Slashdot Mirror
Creating Your Own Printer?
hajo asks: "I am in need of a Large (60" plus) printer which can print onto any thickness material for a specific art/robotics project. I loved the earlier Slashdot story where the two students used two motors and an inkjet can for large mural prints; but I need a higher quality end result. I can build a plotter mechanism with two PC controlled stepper motors; But I would like to find out how to use head the parts from a cheap inkjet printer. Where can I find info on the hardware and drivers for such a project. I have a hard time believing that I'm the first who wants to use the ink jet head parts of a printer to do something with them. Any hints, tips and URLs deeply appreciated. I believe this project will make for an interesting read and as thanks for any help I will keep the Slashdot community informed of any results."
Some guy named Gutenberg made his own printer centuries ago.
> ...which can print onto any thickness material...
How thick is "any thickness"?
Ten meters? More?
This is potentially a very big printer unless the entire printer sits on the top of the material in question.
Go here to create your own Slashdot dis
We do something kind of similar at work, where we print stuff 24/7 on huge presses running hundreds of feet per minute.
For some of the variable imaging there is row of hacked inkjet cartridges, I think Lexmark with the removable ink-sponge cartridges. They snap off the ink cartridge, and snap in a custom plate with a tube running to a pump and bucket of ink. The electronics are all custom.
I haven't worked directly with them, but even if I did I couldn't tell you any more. I guess you could take it as proof that it's possible...just grab a printer and start playing with a multimeter and power supply. Once you get the right voltage and pin mapping, you're ready to design a solution. If you talk to a professional ink supplier, they may be able to get you an ink formulation that will work in the cartridge...it has to be special. These can work for a long time at a pretty good speed.
...
I found something on the use of car alternators as stepper motors (used at least once to drive a router as a "print head") at http://www.tinaja.com/glib/resbn46.pdf. But it's got no details. (The bizarre porn sites showing up in the results of my search have to be seen to be believed.) The stuff at tinaja.com seems to be the most relevant; if you are trying to sling a big print head over a large surface and you aren't too concerned with sub-millimeter accuracy, that's probably your best bet.
Time is Nature's way of keeping everything from happening at once... the bitch.
I don't know that I'd try to use commercial print heads at all, because these are precision devices designed with the idea that they have complete control over the print media - level and movement. You're dealing with devices that operate in terms of thousanths of an inch. I just don't think you're gonna get that level of precision from a custom built device and it would be reflected in the output. In particular, I wouldn't think Color would work very well.
It's fairly simple to build something providing the surface you're printing on is flat. You can build a sturdy 2-d frame to hold the printheads across a flat surface, it's the variable height stuff that is a problem. You could rig something up with sensors that could move the print-head up and down, but I wouldn't try it in anything other than unidirectional myself.
For something of this size, I think you'd be better off trying this with something old. Modern ones would kill you in trying to get the head to be just close enough and at just the right location along the surface.
"Why don't you just use a paintbrush?"
There, that should get me a +3 Insightful.
"Derp de derp."
Commercial inkjet carts will be worthless in this application. Very large format prints are typically done at 70 dpi, I've seen billboard proofs as low as 15dpi and they look great (well, when seen from billboard distance).
You want something that can blast out huge dots, not microscopic 1440dpi dots. Plotter mechanisms are difficult to engineer with precision at that size. Most of the largest format printers use a rotating drum like Iris inkjets. You should see Metromedia's custom printers, they use drums the size of railroad boxcars. It is much easier to keep a drum spinning at a constant speed and just run a printhead past it at a fixed speed, than to accurately advance paper in fixed increments through a conventional printer mechanism. Trust me on this, I used to be an Iris technician.
Your Own Printer
This is not unlike asking, "I want to take the drive electronics off the hard drive and read and write to the platter with the existing head and motors. How do I do that?"
Not only are different print heads from the different manufacturers driven differently, printheads within a single manufacturer are driven differently. There are several ways to eject a drop of ink from the head, and several ways to adjust its trajectory in the air. I suspect the industry has largely moved to one or two methods that are similar, and I also suspect that the driving electronics are very similar, at least from a high level view.
What you'll find, however, is that each printer has a specific ink, a specific nozzle, and a specific distance to the media, such that you will probably find as many similarities as you will find differences between two printers.
Why is this important? This is the reason why very few people have done this sort of endeavor.
However, you are probably looking for a low resolution (100-200 dpi), and are only looking at doing one of these things. so go get a cheap printer and an oscilliscope and read the voltages present at the head when it's printing. You will likely need to duplicate the voltage and pulse length, but you can probably ignore the pulse shape and perhaps even the slew rate since you are going to have a lower resolution and you won't care if the thing breaks down in 1000 hours rather than 10000 hours.
If you trace the circuitry, you might find they use separate chips to drive the head, and will produce the right logic levels given a simple digital command - some judicious signal hunting will tell you all you need (or some good data sheets). This might give you an edge, since you wouldn't have to create your own drivers.
If you want to try and use more of the printer, then you can hack the sensor that tells it how far along the track it's gone. Just expand it (digitally or by building a new sensor) and then hook its stepper outputs to a higher current drive and have that drive the real stepper. Just magnify everything. You'll have to deal with a low ink output, though, so there's always some gotcha.
Then you get to enjoy trying to keep the print head exactly x millimeters from the paper across the 60" width. Usually this is done with a long, large, rigid drum os some sort and some tensioning drums.
You didn't give us much info though, so I can only answer the question you have, not solve the problem you have. What about your project makes a pen plotter type system inapropiate?
-Adam
I would recommend you print on smaller sheets and transfer the ink to the surface, rather than try to create such a large printing device. I do know they make sheets designed to "iron-on" transfer to other surfaces; whether the surface you plan to use is appropriate, I don't know.
An inkjet printer precisely moves a print head back and forth above the paper path. That's the part of the printer you want to keep. Another part of the printer precisely moves the paper, you want to replace that part of the printer.
Find the wires that lead to the stepper or servo that precisely moves the paper. Replace that with a larger motor that can precisely move what's left of your printer. You will probably need to boost the power. You can use the old stepper signals to drive opto isolators that drive some big MOSfets. (or a simpler circuit of your choice, that just happens to be how I used my Lego Mindstorms to control some much larger motors. Google for "y3md".)
If you can't dissassemble your printer so that the printhead can scan across your workpiece, start over with another printer. (that's why it is good that the printers are so cheap.)
The next challenge is getting each pass lined up with the previous pass. This is very hard. Don't even try. Instead aim to overlap each pass. At the start of each pass, precisely measure the overlap, and generate an appropriate image to render. It is far easier to measure something than to move something.
The article is still there (http://news.bbc.co.uk/1/hi/technology/3045158.stm )
2 35&mode=thread&tid=133&tid=137&tid=186&tid=194 )
and titles "Giant printer goes on show"
and related slashdot entry is Giant "Inkjet Printer" (http://slashdot.org/article.pl?sid=03/07/07/1117
What? He begins sentences with conjunctions. He employs sentence fragments and makes you guess the meaning. He uses imprecise and vague language (how do you flip "half of" a switch?).
Don, hire a proofreader! It's too painful to read!
For me I think the drivers would be the hardest part to come up with. I could tinker with wormgears and other such hardware all day. But I'm not much of a programmer.
I've hit Karma 50 and gotten a Score:5, Troll... I win!
pun! seriously, though, the simplest way i can think of is to hack a plotter mechanically by replacing the paper roller mechanism with a sturdy stationary frame, so that instead of the paper moving under the plotter heads, the whole plotter moves over the paper (or whatever surface). you could then just sit the frame on the surface and the plotter would move over it, drawing inside the frame.
media thickness or stiffness wouldn't matter because the device sits right on the surface (as long as the surface is flat). if the piece is bigger than the frame, plan it out and split the graphic into tiles (have the plotter draw some light hashmarks that you'd match with marks on the frame so that the tiles line up). it should be simple to program automatic segmenting.
if it makes the concept any clearer: you'd basically have a logos turtle (or a gang of colored turtles), running around on top of your medium. or is it a giant etch-a-sketch?
good luck. however you solve the problem, i hope you post the results!
"This is not a sig." -- R.
I missed whatever earlier article the submitter was talking about... any got a link?
Buy a printer to do it, they exist, they're expensive.They can use ink designed to print on your surface.
Buy a plotter. Glue Paper to surface, laminate or paint it if required.
Attach your printer beast to a CNC rail system. They make home use ones, it might give the rigidity, speed and location control you want.
1 Will work, and can be adjusted, it is very expensive.
2 Standard inks are designed for paper, they are okay for special transparencies. General surfaces they suck.
3 This is a hack, getting proper alignment will be difficult and expensive. Better to use a real solution.
Of course there is the obvious Designjet . But that isn't as fun as putting a scope on the lines to an inkjet head to get an idea of what it takes to make dots. Another idea might be to use an old dot matrix head. Or like the spray paint can printer, but use an airbrush head for finer dots. As far as the programming, once the image is in a bitmap, it would be straightforward to loop through the pixels while stepping the motors.
This project sounds like fun, let me know if you happen to be in the Atlanta area.
I looked into this a while back. Unless you have the ability to make tiny springs and gears, mold your own plastic and metal, you're not going to be able to make one of these things for less than you can buy one (or three!).
The specs for the off-the-shelf printheads are unavailable. And the interface is designed to be difficult to reverse engineer to prevent 3rd party printheads from becoming available.
a simple 3 axis BASIC STAMP controlled arm connected to a paintbrush ought to do it.
use a roller feed off an old printer to pull the sheet along and mount the arm (or multiple arms for different colours holding airbrushes) to do it.
I have a hard time believing that I'm the first who wants to use the ink jet head parts of a printer to do something with them.
You aren't.
Take a look at Output magazine. It's the trade rag for the digital print finishing industry.
Ink jet printers have been used for everything from cake decoration (photo quality image made of frosting. Yum, Yum.), to etching brass and wood.
You can probably buy what you're looking for without having to be creative or inventive at all. Though that might take the fun out of it.
Next, you will need to rig a larger ink reservior for the print head. This should be pretty easy if you have any mechanical aptitude and access to a machine shop. You also need to secure a source of compatible ink.
I'm tempted to say you should skip the print head altogether and use a simple spray nozzle (disassemble a cheap airbrush with a fine nozzle and actuate the needle with a solenoid), but this would mean that you would be working at very low resolution (maybe 10 or 20 dpi). You will still need to maintain a reasonably constant distance from the media, but you tolerances will be less demanding and you will have a wider choice of inks.
Finally, you need a three axis robot to position the head over the media. You can sense the distance from the media optically (a couple photodiodes, some lenses and a diode laser should do the trick). Getting exact positioning of the head in the X-Y direction is a bit harder (good mechanical positioning over 2 meters will be expensive) but you can probably use an ultrasonic system and get reasonable results (there used to be digitizers for CAD use that had an ultrasonic clicker in the pen/puck and a pair of microphones on perpendicular edges of the tablet. Either the delay from click to reception, or the amplitude at the mics, was used to determine the pen/puck position). Just put a ultrasonic clicker on the print head, and click a couple of times on each X-Y positioning operation to fix the head location.
Assemble the above parts, along with some custom hardware and software for interface and control, and you should be good to go.
A BBC micro and a turtle is all you need. Simply write a postscript - turtle filter (we did the reverse as Bsc coursework) and away you go. Want colour? Use 3 turtles.
The reason we were called in to design it was they had an existing system where it printed the page, trifolded, and stuffed... but then had to be labeled seperately. All they wanted was to modify the stuffer to print the envelopes as they went in.
In the end, the printing was trivial but the control system was not. The final solution was to buy a standalone envelope printer and print the stack, and load the stack into the stuffer. Syncronizing the two wasn't as bad as trying to do it on-the-fly.
There was an article on Slashdot a few months ago with a printer that you kinda "waved" over a surface and it would automatically print dots where they were supposed to go when you waved over the correct position.
From the demos that I saw on the article, it printed some pretty good images.
It was about the size of an old hand-scanner.
Look around, I'm sure you can find it. I just don't remember the name. It would be perfect if you need to print text/basic images on any surface.
- It's not the Macs I hate. It's Digg users. -
I would build it in the style of a CNC machine. There is plenty of documentation available, I actually know some one who built his own little cnc out of mdf a dremel some stepper motors and a lot of hacking. The beautiful part is the parts were cut on another CNC machine. In addition the controller software is freely available. The suggestion to use airbrush is actually a good one. I have heard of a CNC spray painter that uses standard spray cans to do computerized grafiti. Using the CNC type model you can have it draw either raster or vector graphics and with some work do continuos tone work. Hell you do it well you have a new job.
Also, for convenience i would think of working with 4x8' panels and doing it table top style and tiling if you need larger.
One of these might scale up to what you need.
I'm in a similar boat and am seeking same.
One pricy but ultrahigh quality option I've considered is sending my files to a company that does giclee printing. These guys seem to have a good deal (the co-op program) http://www.crimsonatelier.com/ , I haven't sampled their services personally however.
Have you looked at any shitty old dot-matrix printers?
I've considered a couple of designs : Big plotter and donut hoverbot.
-Big plotter is obvious. I've seriously considered doing away with any mechanically-indexed-coordinates type control due to the pricyness of the parts. How about a print head that checks it's position with lasers looking at reflecters? You could use REALLY cheap stuff for your frame then, like washingmachine motors and clothesline.
-The idea I'm looking hard at is a donut-shaped hoverbot. It would hover on some fans under the donut-body and the print-head would reside in the center of the hole. It'd move around via a few more fans. Mechanically very simple. The air-cushions might not fuck up the wet paint. It could keep track of it's position with a spinning laser-eye thing or radio beacons or something. It'd be alot simpler and more foolproof than a plotter. More portable too. Use it on a breezeless basketball court.
U got images of your stuff on a site?
who has one of these. I used one in the CIT lab here at UT, but I think most Kinko's or drafting places will do it for you. I've done several large art prints with these, and they can't be beat. Building your own is a neat idea and all, but unless making your own large format printer is part of the piece, I'd stick with the professionals. No need to reinvent the wheel, man.