On ink sedimentation, agitation, clogs, and Hamlet

[cybis]

Update on the pure yellow: we have separation. There is a distinct layer on top with a volume of about 7%. Unfortunately, pure yellow is still too opaque to clearly observe what is going on below it. I'll run more tests next week.
 

[John Nollendorfs]

This looks similar to the 3rd party pigmented ink separation I saw 8 years ago from sitting for one year. But mine had a chalky looking upper layer.

But like a wrote before, the thing that the centrifuge won't show is the clumping of the pigment particles. Obviously you would see that type of result in the screen of the damper though!

[cybis]

...the thing that the centrifuge won't show is the clumping of the pigment particles. Obviously you would see that type of result in the screen of the damper though!

Great, now I need a (stronger) microscope. Thanks John

[deanwork]

I don't know about you guys, but I have been using Epson large format printers for about 12 years and I have always removed my ink carts and gently shaken them from time to time to assure pigment suspension, especially in times when I'm not running a lot of media through it. With the earlier models especially is was really the only way to go.

 My opinion is you should always do this regardless of the ink used. If you are in a strict production environment where you are going through several carts every month, the action of the printer is most likely sufficient to keep the ink compounds blended. But letting a printer sit for months with little or no use is inviting disaster, as Ansel Adams was fond of saying.

The only printer that I have ever owned that I would say does not require this is the Canon IPF series. That printer really agitates the carts automatically in a forceful way. Each time you put in a new cart it does it for a long time.  Also the carts are facing down and don't require a lot of excess pressure to move the ink into the holding tank and heads. But if the Canon was unplugged and left idle for a length of time even they might need to be manually mixed by shaking the carts by hand, though I doubt it considering how aggressive that auto shaking process is. But I would do it anyway. With the Canon you would probably end up with ink drying in the nozzles and ruining the head before the pigments became unsuspended with the base.

john

[enduser]

Also, with the Canons, the inks are "push-pulled" in and out of the carts from time to time, according to the service manual.

[cybis]

While waiting for equipment to arrive in order to test the inks further, I’ve been playing around with the dampers and I think there might be a problem with their design.

The dampers perform three functions:

• 1.   Filter the ink. I don’t think the filters are problematic, but on the x900 the filter ‘windows’ are on the downstream side (unlike the filters in Brian’s 7880 picture), so it’s impossible to see if there are particles accumulating there. The filters are also a lot smaller than on the 7880. I reversed the flow of ink over the filter to try to catch any particles in a paper filter but couldn’t see any, which is good.
  
  
• 2.   Damp the ink pressure. The thing I never realized until now is that the main role of the dampers is not just to damp, but to completely depressurize the ink before it reaches the printhead. The ink reaches the ‘high pressure’ side of the damper at approximately 4.5 psi of differential pressure (approximately 5 psi in the ink cartridges minus .4 drop from the height difference between the cartridges and the dampers). By a system of valve, spring, membrane, and suction, the absolute pressure on the ‘low pressure’ side of the damper is less than the atmospheric pressure; the pressure differential is slightly negative, somewhere between 0 and negative 0.2 psi. The printhead never receives positive ink pressure. That means the printhead needs to suck ink from the damper’s low pressure side. How does it do it? Capillary action is the only mechanism I can see doing the job. Specifically capillary action in the very fine piezo channels.
  
  
• 3.   Air trap. This is where I think there might be a design problem. The low pressure chamber of the damper is designed to trap air. When/if a mixture of ink and air enters the chamber, air rises to the top and ink sinks to the bottom. Ink is sucked by the printhead through an opening at the bottom of the low pressure chamber. Therefore only air free ink should be able to reach the printhead.
  
  Except that, in my view, there is a problem. Once there is a mixture of air and ink in the chamber, the air is trapped there forever! It has nowhere to go. The fraction of air to ink in the low pressure chamber can only increase over time during normal operation. When the printer is new, the ink lines, damper, and printhead are filled only with air. Priming the system, I assume, is achieved by creating a vacuum at the capping station. If the vacuum is strong, it will remove the air from the low pressure chamber before ink reaches it. When the ink reaches the vacuum of the low pressure chamber, it will completely fill that void. A new printer primed for the first time with a prefect vacuum will have an air free damper. If the vacuum was not perfect, ink will not completely fill the chamber and whatever air was left will be trapped in the chamber essentially forever. If the priming was perfect and the chamber was initially air free, but air subsequently finds itself in the ink lines, it will accumulate in the damper.
  
