Canon i9900/9950 Color Problem

Started Jun 29, 2010 | Discussions thread
bobsokay Regular Member • Posts: 227
Re: Canon i9900/9950 Color Problem

Hi Rick,

I am very excited as I have now fixed my printer! But the story is long and painful and I hope you will bear with me. Up front I want to immediately state that my whole focus on the media sensor turned out to be wrong.

The clues pushing us towards the media sensor made perfect sense, but ultimately we can use the same clues to draw a completely different set of conclusions.

The short answer is that I needed a new print-head for my i9900. I hope my experience saves somebody some trouble. The story goes like this...

The clue that drove several of us to suspect the media sensor is that printed photos on plain-paper with the settings selected for plain paper all looked correct. When we switched to any other paper, the colors came out crazy. The only thing which knew the paper type is the media sensor, so this seemed a logical conclusion. Even Canon puts into their service manual, a debug path that talks about wrong colors and replacing the media sensor.

But here is the interesting part. The defect is not the media sensor, it is the print-head! I replaced my i9900 print-head with a new one and instantly cured my magenta shift. It took me a long debug path, the purchase of a second defective printer (with a different failure) and a lot of head-scratching to come up with an explanation that makes sense.

Please feel free to follow along, ignore the rest of this, or offer comments.

Thinking that I had a broken media sensor, I ended up purchasing a second i9900 on eBay that stated it was for parts, since the print-head was indicating as broken. When I received the second broken i9900, I moved my original print-head over to that printer (let's call it #2) since the problem with that printer was a totally failed print-head (5 orange flashes on the indicator LED).

I was absolutely astonished to see that printer #2 now worked fine--except it showed the exact same bizarre magenta shift. In other words, the color-shift problem had followed the print-head to the new printer.

I studied the test prints for a long time and sat thinking about the issue. Suddenly it dawned on me that the problem I was seeing was not the classic clogged print-head, where some colors are offset because something is missing, but rather that I was seeing the presence of extra ink, including some areas that should have been nearly white or very light gray or blue.

The service manual shows that the connections to the print-head look like a shift-register input, with odd and even data pins for each of the colors, and a master shift clock plus a latch input. Heating is then triggered by heat-enable pins.

It finally made some sense. My logic runs like this: what if the print-head developed a short-circuit (or nearly a short) between the data input pins of two different colors? The presence of a "spray" command on one color would tend to also cause the other color to be applied for every single bit shifted into the head array.

This would make the colors too strong and "muddy" which is exactly the description somebody else gave for the same magenta shift. That person had also noted that the problem came and went intermittently, but seemed to get worse over time. This would be consistent with a growing defect, where metal migration or contamination would cause an intermittent and increasingly conductive path between two nodes in the circuit.

But now the big question: why the difference between plain-paper and other papers? Why doesn't that point to the media sensor? The answer might be in this odd/even distribution of ink-spray points. What if the short exists for only an odd or even path? And what if Canon chooses to only spray using one-half of the ink-spray points when printing on plain paper? This would be a very logical thing to do, given that plain paper won't support the higher resolution anyway! If plain-paper prints only use the "good" half columns, one would never see the shorted extra ink-spray and plain-paper prints would look fine. But on good papers, all of the heads would be spraying and we will get the extra ink and muddy appearance.

I got the final confirmation today when I received a brand-new print-head in the mail and popped it into my #1 printer. The color shift was instantly corrected.

Note that the classic nozzle-checks will not show this issue since they are aimed at finding clogged print-heads. However, I intend to check to see if those patterns are showing a color shift as compared to a known good print-head. My suspicion is that the nozzle check shows good patterns, but in somewhat bad hues.

Anyway, if you have followed this far, I hope that you can get your printer going again, and don't need any expensive repairs. My adventures have been as long as this posting and much more frustrating. The problem can be intermittent (which might explain why I was sure that the media sensor bit was related) and Canon does not describe such a failure in their literature, since it does not fit the most common scenario: clogging or open-circuit failures due to thermal stress.

I believe the short-circuit scenario also makes some sense as a failure mode. The print-head sits there all of the time with wet ink adjacent to active circuits and heating resistor points (the thing that actually boils off the ink bubbles). The most common failures will be 1) clogging of the ink nozzle tubes due to drying and 2) burnout of the resistor heating elements when there isn't enough ink to cool them down or just due to thermal stress. A short will be much less common, so Canon probably does feel it is necessary to cover this lower-rate failure mode.

Good luck with your printer, and Best Regards,


Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow