Hacking an Inkjet printhead

[ali_asadzadeh]

Hi,
I have a brilliant Idea about building a custom printer, I want to know if we have a standard regarding the inkjet printer heads?
Do we have a such thing? can we control them? do you have any idea?

[DmitryL]

There are just about 10 inkjet heads manufacturers in the world. They produce about 200 various industrial inkjet head types. There is absolutely no standard at all.
BTW, industrial stuff is quite expensive.. You can start with contacting head manufacturers (like Ricoh, Xaar, Seiko, Dimatix) and asking them for price quote and datasheets.. Good luck

[Mechatrommer]

I have a brilliant Idea..

right... this is my damaged epson printer head, six color, 180 nozzles for each color in 0.14mm pitch... altogether 1080 nozzles controlled by 29 pin cable... maybe someone will have a way to figure that out... i have lower end print head that maybe consist of 1/10x nozzles from that, but they are stuck somewhere in my junk store... have no time to dig them out, maybe some day.. (2nd picture is the piezo elements (mostly damaged in the center area because i accidentally scrap them off with my finger) pulled out from blue arrow in the 1st picture...) btw, canon use heat element instead of piezos, so they maybe entirely a different fun..

[CM800]

The only printhead I know of that is openly documented is the  C6602x.

[ali_asadzadeh]

Thanks for the hints, But Do not give up so soon

I have found this one, but it's pricey (250$) and have low res of 200DPI.

http://d1.amobbs.com/bbs_upload782111/files_36/ourdev_619324HXZTX0.pdf

So do we have something better? with better resolution, cheaper and the ability to transfer the Ink with an external 1 litter tank of Ink! apparently I need a lot of printing

[Mechatrommer]

Nice documentation, i will need 6 of them that makes $1500, excluding the receiving and ink feeding assembly, xy gantry paper feed etc. That 200dpi head is good enough already similar or close to the printhead i attached earlier. Its photo printing resolution grade no wonder its expensive... its maybe worth the effort considering hi volume ang large format photo printer may reach 5 digit price..

[cdev]

Back in the day I had many artist friends and many of them owned large format inkjet printers that as far as I know did not use cartridges of any kind. They filled them with bottles of ink.

I had some friends who sold software that controlled them. This was 25 years ago. The prints their printers made were absolutely gorgeous. Crappy consumer inkjet printers are designed to break so you spend more money. Its a false economy.

[DmitryL]

As a rule of thumb, industrial inkjet heads cost is about $1 per nozzle And it is not uncommon to have around 2000 nozzles per one head...
And you will need ink supply system (possibly with flow/pressure/temperature control). Also  head driver electronics (Arduino is unlikely to fit here). And software.
And mechanical transport. Also you will need to persuade head supplier that you are worth dealing with.
Getting proper ink might be an issue too.
Though some people make low grade inkjet printers, google might help here.

[ali_asadzadeh]

Nice hints, I need to do Color printing at most twice the price of a normal commercial inkjet printer (around 150$), so any Ideas are so damn welcome, if we could hack a Color print head or find another cheap way of doing it, we have lot's of new Ideas and toys to deal with. Please participate in the discussion and bring your new Ideas for hacking the Head

[Kalvin]

There are quite many links available about hacking the printhead. For example here are some:

http://ytec3d.com/forum/viewtopic.php?t=47
https://ytec3d.com/forums/topic/step-1-hacking-the-cn642a-inkjet-printhead/
http://www.piclist.com/techref/pcb/etch/custom-vs.htm

[tooki]

Back in the day I had many artist friends and many of them owned large format inkjet printers that as far as I know did not use cartridges of any kind. They filled them with bottles of ink.

You are probably thinking of the Iris continuous inkjets. They use pumps to create a continuous page-wide curtain of droplets, the unwanted of which were electrostatically deflected into an ink dump. They printed at 1800dpi back when consumer inkjets struggled to apply 300dpi cleanly. Even today, many large-format inkjets use bottled ink.

I had some friends who sold software that controlled them. This was 25 years ago. The prints their printers made were absolutely gorgeous. Crappy consumer inkjet printers are designed to break so you spend more money. Its a false economy.

They're built down to a price, not to break. The printer companies lose money on cheap machines, remember? They're making ALL their money from the ink cartridges, so it's in their best interest to make the printer last longer while being cheap to make.

When's the last time you had a printer break? I've never had an inkjet printer fail, other than an Epson with a clogged head that no amount of cleaning would clear. In terms of mechanical reliability, the printer industry has done incredibly well, especially given the cost constraints. I've replaced printers because I wanted an upgrade, not because it died.

I love the idea of basically dirt-cheap ink in a costlier printer, but most people don't want to spend $600 up front, they'd rather spend $150 up front and amortize it later via expensive ink. In the end, the TOC is probably not very different.

[tooki]

Hi,
I have a brilliant Idea about building a custom printer, I want to know if we have a standard regarding the inkjet printer heads?
Do we have a such thing? can we control them? do you have any idea?

By pure chance, I spent most of yesterday and the day before reading hundreds of pages of old HP Journal articles about the development of inkjets. (Issues May 85; Aug, Oct, and Dec 88; Aug and Dec 1992; Feb 94)

In a nutshell… no way is a hobbyist gonna be able to drive an inkjet head optimally. The timing and electrical requirements are intense. Printers drive them using custom driver ICs that you simply can't get, other than tearing apart a printer. And the mechanical requirements are insane; it's frankly miraculous that they can be made affordably, never mind cheap as they are today.

I dunno what application you have in mind, but chances are you're better modifying an off-the-shelf printer.

[ali_asadzadeh]

Thanks, the links was fantastic,But I wonder how do they do Gray scale a color? it seems all the heads can do simple binary color!

[amyk]

Epson printheads have been reverse-engineered and driven acceptably with third-party electronics; more information on the protocol here, which seems to have not changed much since the very earliest models:

http://techref.massmind.org/techref/pcb/etch/custom-vs.htm
http://dspace.mit.edu/bitstream/handle/1721.1/85435/870681346-MIT.pdf?sequence=2_1
http://web.archive.org/web/20120210162510/http://jeffreygough.com/files/27c3-file-print-electronics.pdf

Not surprisingly, it's based on a shift register and operated similarly to an LCD or EPD driver.

Keywords to search for: "Epson" "printhead" "NCHG"

[tooki]

Thanks, the links was fantastic,But I wonder how do they do Gray scale a color? it seems all the heads can do simple binary color!

Correct. Shading is done with halftoning, where dots are spaced apart to produce lighter shades (e.g. 25% gray is done by only placing roughly 25% as many dots as for black, though in actuality you must apply a gamma curve because human brightness perception is nonlinear). In addition, some printheads can modulate the dot size itself. (Modern printers often have multiple heads for the same color, for different dot sizes, allowing for smoother color gradation.)

No offense, but you don't seem to know even the most rudimentary basics about printing, never mind the specifics of inkjets. How do you ever hope to build a printer if you don't understand the fundamental principles of ink on paper?

[Mechatrommer]

Modern printers often have multiple heads for the same color.

modern printer use variable droplet size or "on demand" from the same head, no need separate head = more expensive.

No offense, but you don't seem to know even the most rudimentary basics about printing, never mind the specifics of inkjets. How do you ever hope to build a printer if you don't understand the fundamental principles of ink on paper?

nevermind all the color theory... once controlling the ink jet from the print head is successfull, everything else should become easier. the rest... color theory, xy gantry like in 3d printers, all the information are easily available in the net already...

[tooki]

Modern printers often have multiple heads for the same color.

modern printer use variable droplet size or "on demand" from the same head, no need separate head = more expensive.

Well, except that not all printers use variable droplet size from a single head, because using multiple heads has some advantages.
In essence, there are two ways to achieve good continuous tone color in inkjet printing, and they can be combined to varying degrees:
1. Maintaining larger droplet sizes, but adding more inks of decreasing density.
2. Maintaining fewer ink colors, but varying the droplet size.

Adding more ink colors obviously increases cost, so it's mostly done in professional photo printers. For example, Canon's top-of-the-line photo printer, the imagePROGRAF PRO-1000, uses 12 inks, with fixed 4pl droplet size, using 1536 nozzles per ink, for a total of 18,432 nozzles. That's one heck of a print head, and a lot of inks to stock. So smaller printers for home use tend to use fewer inks, but varied droplet size. The droplet sizes can be achieved either by using separate nozzles as Canon tends to do, or by varying the droplet size from a single nozzle, as Epson does.