  Let’s assume that a power cleaning creates a perfect vacuum at the capping station, will it get rid of the trapped air? I don’t think it can. If the vacuum cannot propagate to the chamber, it cannot remove the air. I don’t think it can propagate there because there will always be at least some ink between the vacuum at the capping station and the trapped air. The depression caused by the vacuum will simply suck more ink in the chamber, which will sink below the trapped air in the low pressure chamber (even if the ink lines are depressurized first, ink will still flow in the chamber).
  
  The only way to remove the trapped air is to pivot the damper 180 degree to that the opening at the bottom of the low pressure chamber is now on top. This implies disassembling the head and sucking the air out with a syringe. Another way is to completely drain all the ink from the ink lines and dampers and thereafter performing an ink charge (praying that the capping station will create a strong enough vacuum). That would be a very costly procedure, several hundred bucks’ worth of ink down the drain.
  
  

Is air in the damper bad? I think so:

•   Air expands and contract much more than ink with variation in temperature and pressure. Let’s say trapped air in the damper warms up during operation. If the amount of air trapped is small, there might not be any associated problems with its expansion during operation as the opening at the bottom of the LP chamber is always covered with ink.  If there is a large quantity of trapped air, expansion and shaking might lead to air being sucked by the printhead and loss of nozzles during printing. In both cases, after shutting down the printer, as the trapped air cools it will increase the depression in the LP chamber. The more air, the more so. This might overcome the capillary action forces in the printhead piezo channels and draw air from nozzles.

•   An increase in atmospheric pressure after shutting down the printer will shrink the volume of trapped air, potentially drawing air in the nozzles.

•   It could be that the x900 are more sensitive to this problem than other Epson printers with the same damper design because of the smaller volume of the piezo channels which would empty more quickly.

•   I have to say I don’t fully understand the dynamic of air and ink inside the printhead. But it could be that once the ‘entrance’ of a piezo channel is exposed to air (the exit being the nozzle hole), the capillary action for that channel ceases. Since the pressure inside the printhead is less than the atmospheric pressure outside, air may be drawn inside the head trough an ink-free channels, causing nearby channels to ink starve as the air bubble grows; you would therefore tend to see groups of adjacent nozzles go missing. The only way to recover is to vacuum the air out with a cleaning cycle which would re-prime the piezo channels. There is a chance that the shaking and bouncing of ‘printing through’ missing nozzles might slash some ink on the entrance of a missing channel, but it’s more likely to just draw more air in the head.

Sorry for the long post. I'm sure I've missed a bunch of things but is the damper design problematic or not?

[Some Guy]

Interesting.

Wonder if one would be able to put a larger syringe, or some sort of vacuum tank with a quick valve, in the capping station's output hoses and do a quick suck on the thing with the head parked there to remove any air in the head?  The pump there may not be enough to do much of anything other than a weak capillary drain.  I also wonder if that pump alone is enough to do a service fill and purge too, maybe the air pressure is more there during that initial set-up step?

I agree that one way may be to put the head in a higher position so it can bleed out naturally, but how?  I could see a bit of air being pulled in back through the head once a ink cart is pulled too and any slight pressure is relieved.  I've switched out carts and had no ink for a while which makes me think air is there, but it somehow begins printing again.  Dunno.

I don't know what goes on with the silly MK><PK switch.  I got a really bad "No ink at all" in the blacks during a nozzle check doing that once.  Took a couple of days before it worked again.  Air?  That whole PK-MK idea needs to be scrapped and just find a better "universal black" ink.  Good to sell a lot of ink though.  Jon Cone recently wrote "that about 35% of the ink ends up in the maintenance cart anyway."  Seems a lot going down the drain.

SG

[cybis]

Here is a schematic of an ink damper.

[cybis]

Wonder if one would be able to put a larger syringe, or some sort of vacuum tank with a quick valve, in the capping station's output hoses and do a quick suck on the thing with the head parked there to remove any air in the head?  The pump there may not be enough to do much of anything other than a weak capillary drain.  I also wonder if that pump alone is enough to do a service fill and purge too, maybe the air pressure is more there during that initial set-up step?

One problem is that even if the pump created an instantaneous and perfect vacuum, there is a strong possibility that it would still not remove trapped air in the damper.

Another problem is that an abrupt drop of pressure at the head might damage the extremely fragile piezo capillary channels. I think a slow ramp up is needed to give air a chance to escape peacefully. The printer's vacuum pump should be sufficient.

