LED "laser" printer illumination bar hacking

[josecanuc]

I recently did a teardown on a color "laser" printer that uses LED instead of a true laser for imaging. Of the many interesting parts inside, there are the LED illuminators that actually apply the image to the drum/belt.

The printer is a Xerox Phaser 6022 http://www.office.xerox.com/printers/color-printers/phaser-6022/spec-enus.html and claims 1200x2400 DPI. I doubt there are actually 1200 LEDs in each bar, and that the 1200 figure is interpolation by varying brightness of 2 adjacent emitters, and/or marketing wank of 4-color * 300dpi per color. I didn't count the distinguishable elements in the emitter strip, but it doesn't look like it is even 300dpi, though it's hard to tell what's the emitting element.

Each of the 4 colors (CMYK) has one of these light bars fed with a 40-pin flat-flex cable. There is a small circuit board with a blobbed chip on it, that then plugs into the light bar with an 80-pin connector.

Questions I have:

• Is this even interesting to play with?
• Is the light visible, or will it be UV or IR?
• Has anyone played with this kind of thing before?
• Is it even reasonable to think it could make a neat decoration driven by an arduino and some shift registers?

[tooki]

The printer is a Xerox Phaser 6022 http://www.office.xerox.com/printers/color-printers/phaser-6022/spec-enus.html and claims 1200x2400 DPI. I doubt there are actually 1200 LEDs in each bar, and that the 1200 figure is interpolation by varying brightness of 2 adjacent emitters, and/or marketing wank of 4-color * 300dpi per color. I didn't count the distinguishable elements in the emitter strip, but it doesn't look like it is even 300dpi, though it's hard to tell what's the emitting element.

No, they really do have 1200 LEDs per inch, so 10,200 LEDs (per color!) for an 8.5” page width. The little things you’re seeing are the lenses, which focus the light from the LEDs on the photosensitive drum. The individual LEDs would require a microscope to be seen.

You absolutely don’t want to vary the brightness in a xerographic toner process, that causes irregularity and blurriness. (Remember trying to make grayscale copies on old analog photocopiers? It was awful. Digital copiers that scan in grayscale and then dither it to pure black and white produce far better grayscale copies.)

But yeah, when you remember how small the individual transistors in modern semiconductors can be, making an array of 1200 LEDs per inch is trivial. I mean, a 12MP smartphone camera sensor packs 4000 photosites into less than 6mm — over 16,000ppi.   

This The attached white paper by Fuji-Xerox explains it in detail. (And indeed, they state that theirs is 10,240 LEDs, so I guess they gave themselves some wiggle room for borderless printing.)

The doc doesn’t say what color the LEDs are, but AFAIK, laser printers use either IR or red, so I’d expect the same from the LEDs.

Whether it’s fun to play with, or indeed how to drive them, I dunno!

Edit: replaced link with attachment, after the link went dead.

[matseng]

You absolutely don’t want to vary the brightness in a xerographic toner process, that causes irregularity and blurriness. (Remember trying to make grayscale copies on old analog photocopiers? It was awful. Digital copiers that scan in grayscale and then dither it to pure black and white produce far better grayscale copies.)

Apparently OKI does 16-level LED brightness in their MC873DNX according to this:

Multi-level vs. Bi-level Printing

Traditional laser printers apply dots using two levels of intensity: 100% black or 100% white. Dots are grouped together (creating halftone cells) to simulate levels of grey. This is referred to as bi-level printing.

By contrast, our multi-level LED printing technology can produce different levels of intensity. This produces up to 16 shades or tones in the range between black and white. This is similar to the effect you get when replacing a standard light switch with a dimmer switch.

Multi-level Technology delivers more shades of grey or color within the same size halftone cell. This results in a higher level of detail and greater color depth, another great feature from OKI Printing Solutions.

[tooki]

You absolutely don’t want to vary the brightness in a xerographic toner process, that causes irregularity and blurriness. (Remember trying to make grayscale copies on old analog photocopiers? It was awful. Digital copiers that scan in grayscale and then dither it to pure black and white produce far better grayscale copies.)

Apparently OKI does 16-level LED brightness in their MC873DNX according to this:

Multi-level vs. Bi-level Printing

Traditional laser printers apply dots using two levels of intensity: 100% black or 100% white. Dots are grouped together (creating halftone cells) to simulate levels of grey. This is referred to as bi-level printing.

By contrast, our multi-level LED printing technology can produce different levels of intensity. This produces up to 16 shades or tones in the range between black and white. This is similar to the effect you get when replacing a standard light switch with a dimmer switch.

Multi-level Technology delivers more shades of grey or color within the same size halftone cell. This results in a higher level of detail and greater color depth, another great feature from OKI Printing Solutions.

Oh wow, I didn’t know that. Thanks for the heads-up!

I guess that with digital, they can control the process well enough to make it viable. (I’m speculating here, but I think that much like phosphors, the xerographic process is highly non-linear, requiring gamma correction. And I further speculate that it’s much more complex for xerography, insofar as you have to deal with how the toner behaves when fusing.)

[nali]

755nm apparently for the Oki LEDs.

Here's some slightly less marketing orientated stuff https://www.oki.com/en/otr/2006/n208/pdf/otr-208-R05.pdf

[Circlotron]

 Maybe you could use it to make some kind of high resolution vertical “thermometer” display, e.g. sound level.

[jbb]

I wonder if that wavelength would make 3D printing resin go off... printing all the way across the bed in one go could be nice.

[Siwastaja]

I guess that with digital, they can control the process well enough to make it viable. (I’m speculating here, but I think that much like phosphors, the xerographic process is highly non-linear, requiring gamma correction. And I further speculate that it’s much more complex for xerography, insofar as you have to deal with how the toner behaves when fusing.)

Yeah, wouldn't have expected that.

A decade ago, I played with trying to make grayscale analog copies. The workflow was:
1) Image shot on B/W film and developed
2) Analog (photographic) prints made at lowest contrast possible (00 contrast on multigrade paper + paper pre-exposure to bias it with slight gray tint)
3) Doing the copies.

What I found out was that one particular analog copier made fairly good and almost consistent results! When copied in large batches, it did heat up and the brightness setting needed adjustment on the fly, but other than that, completely usable grayscales - better than what I could achieve with my crude DIY (analog) halftoning setup .

Then, with another copier, the contrast was simply too high, and the uniformity was so bad it was unusable. With that machine, the results looked like how the old-school analog copies of photographs indeed usually look.

So while I'm a bit surprised they are doing this, it's not definitely impossible since I was able to kind-of do it completely analog & with some luck. Done right, it would be a great benefit in printing photographs. Maybe they have a feedback, scanning the output to make the tones match when the drum ages etc.?

[Zero999]

The printer is a Xerox Phaser 6022 http://www.office.xerox.com/printers/color-printers/phaser-6022/spec-enus.html and claims 1200x2400 DPI. I doubt there are actually 1200 LEDs in each bar, and that the 1200 figure is interpolation by varying brightness of 2 adjacent emitters, and/or marketing wank of 4-color * 300dpi per color. I didn't count the distinguishable elements in the emitter strip, but it doesn't look like it is even 300dpi, though it's hard to tell what's the emitting element.

No, they really do have 1200 LEDs per inch, so 10,200 LEDs (per color!) for an 8.5” page width.

Expanding on that. Colour printers have four print heads: one for each colour.

[amyk]

• Is it even reasonable to think it could make a neat decoration driven by an arduino and some shift registers?

It is almost certainly itself a shift register; I'd guess 10 parallel strings of 1024 pixels each. As for how to drive it, you may want to look for a service manual (there's a PDF of one chapter floating around on the Internet, but I haven't found the whole thing) or if the unit is still working, use a logic analyser/oscilloscope on the signals. The protocol won't be very complex but you still need to figure out the details.

[josecanuc]

Appreciate the input, everyone. I can imagine, if there are 10k+ elements in one strip, and 2400 lines per inch perpendicular, this thing would have to pump out 23 Mbits for every inch of paper travel. Specs say 18 pages/min => 3.33 seconds per page => ~77 Mbits per second for each color, assuming only binary LED states. I can only hope that if it's a shift register, the clock is allowed to be slower!

Unfortunately, the unit is no longer working, otherwise my wife would not have allowed a teardown!

[matseng]

And it probably can be even worse than 77Mbit - some units have subpixel timing adjustments to fix any imperfections of the placements of the LEDs in the ledbar....

[tooki]

Wanna see insane data rates? Look at HP’s pagewide continuous web inkjets. Several feet wide at 1200dpi, with feed rates of several feet per second, with 8 inks. I did some back-of-the-envelope calculations a few months ago and arrived at a couple of GB per second! (And those things support variable data, so it’s not just the same buffered, pre-rendered data over and over, necessarily!!)