Using separate nozzles means much more tightly controlled droplet size and shape, compared to variable-size droplets from a single nozzle size. Another huge advantage is that separate nozzles means you can do a single pass for all the available dot sizes, whereas with the single nozzle approach, layering the droplet sizes would require multiple passes.

For years, and today still, in its better home/SOHO multipurpose photo printers, Canon uses heads of 512 nozzles for color inks, and 1024 nozzles for pigment black. Look at the Canon TS9020 from last year: it's got 1024 nozzles for pigment black, and 5632 nozzles for color. It uses CMY and photo black and gray inks, so those 5632 nozzles are allocated among the 5 dye inks. 5632/512 is 11, meaning 11 print heads. Given how Canon has typically allocated them in the past, I expect something like this configuration:

5pl (or larger?) pigment black
5pl cyan
2pl cyan
1pl cyan
5pl magenta
2pl magenta
1pl magenta
5pl yellow
5pl photo black
1pl photo black
5pl gray
1pl gray

You can see this on the nozzle test pattern from the manual:


On the photo black and gray, I can't tell for sure which nozzle sizes they are, since I only have the low-res reproduction of the test pattern, not an actual printout and a microscope to compare with. But it's clearly visible (and aligns with the stated nozzle count) that there are two heads each for those inks.


Sadly, on the newest models that came out this year, Canon is no longer publishing any nozzle configuration specs whatsoever.

The current Epson XP-960, a six-color printer, also states no nozzle counts, but says "5 ink droplet sizes, as small as 1.5 picoliters". Its nozzle test pattern in its manual is significantly simpler, showing just 6 heads of unknown nozzle count:



The Epson SureColorP600 photo printer is specified as 2pl minimum droplet size, with 3 available drop sizes, in a 180x8 nozzle configuration.

HP gives no info whatsoever.

Brother is slightly more forthcoming. The MFC-J885DW for example is a 4-color printer, stated as having a 210 nozzles per color of unspecified droplet size.

No offense, but you don't seem to know even the most rudimentary basics about printing, never mind the specifics of inkjets. How do you ever hope to build a printer if you don't understand the fundamental principles of ink on paper?

nevermind all the color theory... once controlling the ink jet from the print head is successfull, everything else should become easier. the rest... color theory, xy gantry like in 3d printers, all the information are easily available in the net already...

If someone hasn't even heard of halftoning before, do you really expect them to be able to understand and implement practical color theory? Or the drive electronics for that matter? Indeed, if they don't understand how the printing process works, how would they even be able to determine that they are driving the print head optimally? Let's be real here...

[DmitryL]

Some printhead manufacturers recently started to inroduce cheap heads for low-end market, e.g. Ricoh GH2220. I think they should be easier to obtain than high-end ones.
Driving these heads is not simple though... And getting proper ink + corresponding driving waveforms should cost a bit.

[Mechatrommer]

If someone hasn't even heard of halftoning before, do you really expect them to be able to understand and implement practical color theory? Or the drive electronics for that matter? Indeed, if they don't understand how the printing process works, how would they even be able to determine that they are driving the print head optimally? Let's be real here...

the color theory or "method" only comes when you have something to play with. you dont have an ink spurt out of a nozzle, you dont have anything to play with, playing theory is useless anyway... at the beginning when we dont have variable droplet, we have different method to spurt ink, and then we have variable droplet, another game (theory) comes in... before inkjet, halftoning is made from mixing the right amount of white and black paint, and a good hand strokes, with inkjet, its different game, there are many bunch of halftoning method for inkjet alone... btw, you can be a theorist master all you like, you can learn all color gamut and all, in the end, you dont have something to support your theory... everybody else who doesnt learn will be more happy with their gamut limited printer... ymmv...

[ali_asadzadeh]

If someone hasn't even heard of halftoning before, do you really expect them to be able to understand and implement practical color theory? Or the drive electronics for that matter? Indeed, if they don't understand how the printing process works, how would they even be able to determine that they are driving the print head optimally? Let's be real here...