As a thought experiment to help think about trapped air, imagine the following setup: a reservoir filled with ink (similar to a cartridge), connected to a chamber containing trapped air (like an ink damper), and connected to a vacuum (such as the vacuum at the capping station). Between the reservoir and the chamber there is a variable flow restrictor or a valve (absent in the actual printer). Between the chamber and the vacuum there is a fixed flow restrictor (like our capillary piezo channels).

Now, what happens as we close the variable flow restrictor?

With the variable flow restrictor fully open the trapped air is barely affected.



With the variable flow restrictor in an intermediate position the trapped air expends but not enough to actually escape the system.



With the variable flow restrictor closed the trapped air can finally escape.



I haven’t discovered anything yet in Epson design that would significantly restrict the flow of ink from the cartridges to the dampers during a ‘vacuum cleaning’. There is definitely no shutoff valve.

 

I agree that one way may be to put the head in a higher position so it can bleed out naturally, but how?  I could see a bit of air being pulled in back through the head once a ink cart is pulled too and any slight pressure is relieved.  I've switched out carts and had no ink for a while which makes me think air is there, but it somehow begins printing again.  Dunno.

Power cleaning with the printer upside down

I don't know what goes on with the silly MK><PK switch.  I got a really bad "No ink at all" in the blacks during a nozzle check doing that once.  Took a couple of days before it worked again.  Air?  That whole PK-MK idea needs to be scrapped and just find a better "universal black" ink.  Good to sell a lot of ink though.  Jon Cone recently wrote "that about 35% of the ink ends up in the maintenance cart anyway."  Seems a lot going down the drain.

It could be air. There is a couple more potential air traps in the K switch.

[cybis]

Here is a short clip showing what the bottom of centrifuged green ink looks like magnified 600x: https://youtu.be/wDTTMrqU3Co
Note the Brownian motion of pigment particles when they are not clumped together and the absence of motion when they stick together.
(The video was shot with a smartphone through the ocular of the microscope, DSLR adapter is on the way  )

[John Nollendorfs]

Interesting stuff you are doing here Cybis! Thinking of designing a new printer too? ;-)

[cybis]

Interesting stuff you are doing here Cybis! Thinking of designing a new printer too? ;-)

  John, those missing nozzles are really ticking me off and I just like to know my ennemies...

[cybis]

I was pretty excited this morning to see the result of an experiment I ran overnight.

I replaced a printhead channel with a single glass capillary tube, filled refillable cartridges with colored distilled water, pressurized the cartridges at 5 psi. I then drew the colored water through the capillary tube and depressurized the cartridge thereafter.

An actual printhead contains 360 ‘capillary tubes’ about 50 x 50 microns, creating a suction force of about 0.5 PSI. The dampers open at approximately 0.05 PSI, so the capillary force is plenty enough to suck ink from the slight vacuum created by the dampers. Note that a single printhead nozzle also ‘sucks’ with a pressure of about  0.5 PSI.

The smallest glass capillary tube I could put my hands on was 100 micron in diameter by 100 mm in length. The suction created by the glass tube is about half of a printhead nozzle: 0.25 PSI.

I also experimented with a 400 micron plastic tube that exhibited essentially zero capillary forces.

When I drew water through the plastic tubing alone, the water would recoil back towards the damper as soon as suction was removed. This demonstrates neatly that there is indeed a slight vacuum in the damper and printhead.

When I drew water through the 100 micron capillary tube, the water would fill the glass tube on its own even without any suction being applied. Suction was needed to draw ink to the entrance of the tube, but once the liquid made contact with the tube, capillary force alone was sufficient to draw the liquid from the damper and into the remainder of the tube. Even with the cartridges depressurized the water would remain in the glass tube, at least for the first few hours…

When I checked the glass tube this morning, about 15 hours after filling it, it was empty. The water was sucked back in the damper! Why? 

The low pressure chamber of the damper happened to contain about 50% of trapped air. Overnight, the temperature in the room dropped by about 4 degrees Celsius and the atmospheric pressure increased by about 5 hPa. Could it be why?

It could be a fluke - I need to repeat the experiment with and without trapped air a few times. But if the glass capillary tube is a good proxy for how the printhead behaves, and if further test runs confirm the first result, I think this is a good candidate for the main mechanism causing Epson x900 nozzles to go missing.

 

[cybis]

Picture of the experiment.