Anatomy of a Canon print head

[ghwellsjr]

This post will describe what I have learned about one of the most popular of Canon's print heads. It uses five cartridges. It is in the MP780, the iP4000 and the older i865, as well as many others. Here is a picture of the print head as viewed from inside the printer:

You can see the five cartridges including four dye BCI-6 inks and one larger BCI-3e pigment black ink.

A nozzle check for this print head looks like this:

What's interesting about this is that even though there are only five cartridges, there are seven sets of nozzle patterns. It's as if there were a photo cyan and photo magenta cartridge. The question is: how does the print head produce the lighter versions of the cyan and magenta?

A partial answer may be obtained by looking at the bottom side of the print head where the nozzles actually reside:

The taller set of nozzles on the right is for the pigment black. Notice the dark area between the two gold regions. This dark area is actually the ink inside the print head between two columns of nozzles. On the other set of nozzles are six similar, albeit smaller dark areas between pairs of columns of nozzles, one corresponding to each of the BCI-6 nozzle check patterns.

I took the print head into work where we have a binocular microscope. If you ever get a chance to look at a print head through a microscope, take the opportunity--it's really awesome. The surface of the print head is a clear epoxy with tiny holes that serve as nozzles. This clear epoxy allows you to see the electronic circuitry, the nozzle chambers, and the ink supply.

Here is a handheld photograph of a portion of the pigment black nozzles, taken through one of the eyepieces on the microscope:

You can clearly see the black ink and how it supplies the chambers just below the nozzles. I say "below" because the print head is held upside down during examination under the microscope, but, of course, when it is installed in the printer, the chambers are above the nozzles. The nozzles are the lighter color dots in the center of each of the dark "cul-de-sac"-like patterns that lead into the major ink supply going down the center. Note that the nozzles on each side of the ink supply are offset slightly so that the effective density of nozzles is doubled. Each column contains 160 nozzles that eject 5 picoliter drops for a total of 320 nozzles for the pigment black ink. The nozzles cover a distance of about half an inch so there are about 600 dots printed per inch. You can also see the smooth surface of the epoxy in the reflected light to the left and right of the electronics controlling the nozzles.

The dye based nozzles are similar except they are not as long, do not contain as many nozzles and are packed into a smaller area. They have a density of about 1200 dots per inch. The nozzles appear to be smaller than those for the pigment based ink. According to the literature, the minimum size nozzle for this print head is 2 picoliter. These would be on the sets of nozzles devoted to "light" cyan and magenta. All the others are larger although they may be smaller than 5 picoliters.

I tried to get some good enlarged pictures of the various colors in the nozzle check but except for magenta and black, they were too light to show up.

Here is a picture of the beginning of the photo black pattern:

If you have good eyes or if you use a magnifying glass to look at the nozzle check pattern for the photo black (6BK), you will see 32 rows of "dots". The microscope image above reveals that these "dots" are really made up of four drops, each of which is emitted by a different nozzle, two from the left column of nozzles and two from the right column of nozzles. (There are also some very small dots scattered around which I believe are not intentional and will be ignored in this discussion.) The top row of "dots" in the image is made from the top two nozzles on the left and the top two nozzles on the right. We could call these nozzles 1 and 2 in each column. The next row of "dots" are offset slightly and made from nozzles 5 and 6 in each column, followed by 9 and 10, then 13 and 14, and so on down to the bottom of the pattern. In other words, two pairs of nozzles are used, then two pairs skipped, then two pairs used, etc. If we add up all the nozzles used, it will be 32 times 4 which equals 128 nozzles.

Again, if you have good eyes or if you use a magnifying glass you can see a slight discontinuity one eighth of the way across the pattern. This is the point where the unused nozzles in the first eighth of the pattern start being used and the used nozzles start being unused. Nozzles 3 and 4 in the right and left columns now make up the top row of "dots" followed by nozzles 7 and 8, etc. This accounts for the remaining 128 nozzles for a total of 256 nozzles, and we are only one-quarter of the way across the pattern. In fact, you can see that the pattern is divided into eight segments with a slight discontinuity between them. It turns out that the odd numbered segments are identical to each other and the even numbered segments are identical to each other. If you have ever noticed an imperfection or what appears to be a clogged nozzle, it will appear four times across the pattern.

I think Canon has been a little deceptive in their presentation of the dye ink nozzle check because unless you look at the pattern under a microscope, you may be lead to believe that each of the 32 "dots" going down the pattern is just one drop from a nozzle and that the eight segments across the pattern multiply together to give you the 256 nozzles. At least it fooled me. Furthermore, since each "dot" is made from four different nozzles, a clogged nozzle may not even be noticeable since it would just make the "dot" smaller.

The yellow pattern is identical to the photo black pattern.

Here is the beginning of the dark magenta pattern:

This is similar to the pattern for the photo black except that the "dots" are made from two drops instead of four and there are 64 rows down the pattern instead of 32. However, the math still adds up: 64 times 2 equals 128 nozzles used in the first segment and another 128 nozzles in the second segment. Again, the first, third, fifth, and seventh segments are identical and the second, fourth, sixth and eighth segments are identical. Also, don't be confused by the very tiny dots you see scattered around; I believe they are unintentional.

The dark cyan pattern is identical to the dark magenta pattern.

Finally the beginning of the light magenta pattern:

The pattern for the light magenta is identical to the pattern for the dark magenta except that the drops making up the "dots" are obviously much smaller. These are supposedly 2 picoliter and the dark ones are supposedly 5 picoliter although the difference looks greater than that to me.

The light cyan pattern is identical to the light magenta pattern.

To summarize, this print head contains:

320 pigment black nozzles
256 photo black nozzles
256 yellow nozzles
512 magenta nozzles
512 cyan nozzles
------------------------
1856 total nozzles

 

[websnail]

Thanks for sharing that exploration into the world of small people ..

Think I may be bookmarking it when I hit any weirdness I can't explain..

 

[videobruce]

Should be a sticky or at least in some FAQ.

 

[ghwellsjr]

I would have liked to have posted an image of what clogs look like in the nozzle check for the dye based nozzles but that requires clogged nozzles. However, kato92 has provided such an image in this post:

http://www.nifty-stuff.com/forum/viewtopic.php?pid=8942#p8942
Link Broken... moderator

Scroll down to the third post and click on the image of the nozzle check. After it redraws in a window by itself, click on the new image and it will enlarge again. Study the bottom nozzle check for the 6BK. You will see that the black nozzle check pattern can be divided into eight segments going from left to right. The first, third, fifth and seventh segments are identical to each other. Also, the second, fourth, sixth, and eighth segments are identical to each other. This will always be the case, no matter how the nozzles are clogged. Of course, if none of the nozzles are clogged, the only indication of the segments is a slight drop in the even numbered segments compared to the odd numbered segments.

Thanks for the image, kato92, and I hope you get your print head working perfectly again soon.

 

[panos]

ghwellsjr, your posts are amazing

 

[ghwellsjr]

Thanks, panos, comments like that make it all worthwhile.

 

[Defcon2k]

This is really very interesting, thank you for posting!

I'm wondering about the following:


1. Are there different nozzles for the small drops and the large drops, or can the same nozzle produce different sizes? -> I guess the answer is A (different nozzles). This leads to next question:

2. How exactly are the nozzle rows arranged on the print head?
Are they arranged symetrically around Yellow to help bidirectional printing?

Or are they arranged asymetrically, like the 2.5pl nozzles on the left side, and the 5.5pl nozzles on the right side?

I'm interested in this, because it might help to understand the nozzle check print. Imagine you have a nozzle check print where "light cyan" is completely gone, but "dark cyan" has no problems at all.
What would it look like if you then press the print head on some soft tissue?
Would it still look normal like this:

Or would one of the two cyan lines be missing completely?



FYI: Quote from the service manual of the iP4000

Print head

Black:
320 nozzles in 2 vertical lines (600dpi), 30pl (pigment-based black)

Color:
256 nozzles in 2 vertical lines per color x 6 (1,200dpi), 5.5pl (cyan, magenta, yellow, dye-based black) / 2.5pl (cyan, magenta)

 

[ghwellsjr]

1. Are there different nozzles for the small drops and the large drops, or can the same nozzle produce different sizes? -> I guess the answer is A (different nozzles).

Correct, each set of nozzles on this print head are all the same size.

2. How exactly are the nozzle rows arranged on the print head?
Are they arranged symetrically around Yellow to help bidirectional printing?

Or are they arranged asymetrically, like the 2.5pl nozzles on the left side, and the 5.5pl nozzles on the right side?

I couldn't tell by looking at the nozzles which were the smaller and which were the larger for cyan and magenta, but they are not like your first drawing because, like I said for your first question, all the nozzles in one set are the same size. I don't think Canon is trying to optimize for bi-directional printing.

I'm interested in this, because it might help to understand the nozzle check print. Imagine you have a nozzle check print where "light cyan" is completely gone, but "dark cyan" has no problems at all.
What would it look like if you then press the print head on some soft tissue?
Would it still look normal like this:

Or would one of the two cyan lines be missing completely?

One of the two cyan lines would be missing completely.

FYI: Quote from the service manual of the iP4000

Print head

Black:
320 nozzles in 2 vertical lines (600dpi), 30pl (pigment-based black)

Color:
256 nozzles in 2 vertical lines per color x 6 (1,200dpi), 5.5pl (cyan, magenta, yellow, dye-based black) / 2.5pl (cyan, magenta)

Well, I never read that the pigment black nozzles were 30pl but they did appear substantially larger than any of the dye nozzles so that would explain why.

Remember, all the nozzles on both sides of each set combine to get double the resolution of each half.

 

[Defcon2k]

Thank you for your reply!

I couldn't tell by looking at the nozzles which were the smaller and which were the larger for cyan and magenta, but they are not like your first drawing because, like I said for your first question, all the nozzles in one set are the same size. I don't think Canon is trying to optimize for for bi-directional printing.

In the service manual for the S600 I read this:

Print head:
The two nozzle arrays for each CMY color are divided symmetrically left and right along the (middle) line. This makes it completely compatible with bi-directional printing.

and then it shows a schematic which looks like this: Bk(even), Bk(odd) | C1(even, M1(even), Y1(even), Y2(odd), M2(odd), C2(odd)

But frankly, I don't know how this is meant exactly, and if it would have any implications for the newer printers like the iP4000.

 

[ghwellsjr]

Now you've got me wondering, Defcon2k. I'm going to have to take another look at the nozzles under the microscope later this week.

I can tell you that the ordering of the colors in your drawings goes cyan, magenta, yellow, black, magenta, cyan and then pigment black.

Also, I do recall that when I had examined the nozzles under the microscope, I could see that the yellow and black nozzles looked larger and the outside columns of nozzles for the cyan and magenta on one side appeared larger than the inside columns of nozzles but on the other side they all appeared smaller. Here is a schematic:


Pig
Blk
O
O
O
Cyn Mag Yel Blk Mag Cyn O
O O O O o o O
o o O O o o O
O O O O o o O
o o O O o o O
O O O O o o O
o o O O o o O
O O O O o o O
o o O O o o O


I knew this wasn't the way it really was because I could see from the nozzle check that there were as many larger drops for cyan and magenta as there were smaller drops. I assumed that the columns that I should change to larger nozzles were the ones adjacent to the other larger nozzles so I concluded that the first four columns (for cyan and magenta) were all larger and the four columns for magenta and cyan on the right were all smaller, like this:


Pig
Blk
O
O
O
Cyn Mag Yel Blk Mag Cyn O
O O O O o o O
O O O O o o O
O O O O o o O
O O O O o o O
O O O O o o O
O O O O o o O
O O O O o o O
O O O O o o O




But from what you are presenting, I'm thinking that maybe the two columns that are larger are the right-hand of each pair for magenta and cyan on the right, like this:



Pig
Blk
O
O
O
Cyn Mag Yel Blk Mag Cyn O
O O O O o o O
o o O O O O O
O O O O o o O
o o O O O O O
O O O O o o O
o o O O O O O
O O O O o o O
o o O O O O O


I have an experiment which I will try in the near future to determine which of these is correct.

I have no idea what the nozzle pattern for the S600 is.