It's in the forums or the net that somebody like you enlighten us and prevent us from not knowing the unknown! thanks for saving me

And please remember, once you could control the head, the remaining is software! and we can solve that later or change it as we wish.

[nali]

It's in the forums or the net that somebody like you enlighten us and prevent us from not knowing the unknown! thanks for saving me

And please remember, once you could control the head, the remaining is software! and we can solve that later or change it as we wish.

Assuming you'll be printing from RGB you'll need to convert it to CMYK, so you'll probably need to read up on color spaces, particularly LAB https://en.wikipedia.org/wiki/Lab_color_space

Loads of info out there about halftoning https://en.wikipedia.org/wiki/Halftone, Google is your friend.

But as has already been said all of this is pretty academic until you actually have something to splat ink on paper.

[Gyro]

I know it's cheating but have you considered re-purposing the mechanics and electronics of an existing printer? I don't know what sort of custom printer you have in mind but maybe you could fake the vertical position encoder inputs etc. so that the existing printer S/W wouldn't notice the different configuration.

Worst case you might be able to decode the control signals between the microcontroller and the head driver while printing a series of simple test patterns.

Just a thought.

[ali_asadzadeh]

Thanks, Realy nice hints

[tooki]

If someone hasn't even heard of halftoning before, do you really expect them to be able to understand and implement practical color theory? Or the drive electronics for that matter? Indeed, if they don't understand how the printing process works, how would they even be able to determine that they are driving the print head optimally? Let's be real here...

the color theory or "method" only comes when you have something to play with. you dont have an ink spurt out of a nozzle, you dont have anything to play with, playing theory is useless anyway... at the beginning when we dont have variable droplet, we have different method to spurt ink, and then we have variable droplet, another game (theory) comes in... before inkjet, halftoning is made from mixing the right amount of white and black paint, and a good hand strokes, with inkjet, its different game, there are many bunch of halftoning method for inkjet alone... btw, you can be a theorist master all you like, you can learn all color gamut and all, in the end, you dont have something to support your theory... everybody else who doesnt learn will be more happy with their gamut limited printer... ymmv...

No no no... please dude just stop.

1. If you don't know how the ink is supposed to be applied, how can you even know that you're driving the print engine properly? You don't have to be a master, but you must understand the basics, and that's not the case here.

2. "Before inkjet, halftoning is made from mixing ... white and black paint" -- NO NO NO! Halftoning specifically refers to using dots of variable size and/or spacing to simulate continuous color. Mixing paints, on the other hand, is continuous color. They are NOT synonyms!!!

3. Halftoning predates inkjet by a LONG time. It was introduced in the mid-1800s and has been used ever since.

It's in the forums or the net that somebody like you enlighten us and prevent us from not knowing the unknown! thanks for saving me

I really hope you look at the HP Journal issues that I listed earlier. That will give you an idea of the magnitude of work ahead of you if you insist on this endeavor.

And please remember, once you could control the head, the remaining is software! and we can solve that later or change it as we wish.

Yes and no -- driving the heads is not a trivial task when you're applying hundreds of thousands of droplets per second (or millions, in the case of Canon's large heads). That's not just software, it requires demanding drive electronics. It's probably not feasible to make the drive electronics fully software-defined.

I know it's cheating but have you considered re-purposing the mechanics and electronics of an existing printer? I don't know what sort of custom printer you have in mind but maybe you could fake the vertical position encoder inputs etc. so that the existing printer S/W wouldn't notice the different configuration.

Worst case you might be able to decode the control signals between the microcontroller and the head driver while printing a series of simple test patterns.

Just a thought.

I suggested modifying an existing printer 2 days ago. It's far more sensible than trying to roll your own.

[Old Don]

Plus ink jet printers don't reproduce all known colors since their ink is "real" vs. theoretical and by using CMYK they can't produce all the spectrum Pantone standard colors in the first place. In fact the need for "K" (black) in the CMYK printing is necessary since adding the other 3 colors to produce black isn't perfect and so "K" black is a cheat to improve the results. Plus you can't take a RGB screen and expect the CMYK printer to match since they use different methods to produce their outputs. Even professional printing on paper cheat and use 3 or 4 color machine that has extra rollers with custom color ink to match a special Pantone color their customer demands on the product.

Maybe you need something like this: