inkjet printers we are speaking in micrometers. This is frist problem. With 0.3 mm noozles dithering dont work.

2) There is an aproach more economical.
In Adobe Photoshop 8.0 you can convert a true color image to 8 color without loss a lot of quality. In function of image with six or 4 color you get good results. For example for cake boxes I think 6 color is enought.
PHOTOSHOP is aplicating dithering algorithms in order to get magically the conversion.
Next step is save the new image to open format png 8 bits or 256 colors.
Open png and save as bmp 256.
Now you have a bitmap with 6 or 8 or 16 colors what ever you want until 256, and ecah byte is a color pixel.

Example
Suppose you want paint a wall of 1mx1m with 6 colors you must save a bmp with 1000mm of widht for 2000 mm of long with an dpi of 25,4mm/0,5 mm diameter of nozzle

You can execute gcode in order to print 1000mmx2000mm of color1. Now you clean noozle and tube and put color 2, and execute the gcode again.
You can use gcode tool_change in order to clean nozzle, tube load the new paint and intro.
Really you are painting lines of 0,5mm of thickness in color 1,2,3...

I think this approach work, for example in order to paint walls. Have you tried convert textures files in 2 or 3 colors, results are very good.

I hope this help

Just to throw an idea out there...

Have you guys ever seen the Blow Pens kids play with. Blow air in one end and a mist of dye comes out the other end. Kinda like a human powered air brush. How about a system along those lines. Maybe you could use an system of felt tip markers and insert them into a stream of air for a specified time depending on density of color needed. Just thinking out loud.

ajv2803959

Seems like this will work but can you give some more information by each step.
take also a look at http://www.imagetogcode.com/
Maybe this will be intressting

Blowing air across the tip of a felt tip pen seems to draw the ink out and spray it onto a target. Of course I was using 130PSI so the pattern is a little on the large side. Just a thought.

pehaps a sharpie could be used with alarger resevoir of ink above it. the air could come from a very tiny nozzle.. maybe it could be shot through a template to make a smaller dot? thats an idea.

my roommate sugested that just for plotting black vector art. an interesting idea. would have to do some tests.

i gotta get some nozzles now...

I think you missed my explanation of the BladeJet. It uses a very thin jeweler's saw blade to move the paint from the reservoir to the air stream. It sounds like a simpler print head design than the BladeJet.

If you're going to use a sharpie you may as well just use an airbrush. The reason the WireJet doesn't use an airbrush is that airbrush inks are very expensive when used on a large scale. If you're going to use paint you have to create a print head that doesn't push paint through a hole. The benefits of paint is that it is low priced, waterproof, durable, and readily available in process colors. Pixation recommend Sherwin Williams tinting colors which are available at any Sherwin Williams store. They tell me they are equivalent to process colors.

sdfine- yeah and their page says to thin 20% with anti-freeze. so thats easy enough, and what i am gonna do. we need to come up with a source for nozzles, metal loops, etc. for the painthead...


i am currently rebuilding the gantry and finishing the moving carriage. i will post pics later tonite. its going well. much stronger with the steel.

The reason for adding the antifreeze is not for thinning but rather to give the paint a longer wet time so the different paint colors blend together better. I spent a 1/2 hour with the developer of these machines last week and got the full explanation of how they work and what materials he uses. I suggest you experiment with the jeweler's blade (as in the BladeJet). It's a simpler design and the blades are easy to find.

Publitime's suggestion to use imagetogcode is a good one to get started printing a grayscale image. I sounds like the software creates a gcode file where the Z axis is moved up and down to represent dasrker and lighter colors. It would be easy to adjust the amount of paint based on the change in the Z axis.

sdfine- you went and met... glen was it? greg? anyways - what a machine he has built eh?!

you saw it work? or you at least saw it? did you mention to him that some dweebs on a message board are trying to DIY his idea?

again, let me say, i would not commercialize anything that is someone elses patented idea! that would be wrong, as well as legally stupid. im pretty sure there are laws protecting reverse engineering for r&d though...

vacpress and I have had discussions on this before in which I've emphasized my distaste for the complexity of the wirejet design as pertains to the design of the electronics and software. Simultaneously controlling a stepper for the Y axis and four for the each of the CMYK colors is a lot of busywork for even keenly optimized software, much less anything a hobbiest would create -- there are a lot of timing issues in keeping four wires moving at an even pace. It's also a six axis design requiring 12 I/O pins and nonstandard parallel port programming (as far as my novice mind understands it).

Two alternative designs come to mind to me, both which require only 4 I/O pins (in addition to the 4 for the X and Y axis):

Pulsed sprayguns would be simplest; multiple pulses of the gun would create a darker color. A range of zero to eight pulses of each color should be sufficient for a full color scale; these could be produced as the paint head continues to step across the area of the media that constitutes the pixel. Probable implementations would suggest that many more steps-per-pixel (and thus spray pulses) would be available for better tuning of color.
Using a wirejet design, drive the wires using standalone motor speed controller(s?) connected to cheap hobby DC motors and pulse the air that hits the wires. Solenoid air valves are quite common on surplus sites. Tuning the color balance could be done either by accelerating or decelerating the individual wire motors or via parameter changes in the software by increasing the pulse ratios of the colors.


I'd love to hear comments on these ideas.

Regarding patents, the rules have been greatly strengthened in recent years. Nowadays a patent holder not only controls who can create devices that implement the patent but also how devices the incorporate the patent are used by endusers. Yes, if you buy a wire jet you could still be restricted as to what purposes you use it for (if it was so desired by the patent holder). A few months ago there was a great hubbub about the licensing of the patented "Stots Jig" that you can dig up with Google that discusses these issues. Interesting reading, but we should probably leave discussions of that crap to the lawyers, not CNCZone.

chagrin. good suggestions..

- pulsed sprayguns is interesting, but the size of a dot any spraygun will make is pretty big. too big. i am basically abandoning this idea. i think i may implement a 1-spraygun setup to do blackand white. that being powered by raster 2 gcode software will work nicely. i wonder if pulsing should be implemented via a seperate pic micro to recieve the Z commands and pulse according to depth? seems complex...

-i have another idea - how about just 1 stepper moves all the wires, and only the color needed shoots? another option would be 1 continuousely moving steppers, and 4 clutches. clutches are plentifull in garbaged printers. i have a BX35(www.basicx.com) that i would be happy to use to control this thing. it has like 25 IO pins, 2 Serial IO lines. Its pretty amazing, and very simple to use. It is 65khz and runs an interpreted BASIC. It has many features, and is easy to program.

a note to all my wonderfull helpers and critics on what i have available:

things i have available:
-4 good new solenoid valves
-Many solenoids of different sizes. some quite large
-A large array of steppers from very small to pretty large
-4 small/medium DC motors with encoder wheels. from inkjet printers. at least 250-CPR
-Many DC RC car size motors. from RC cars and from Inkjet printers. same size motors.
-4 Medium sized 24v, 1.5amp servo motors with decent looking encoders and pressfited toothed belt cogs. tese are from large old okidata inkjet printers.
-Many sizes of gears and cogs.

Thanks for the help!:rainfro:

Vacpress,

Sorry for the misunderstanding. No I spent 1/2 hour on the phone with him (Dean). I called him to ask a few questions and we ended up speaking for quite a while. BTW... I read all the patents listed on the BladeJet section of the pixation web site and none of them were specific to the BladeJet (they covered the WireJet). Intrestingly enough I came across a 2002 patent for a generic gantry printer to print images. I couldn't believe that someone could get a patent since these have been on the market for over 10 years.

patens have always seemed 1 part intellectual property protection to 1 part idea-theft legitamiztion and semi-legalized monopolization of commonly used ideas.

any idea where to find cheap nozzles? im gonna check harbor-freights replacements.. i think they are like $2-3 though.. not quite cheap, but cheap enough.. just gotta wait to get some $$$ in.

http://www.mcmaster.com/catalog/110/gfx/large/g1bga3l.gif
I Found this image on mcmaster carrs website. it describes some spray patterns. any idea what would be best to blow to paint off a thin metal hoop or wire?

i have some very small. like 1/64 drill bits. could i make my own nozzles? i may try. i may make a plate with 4 1/64 holes all drilled at slightly different angles, and hook it to the aircompressor and see what they do to a thin coat of paint on a medium guitar string..

the mcmaster carr nozzles(pages1898-2000) are pretty costly.. but maybe worth it.. i am gonna check the hardwarestore for nozzles to misuse.

I have tried to use MCmasters nozzles but most of the good ones require pretty high pressure from what I remember

hmm. i have a 110psi air compressor...i dunno aobut those mcmaster nozzles. i think i want to make my own if its feasable. i may try tomorrow.

The new metal is much nicer in all ways, so is my construction.

here is a closeup of the carriage. i have added a db25 breakout board and connector cable. i hope this is enough lines! if not, i can add more easily. another 15 is possible, using a db15 connector. i may consider putting the driver boards on the carriage to avoid long cable issues.

I think this assembly is comming out nicely.

The new gantry looks good vac.
Todays wacky ideas:
You might be able to make a nice loop buy etching some metal shim like a circuit board. A wheel with a hundred or so fine spokes might hold promise.

Maybee you could mask off the spray gun so only a small fraction of its patten escapes and recycle the rest back into the supply.
Early ink jet printers had the ink jet running continuously and deflected the jet electrostaticly to stop it hitting the paper. I'm a movable mask perhaps rotating could serve the same function.
Is there anything acrylic paint doesn't "wet" that you coat a belt or wheel with leaving small dots where the paint sticks.
A chain of gears meshing together might transfer a constant small amout of ink to the far gear, the amount of paint transfered would depend on the film strength and the pressure at the mating surfaces.

To do the pulsing with a gcode setup up i would use coolent/spindell on/off commands.

Take a look at this guys 1 pin dot matrix printer.
http://www.sentex.net/~mwandel/tech/printer.html

The guy is a genius. Look at his other hacks from the mid 80's.

Eric

Another option, this is how the "pros" do it, maybe it can be simplified? see link below, also What about a stainless steel needle from a syringe as a dispensing tip?.....food for thought........

http://equipment.loctite.com/standardequipmentdetail.cfm?EquipmentItem=98084

tachus - i really like the etching idea- though i was picturing something that is held in some sort of device that makes the connection from stepper to hoop very direct.. like.. a rubber wheel... and the hoop could be held captive in small bearing blocks..

i will do some 3D models of this soon...

i need to figure out where to get metal hoops and wire loops... soldering would make a nasty little spot..

So you are planning on a rigid loop? Will probably make things simpler. Methinks i need to wait for the models.

I have a gantry and even the source code for the driver software (gimp-print) for a flatbed printer but can't find a printhead with a reasonable cost. What printhead are you going to use?

kdoney- homebrew.


basically its gonna be like the www.pixation.com machine. 4 metal hoops , driven through a paint resevoir, and blown off onto the work surface with small, focused air jets.

gimp-print - linux opensource? works with the gimp graphics editor? i have seen reference to linux opensource inkjet drivers called something like that?

what are the details of your setup? are you going to put the gantry on wheels, or is it a standard gantry-style cnc machine?


for more info on my(our) project, read the various threads about it. there is LOTS of info here now.. enough to start anyone in the right direction - which was a big goal.

current progress: its close to running its first gcode - just gotta get together some driver boards, and gear the main drive wheels down at least 4:1.

Here is an overhead view of my work area - on our kitchen counter.

the machine drew a fairly perfect rectangle being driven by 1 unipolar driver board and a function generator. this image is of the current state of the printhead carriage. i have attatched a small, 1.25" Z axis that has a spring loaded sharpie marker holder in it. the z axis is a directly driven rack and pinion system made of parts from an old okidata dot matrix, and a rack gear from a HP inkjet.

Some more nice work Vac! How is software research coming?

vacpress
Been following your progress for some tome, good job.
It's really good to have such confidence in your work to start it up and run on your kitchen counter. If I did that I gurantee that the rectangle would have come ont in the x-y direction instead of the Y-X direction and it would have wound up on the floor and my wife saying "I told you so"
Good luck and keep thoes good pictures comming.
Hager

vacpress

is there any driver jet.
I looked everywere but found nothing jet.
I have talk to someone and maybe he can write a image to G code convertor.
Keep looking for the driver.
The driverbourds on your pictures abouve are the selfmade?
By the way you work looks great.

Here is the perl script I posted earlier setup for black and white with a pen which goes up and down.

#! /usr/bin/perl
{
# Put the GCode Snippets in Here

my $SetUpCommand="";
my $MoveXRightCommand="G91 X1\n";
my $MoveXLeftCommand="G91 X-1\n";
my $MoveYCommand="G91 Y-1\n";
my $CyanOnCommand="";
my $CyanOffCommand="";
my $MagentaOnCommand="";
my $MagentaOffCommand="";
my $YellowOnCommand="";
my $YellowOffCommand="";
my $BlackOnCommand="G90 Z-1\n";
my $BlackOffCommand="G90 Z1\n";

my $buffer;

my $XPixels;
my $YPixels;
my @ScanLine;
my $XPosition;
my $YPosition;
my $ImageLength;
my $ImageDepth;
my $RasType;
my $MapType;
my $MapLength;

my ($InputFile, $OutputFile) = @ARGV;

open(INF,"<$InputFile") or die "\nCan't open input for reading: $!\n";
open(OUTF,">$OutputFile") or die "\nCan't open output file for writing: $!\n";
binmode INF;

read(INF, $buffer, 4); # read first 4 bytes and check for majic number 0x59a66a95 = 1504078485

if (1504078485!=unpack(N,$buffer)){die "Input does not look like Sun Raster\n";}
read(INF, $buffer, 4); # get width
$XPixels=unpack(N,$buffer);
read(INF, $buffer, 4); # get height
$YPixels=unpack(N,$buffer);
read(INF, $buffer, 4); # get depth
$ImageDepth=unpack(N,$buffer);
read(INF, $buffer, 4); # get Length
$ImageLength=unpack(N,$buffer);
read(INF, $buffer, 4); # get RasType
$RasType=unpack(N,$buffer);
read(INF, $buffer, 4); # get MapType
$MapType=unpack(N,$buffer);
read(INF, $buffer, 4); # get MapLength
$MapLength=unpack(N,$buffer);
read(INF, $buffer, $MapLength); #throw away Map

print "Image is $XPixels wide and $YPixels tall\n";
print "$ImageLength Bytes Long, $ImageDepth bits \n";
print "Rastype: $RasType, MapType: $MapType, MapLength: $MapLength \n";


print(OUTF $SetUpCommand);

while($YPosition<$YPixels)
{
print "Processing Line $YPosition \r";
read(INF,$buffer,$XPixels); #get one line
@ScanLine=unpack("c$XPixels",$buffer);
if($YPosition%2){@ScanLine = reverse @ScanLine}; # if Y is odd reverse to print right to left
# don't need the following line if we are dragging the tool
# if($YPosition%2){print(OUTF $MoveXLeftCommand);}else{print(OUTF $MoveXRightCommand)};
for(@ScanLine)
{
if($_ & 1){print(OUTF $BlackOnCommand);}else{print(OUTF $BlackOffCommand);}
if($_ & 2){print(OUTF $CyanOnCommand);}else{print(OUTF $CyanOffCommand);}
if($_ & 4){print(OUTF $MagentaOnCommand);}else{print(OUTF $MagentaOffCommand);}
if($_ & 8){print(OUTF $YellowOnCommand);}else{print(OUTF $YellowOffCommand);}
if($YPosition%2){print(OUTF $MoveXLeftCommand);}else{print(OUTF $MoveXRightCommand)};
}
print(OUTF $MoveYCommand);
$YPosition++;
}
print(OUTF $BlackOffCommand);
print(OUTF $CyanOffCommand);
print(OUTF $MagentaOffCommand);
print(OUTF $YellowOffCommand);
close OUTF or die "Can't close $OutputFile: $!\n";
close INF or die "Can't close $InputFile: $!\n";
}

Make a definition file for the "Output Device" called 10x10.upp and put in GhostScripts lib directory

-supModel="SUN rasterfile, 32 Bit, 6+1 Colors (CMYK-Error-Diffusion)"
-sDEVICE=uniprint
-dNOPAUSE
-dSAFER
-dupColorModel=/DeviceCMYK
-dupRendering=/ErrorDiffusion
-dupOutputFormat=/SunRaster
-r10x10
-dupMargins="{1 1 1 1}"

The -r10x10 sets the resolution at 10 dpi although this gets scaled by the gcode later. And the margins are set as small as possible.

Run GhostScript to make a raster file using one of the sample files which comes with Ghostscript (Golfer.ps)

ghostscript @10x10.upp -sOUTPUTFILE=golfer10x10.ras -sPAPERSIZE=a0 -dEPSFitPage golfer.ps -c quit

the sunraster file converted to jpeg below

run the raster file though the script
makegcode golfer10x10.ras golfer.nc

checking the output in a toolpath viewer it's not quite right

looking closer the problem is dragging the tool produces a sausage shaped dot which overlaps a white dot between to pixels.

Modifying the script so that the tool just moves up and down without being dragged changing the lines

my $BlackOnCommand="G90 Z-1\nG90 Z1\n";
my $BlackOffCommand="";

and uncommenting the following line which corrects for th pixel and the end of the line

if($YPosition%2){print(OUTF $MoveXLeftCommand);}else{print(OUTF $MoveXRightCommand)};

The image now looks much better

I can post some extra detail here on how to get Ghostscipt and Perl installed so this works if anybody is interested. There are probably easier ways of doing this for a monochome image but this method will also easily work for color and different marking devices.

ahh - totally sweet

i am ordering the xylotex boards i need tomorrow - 3 of em, and getting my 1 repaired.. i am also going to finish the major stuff left to do to the moving carriage - mostly adding 4 more ballbearings to hold it to the rail, as it can lift up as is - i should have put the rails in a diamond configuration like a shopbot - the next bigger one will be that way.


so - is this script basically the image to gcode concept?

i like how it so raw, i could probably build a real looking and behaving print driver around it. i think i can make a windows program run the scripts in a shell? i have done that before with VB4.0 when i was 15. i was using batch files though...

tachus - can you post a short narrative on your code, how it works, etc?

thanks alot! this thing is getting there....

Good job Vac. I am wondering why the gantry is so high.
Couldnt it be very low to the ground?
That way you have more stability.
I guess It looks better been able to see the spray nozzle move though.

When you posted a poll about what you should build next, not many people voted for this, I am glad you went against the trend and made this. I was hoping for this to be made.

This will be a great project for others to build once its finished.

Vac this is the Postscript -> Image -> Gcode concept allthough it's easily modified to produce some new language invented a on board microcontroller. GhostScript should be able to be set up so that clicking print makes the raster file and then you would need to run the script. This could all be automated with a little extra effort.

The upp file sort of the printer driver for ghostscript here. It can have some other stuff in it like gamma correction if you get that far.

The top part of the script contains the gcode for each step like move right one pixel and you can paste in anything you like here. Next the script opens the input and output files. It reads the small header on the file and checks it is the right kind of file, extracts the X and Y pixel dimensions and prints them out.

The rest of the raster file consists of binary bytes where each byte represents one pixel. Of the 8 bits in each byte four of them indicate wether the corresponding CMYK pixel is on or off. So the script reads the file in grabbing one line line at a time. It grabs a entire line so it can reverse the line to print right to left on every other pass. Reading the Bytes from the buffer one at a time it checks each bit and puts either a pixel on or off command in the output file, then a X move command (either left or right depending on wether it's a odd or even number scan line). At the end of the line it inserts a Y down command gets a new line and so on until it has run out of lines.

The script isn't very clever and will for instance insert a pixel off command even if the pixel was allready off, but this doesn't actually matter and i didn't see any reason to make the script more complicated than it needs to be.

I have been doing a bit more experimenting. I did a few calculations and came up the transfer curve for a round dragged tool and put it into the upp file:

-supModel="SUN rasterfile, 32 Bit, 6+1 Colors (CMYK-Error-Diffusion)"
-sDEVICE=uniprint
-dNOPAUSE
-dSAFER
-dupColorModel=/DeviceCMYK
-dupRendering=/ErrorDiffusion
-dupOutputFormat=/SunRaster
-r10x10
-dupMargins="{1 1 1 1}"
-dupBlackTransfer="{
0.000 0.060 0.120 0.180 0.230 0.280 0.340 0.390 0.450 0.500 1.00
}"

Now the image is correct for the dragged tool and should recreate gray scales accurately.

tachus42

You did a very good job i think.
I want do some experiments with wath you did.
I never have worked with the application you use si i now nothing about theme.
Can you give me more info about how i have to work .
Are there also free downloads ( trials) for the programs?


jurgen

All the programs are free.
Get GhostScript here: http://sourceforge.net/projects/ghostscript/ (get the 8.14 version) and get Perl from: http://www.activestate.com/Products/ActivePerl/
You can cut and paste the perl script into a text editor and save it as something like makegcode.pl. In windows the ghostscript excutable is called gswin32 so where i have put ghostscript in my examples you will need c:\gs\gs8.14\bin\gswin32

Has anybody given thought to making the printhead "fly" constantly in x,y,z to print on 3d objects? The fellows at http://www.stm-usa.com/
are milling out topography and then using a cnc machine to print on top of it. How would you combine the color data with the 3d data? Just curious if any of the talent in this forum has wondered about this.

I got a bit carried away with the "proof of concept" script and added a few features. It now makes a attempt at the shortest tool path and automatically selects the right resolution for the tool size and required output size. Can be set up so clicking print from any application makes a gcode file automatically.
I uploaded the file to the downloads section with some instructions on how to set it up.

Anybody want to give it try and suggest improvements?

I'll try and attach the files to this post

Tachus - wonderfull! You are truly a great help!

now i CANNOT stop work on this. inspiration.

great install documentation, by the way. so, the next piece in the chain is... color, and motion control.


i am going to a friends wedding today, but when i get back tomorrow, i will give it a shot.

3 Xylotex boards to run this beast are on their way - but someone with a plotter table could try it now with a sharpie?!

anyone?

again - thanks - you are obviousely spending some time on this question! i hope you decide to build a painting machine eventually.

gmtwo - it is "easy" to imagine how such a system would work. it would use the same 3d info that was used to carve the model to keep the painthead on a constant plane to the thing.

i bet its done 1 color at a time.. with a single airbrush or something..

having built some models - i would imagine there is still LOTS of hand and finishing involved in both the physical piece, and the painting.

that project needs a real budget.

woohoo - the xylotex boards are here

and i got some 8" hard rubber wheels and 5:1 reduction toothed belts, cogs, and pinions.

Ive got a bunch of parts to build now for the main wheel drives. Direct drive just wasnt going to work. i would have needed much larger motors, and the accuracy would not have been that great.


Anyone tried Tachus's software? I will try it with a pack plotter later tonite - probably kcam4 or mach2's built in plotter.

Tei! Can you give it a shot on your mill?

vacpress - great progress - and the boards are in - you'll be busy!!!
Say, how 'bout you clear your inbox - I can't get thru with a pm to ya!
Cheers - Jim

doh.

yeh- i have to start an archive file or something, cuz i have lots of very helpfull code chagrin, tachus, etc. have sent me clogging it now.

ill do it now

vac

Did you have tryed the software yet?
I hope we will get a working software.

i have looked at the docs... i am gonna try it tonite.

im also gonna build a simple unipolar step driver board for the Z axis pen holder, and for a stepper driven paint gun trigger. This board is gonna ride on the carriage.

I am being hesitant to do the main drive untill i goto a surplus shop and look for some componenents or wheels to simplify it. As it is, it willtake like 4-6 hours to install the parts - thats time i dont have.... Sad as that is.

The wheels i got are heavy and much better than the light, weak foam sports equipment wheels i had. these wheels have the advantage\disadvantage of having some bearings press fit in the center. These bearings are very sloppy, so i may clamp them reallty tight on a drive shaft between 2 bolts, so they dont turn.. I think that will work...

I have added a GUI window to edit the script settings. There is a small change in the way the redirected port is setup - check the instructions.

tachus - Great!

i have all the aprts and PCBs now - perhaps tonite i will have it moving... the main drive is proving tricky.. i am trying to keep it economical - difficult..

i saw some wierd steppers with 5:1 reduction built in, but they seemed to only be about 40 steps\rev?

anyone know anything about stepeprs with less than normal steps\rev?

http://michael.tilp.org/painter/painter.pl

Definitely a work in progress - right now it just spits out comments. Expect the script to die horribly if you don't enter in parameters that *I* expect ;). I also seem to be getting some funniness in the RGB to CMYK conversion so I'm going to have to write a visual aid to help tune it at some point.

I'm still lost on the GCode - I've found some incomplete references that will get me through the basic movement of the "printhead" but I have no idea what to use for GCode when I need to print a splotch/spray of color. Using Z motion works, but only for one color - I need to be able to create output for four colors. I can think of possibilities (commands for "spindle on", "mist coolant", "flood coolant", and ?) but ... dunno.

Does anyone know of a complete GCode reference on the web?

vacpress, I also need additional detail on your planned method of control for your paintguns so I can write this to fit it.

I thought the color control could be done with coolent control messages etc too, but looking a bit closer it might not be possible without alot of funkyness. There's flood and mist coolent GCodes but only one coolent off command for example which makes things kind of tricky. I suspect going the G Code route will just be a step on the way to a version with it's own controller accepting some "slightly like GCode" language. But there is another posibility Vac's proposed marking mechanisim has a stepper for each color which needs to be advanced to draw a pixel. There are 6 axisses avalible XYZABC or XYCMYK - just right.

I really feel, that you guys might be better off getting away from the g-code reference. You already know the travel path of the head. Each command is one step over. Spray color. One step over, Spray color. Reach end of path, move one row, return in other direction. The motion could be programmed into the on-board pic controller to give the given pulses to the steppers as needed for travel. After each step the “print driver” needs to send 4 numerical codes. Each a value for the given color: C – M – Y – K. The value is the density of each. This could be translated by the onboard controller to be either duration of spray – or maybe variation of pressure – or maybe both depending upon experimentation.
My current program only reads RBG values from a BMP and outputs them. I’m really looking into this ghostscript thing to handle all file formats, but have not had the time. Thanks tachus for the info, I’m looking forward to figuring it out but real work is getting in the way.
Likewise I have been trying to develop a 4 color ‘spit’ jet print head I was hoping to deliver to vacpress to mount on his gantry, but shop time has been difficult to sneak in with sold machines here taking priority. (Silly thing about trying to keep paying customers happy!)

I really feel, that you guys might be better off getting away from the g-code reference. You already know the travel path of the head. Each command is one step over. Spray color. One step over, Spray color. Reach end of path, move one row, return in other direction. The motion could be programmed into the on-board pic controller to give the given pulses to the steppers as needed for travel. After each step the “print driver” needs to send 4 numerical codes. Each a value for the given color: C – M – Y – K. The value is the density of each. This could be translated by the onboard controller to be either duration of spray – or maybe variation of pressure – or maybe both depending upon experimentation.
My current program only reads RBG values from a BMP and outputs them. I’m really looking into this ghostscript thing to handle all file formats, but have not had the time. Thanks tachus for the info, I’m looking forward to figuring it out but real work is getting in the way.
Likewise I have been trying to develop a 4 color ‘spit’ jet print head I was hoping to deliver to vacpress to mount on his gantry, but shop time has been difficult to sneak in with sold machines here taking priority. (Silly thing about trying to keep paying customers happy!)

Great thread! It looks like there has been a lot of progress made on identifying the RGB values of each pixel and knowing where the pixel (X/Y) needs to be placed. I have a PC controlled mill that could very easily be programmed to follow a raster pattern and trigger a blast of color at each spot. I can also sort of imagine how you could use an airbrush to deliver the right color to the paper at the right time.

What I can't get my head around is how to build a reasonably high-resolution picture from these spots. The spray equipment I am familiar can either produces a round or oblong spots with sharp edges or a fan that has great definition in one direction but a very soft edge in the other direction.

With round or oblong spots you can print a fish scale pattern with overlapping dots. The smaller the spots the clearer the picture. Is this the plan or is there a way to get squarer spots?

The wire jet Vac is modeling his machine after doesn't have good resolution only 2.5 or 5 dpi. Each pixel however is very transparent and several hundred need to be put in more or less the same spot to get 100% saturation. This means that each pixel has a full range of color as opposed to a normal printer with small dots where each pixel is one of seven colors. So photographic images will look ok when you stand back a bit but sharp edges won't look too good close up.

I finally found the airbrush the Pixelation.com machine was modeled after. Here is an annotated picture of it.

Steve

http://www.datagems.com/images/paasche.jpg

The script is now able to compensate for the geometry of the jets on the print head, also there is a option to generate a pixel on command for every on pixel instead of when it changes from off to on.

Sorry for the lack of comments in the script for anyone who want's to see how it works. It made my head hurt working out all the logic for it without trying to explain it to anybody else, any questions gladly answer though.

No instructions included with this file see the previous release.

Sorry i havent been around!!! I am still alive, and will be back.


2 weeks till semester ends. lots of work.. if your bored, check out one of my designs(unfinished)

www.vacuumstudios.com/Demo2.html

Yes I had noticed ther had been a bit of vacuum in the forum without you :)

ok.

i got some free time tonite. i am going to try and get the thing going.. all i have to do is make some wheel mounts on the lathe, attatch some steppers, run some wiring, and try it out..

the res. will be lower untill the 3rd xylotex arrives.

vac

I need a 3D text.
Can you fix that for me?
The text i need in 3D is

Belettering
PUBLITIME

jurgen

publi -

sure.

need more details!

explain your whole idea.

vac

It doesn't mather what kind of letters you use.
I just need 3D letters for printing.
I do not have to mill the letters.
Mab you have a idea how to make.
It just have to look very nice.


jurgen

is that a poem?

Has anyone made any more progress on this?

sdfine

HI EVERYONE!

it has been a long time, but i will eventually be back. i have been traveling and studying in europe for the last 8 months..i am writing this from rural france near a music festival im camping near with some friends and my new french girlfriend.

however, i have a new project planned, and will soon finish the first painter. there is lots of hard work others donated in the form of scripts and code. i will be digging that up and implementing it soon. thank you for the help, again.. especilly the guys who did the scripting examples...


in the meantime, i will mention that i am going to start a second plotter type device right away, designed to make wallpaper and cut the scrap edges.. this is part of a larger goal of producing a line of home decoration products produced using home-engineered equipment..

another idea, a giant scale "3d plotter" that uses that cool expanding insulation foam stuff that people use for filling the molding around doors. this is going to hopefully have a work area of 4' x 8' x 100' and is intended as an art project.. we will see.. i imagine it plotting data in 3d in realtime on a large scale... just like the painter project, but bigger,and 3d...

ciao, grazie

anyone has decent photos of wirejet or blade jet harware?! like pixation machine..i cant find any decent photo to see how and what they did...