Clogged Canon print head

[ghwellsjr]

This image shows a surprising result - the 5 pl and 1 pl nozzles are located side-by-side in the same bank of nozzles. There are two duplicate nozzles sets for magenta and cyan, giving 256 large and 256 small nozzles for each color. I guess that this makes sense, since it keeps the same ink flow in each channel. The "heaters" are the same size for both sets of nozzles. Note that the 1 pl nozzles are located between the 5 pl nozzles.

Wow, Grandad35, awesome work!

Check out this thread for an in-depth discussion of the issue of why the nozzles on each side of an ink channel contain two different sized nozzles:

http://www.nifty-stuff.com/forum/viewtopic.php?pid=10554#p10554

I think our brilliant contributor Defcon2k figured it out, if you read all the way to the end, although I haven't had time to investigate it further to determine for sure if he is right.

This is done to overcome an unavoidable problem in inkjet printers which is that when a nozzle is fired one time, the ink comes out in about three droplets each smaller than the previous one. If the nozzle were aimed straight down, it would create three separate dots, but by aiming the nozzle at an angle back away from the direction of travel, the drops have no horizontal velocity and end up on top of one another, creating a single dot on the page.

Now by having two sets of nozzles, one aimed in one direction and the other aimed in the opposite direction, you can fire one set when the head is traveling to the left and the other set when the head is traveling to the right, thus giving you nice clean dots.

The nozzle check is done with the head traveling in just one direction so half the firings result in lots of extra smaller dots appearing on the page.

 

[Grandad35]

ghwellsjr,

It was good to reread your previous posts on the subject of the small/large sets of nozzles.

I should have specifically noted the orientation of all 4 sets of the 1 pl nozzles, but can't make a definitive statement until I get a chance to look again under the microscope. My memory is that all 4 sets of the 1 pl nozzles were on the same side as the pigment black nozzles, and were not placed symmetrically about the center.
8/10/07 Edit
My memory was incorrect - the smaller nozzles are symmetrically placed, and always toward the middle of the print head.

FYI: Quote from the service manual of the iP4000
snip...
Black:
320 nozzles in 2 vertical lines (600dpi), 30pl (pigment-based black)

This spec isn't for the ip5000, but the 30 pl value is consistent with the larger black nozzles and heaters. Using the vertical blue bar (1/600" = 42 microns long) as a scale, the black nozzles are about 25 microns in diameter, the 5 pl nozzles are about 12 microns and the 1 pl nozzles are about 6 microns.

From what I read, the concept of each nozzle shot generating 3 drops of ink was only shown in Defcon2k's animation:

Was there other documentation for this phenomenon that I missed?

 

[ghwellsjr]

It wasn't based on any documentation, it was a conclusion based on these photographs:

http://www.nifty-stuff.com/forum/viewtopic.php?pid=10642#p10642
http://www.nifty-stuff.com/forum/viewtopic.php?pid=10686#p10686

In the first photo, all the nozzles are untaped so you see the result of both sets of magenta nozzles firing during a nozzle check. Notice that there are lots of extra smaller dots scattered around. If you look carefully, you can see that the larger dots cluster in pairs and the lower of the two in each pair has one or two smaller dots to the right of it.

In the second photo, one set of nozzles is taped over so now you are seeing only the upper dot of each pair in the first photo. There are no extra dots.

The last photo is with the other set of nozzles taped. Now you can see that each nozzle produces one or two extra dots and none of them is a clean single dot.

The question is: why do the nozzles in one set produce a clean single dot on the page while the nozzles in the other set produce extra dots?

Defcon2k suggested that this had to do with the direction of travel of the printhead and I suggested that this could be explained if the nozzles in each set were angled to eliminate horizontal velocity of the ejected dots for the correct direction.

This has nothing to do with symmetry around the pigment black nozzles. It really has nothing to do with any kind of symmetry. It probably has to with the way the nozzles are manufactured. Possibly it is easier to angle all the nozzles (half of them larger, half of them smaller) in the same direction on both sides of a given set of nozzles than it is to put all the larger nozzles in one set, half angled one way and half angled the other way, and all of the smaller nozzles in the other set, half angled one way and half angled the other way.

This is pure speculation and I had intended to run some more tests to confirm this but I just haven't had time. I was hoping to be able to produce some kind of sparse printout of one color where the individual dots could be seen and then print it with different quality settings at different speeds and both bi-directional and uni-directional. If necessary, I would cover up one set of nozzles. My hope was to see some printouts with no extra dots and them some with extra dots because if this theory is correct, it would only work at one speed and in the proper direction.

It might also be possible with really good microphotography to determine if the nozzles are angled.

 

[Grandad35]

Ghwellsjr,

I ran some tests on my i9900 (768 nozzles, 2 columns of 384 for each color - 3x as many as the ip5000). The nozzle checks look like this:

As before, click on the thumbnail to get a larger image, then click on that image to get an even larger version (1200x1200 pixels). It looks like there are 4 individual (and approximately equally sized) ink drops in each cluster, with the spacing of the clusters being 4 nozzle spacings (4/600 inch). I don't see any evidence of extra drops in this Cyan scan (the magenta and black scans looked the same).

These images were scanned at 12800 pixels/inch, then resized to 12000 pixels/inch to get exactly 20 pixels/nozzle spacing on each image, with each image representing a 0.1 x 0.1 inch section of the print. The light "levels" were also greatly enhanced to bring out the relatively light original I opened the cover while these tests were being printed, and every nozzle check was printed from left to right.


This is a portion of the extended nozzle check, scanned at the same scale. A portion of the previous nozzle check was superimposed on the right side to make it easier to make a comparison. The horizontal bars are spaced every 32 nozzles, then every nozzle is fired (you can see 32 distinct ink dots printed in each vertical bar - 2 columns of 16).

Again, there is no evidence of extra ink dots in this area. For some strange reason, every extended nozzle check is printed from right to left - exactly the opposite of the nozzle check.

These tests were printed on Kirkland Photo paper, in case that makes a difference.

 

[Grandad35]

It might also be possible with really good microphotography to determine if the nozzles are angled.

I decided to put Zack23's print head under the microscope again before proceeding with the destructive tests. This animation was generated from several shots where the only difference was that the microscope's stage was moved down (away the lens) between successive shots (there was no lateral or vertical movement of the print head between shots). At the start of the animation, the focal plane was low enough that nothing was in focus. As it progresses, the bottom of the nozzle comes into focus, as well as the circuit layers. Further along, the exit of the nozzle comes into and then out of focus. I wish that I could say how much the stage moved for this sequence, but I have no idea. Finally, on the last shot, two circles were added showing the size of perfectly round 1 and 5 pl ink drops (these are 5 pl nozzles). The ratio of the drop diameters is about the same as the ratio of the nozzle diameters (shown in a previous post).

There have been a number of instances reported where one or both sides of a set of nozzles fails for exactly 1/2 of the nozzle. It has been theorized that this might be caused by an electrical failure. This opportunity was taken to look at the circuitry at the midpoint of the nozzles for anything that might explain this failure mode. The following 2 thumbs (click on them for a larger image) show the right and left edges of the pigment black nozzle set (160x2 nozzles) at the midpoint. Both sides show a break in one of the main traces at the midpoint, and a connection with two external traces from that point on.



The following 3 thumbs were taken from the dye ink nozzles. This image shows a break in some of the traces at the midpoint (this is the only place where this happens on the color nozzles).

This image shows another difference in the design that happens only at the midpoint. There is a discontinuity in the design at this point, with a slight extra space being added.

This image shows a discontinuity in the traces on one edge at the midpoint, similar to what was seen on the pigment black nozzles. There was no similar break seen on the other edge.

 

[ghwellsjr]

Grandad35, again, awesome work.

I don't see any evidence of angling of the nozzles, is that what you concluded?

 

[Grandad35]

I don't see any evidence of angling in the nozzles, is that what you concluded?

Yes - neither the 1 nor the 5 nor the 30 (pigment black) picoliter nozzles showed any indication of being angled.

In the last post I forgot to mention that the 1&5 pl nozzles ARE located symmetrically about the center of the print head, with the 1 pl nozzles always being inside of the 5 pl nozzles.

I should have also drawn the conclusion that there is something happening at the center of each column of nozzles indicating that (probably) the power supply to the nozzles is split at this point. This could explain the loss of exactly 1/2 of a column of nozzles as being an electrical problem in the print head. Unfortunately, such a failure in a power trace also has the potential to damage the mother board, so a new print head is not a guaranteed fix for this problem.

While looking at the animation again, I noticed something that I hadn't noticed before - that active etch extends around the nozzle, but at a deeper level than the other etch. Being deeper, it is out of focus on the other shots, so it wasn't obvious.

It is possible that this is the actual heater, and that the device previously described as the heater could be a power transistor. Perhaps someone who knows what these devices look like could chime in on this.

 

[ghwellsjr]

Maybe having the heaters on one side of the nozzle could impart an angular direction to the firing of the drop. Are the heaters for the corresponding column of the same color on the other side of the nozzles?

 

[Grandad35]

Maybe having the heaters on one side of the nozzle could impart an angular direction to the firing of the drop. Are the heaters for the corresponding column of the same color on the other side of the nozzles?

I didn't look at the circuitry UNDER any of the other nozzles, as I didn't even notice that it was there until after the individual "depth" shots were combined into the animation. However, the two columns of 5 pl nozzles for each color are located symmetrically about the center of the print head. The circuitry beside each nozzle is also symmetrical, so we can probably assume (always a dangerous word) that the heaters are also symmetrical, even if they are under the nozzles . If this system is designed to produce an angular ejection path, the two columns of nozzles should have opposite ejection angles.

I should also mention that my previous printing tests were on an i9900 that produces 2 pl droplets. The surface tension forces divided by the droplet weight are lower on a larger (say 5 pl) droplet than on a smaller (say 2 pl) droplet, so large droplets may be more likely to generate "dwarf twins" than small droplets. Just food for thought.

 

[berryrice]

Hello.I just picked up a i9900 super cheap as the old owner said the printer was just printing out lines. So I thought for 50 bucks it was worth a shot to attempt to clean the head as I've done with my i860,i960, and MP780 without a hitch. So after reading the 12 pages of postings and trying different methods I'm just about ready to give in and buy a new print head BUT I want to make sure it is the issue. The i9900 "sounds" and tracks properly but give out next to nothing on the printout, just a few pink and blue lines.....just enough to create a silhouette of the picture (about 90%+ is white). So I think I have either a MASSIVELY clogged head or something more is wrong with the printer its self. Anybody have any suggestions or other issues with the i9900 that may cause what I'm describing before I shell out for a new print head. Thanks in advance!