Epson 7900 from the inside - out

[Eric Gulbransen]

At this point, I'd be very surprised if anyone at Epson seriously thought it was "time" for them to do anything of the sort on this particular case...

I'd like to make myself clear on this - I do not hold Epson responsible, I do not hold Dan responsible, I do not expect or even hope that either of them should come to my rescue.  In fact Dan offered and I denied him.  It's a used printer, it's out of warranty, there is no service contract, I had it shipped across the country, AND I worked on it.  At this point I may as well have built this printer out of spare parts from Stanford's robotics classroom down the street - I am on my own.

What would be great though is someone with lab-coat-like Epson knowledge chiming in to help us better understand what we are all up against considering these X900 heads.  Because as time goes on we are only going to need more knowledge and more help.  Could be it's impossible for them to nail the cause of our specific problem to the wall, but I bet it's very possible for them to eliminate some of the ideas which we have considered, but they likely know are non-sense. 

Like Mark suggests, I am sure they have angles to consider here which some of us can't even imagine.  So be it.  Maybe we'll never hear a word.  But it sure would be helpful.

[Wayne Fox]

Based on past LuLa readership, this thread will probably hit 30K+ views, that is too big to ignore...

While I agree this site is certainly one of the most visited photography sites on the internet, for good reason, are those "unique" visitors or are they just visitors?  (meaning I count as 30 or 40 of those visits since I've checked this thread once or twice a day since it started, just like many others).

It still does represent a pretty large group of interested parties, but all it has really shown is a few users out of thousands have had head failures which may not be any larger of a % than is typical of earlier epson printers, just means no one has done this extensive of a discussion about them.

I will admit it does seem strange, as head failures in most other epson printers seem rare (they just seem to run for years), and the LLK channel does seem to be the most common channel affected. Still it's an extremely small percentage.

This has been a very interesting thread and I think Epson should perhaps look at doing some changing, for example I see no reason they shouldn't recommend replacing the wiper every 12 months.  It appears easier to change than the cutter blade.

[gwhitf]

This has been a very interesting thread and I think Epson should perhaps look at doing some changing, for example I see no reason they shouldn't recommend replacing the wiper every 12 months. 

They have nothing to gain, and everything to lose, by even acknowledging any issue. I doubt you'll ever hear a peep out of Epson.

[Mark D Segal]

They have nothing to gain, and everything to lose, by even acknowledging any issue. I doubt you'll ever hear a peep out of Epson.

I don't think that's the operative consideration. My sense of it is that if they were of the view that the wiper blade should be a user-replaceable consumable like a cutting blade, inks or the maintenance tank, they would have provided clear instructions in the manual for doing so. Of course nothing prevents them from rethinking this down the road, but I imagine if they were to do so at all, it would be on the basis of accumulated world-wide performance data on the machines.

[Eric Gulbransen]

You're not the only one Wayne.  My genius buddy Steve reads this thread every day as well.  But I'm starting to wish he didn't..

We bought this 7900 together, Steve and I.  We set it up at my house next to my heat press machine.  Oh the printing dreams we shared.  He even brought his rolls of paper over.  But it's been a couple of months by now with nothing good come out of this machine yet.  So last night after my genius buddy came over for dinner, he left with all his paper. 

...heartbreaking.  I think he's lost hope.  And he's way smarter than me so if he's lost hope, oh brother that's not a good sign. 

I don't know a lot but I do know our 7900 will print again.  Sometimes I think it's an advantage not being so smart. 

[chaddro]

If you google: "Epson Micro PiezoTM Print Head Technology" you'll find an Epson Published white paper on the new print heads. Although it's really an advertising piece for marketing the new print heads, the comparison to the old 9800/7800 heads is interesting. They tout how much better the new generation heads are by showing the relative size of nozzle and chambers on page 8.

A single piezo layer 1 micron thick vs 16 layers 25 microns thick. Hmmm... I have no idea how tough these new heads are but it makes you wonder.

That tiny chamber make me wonder if despite the back flushing that Eric did on the head, that ink was still plugging said nozzles and that the clear nozzles were simply allowing for the back flush... and the possible dangers of doing just that... 1 micro is thin? Wow. That head may have needed DAYS to soak in the proper cleaning liquid. Unless, of course, there was some mysterious electrical failure.

[clic]

Could you please post the link to that paper, that sounds interesting, and when I google this, there are quite a few docs that pop up, and are not the one you seem to mention.


Thanks.

[chaddro]

Hope these links paste correctly. First is the consumer version, second more scientific published by the designers of the heads.

http://pos.epson.com/lp/epson-color-works/pdf/Micro_Piezo_White_Paper.pdf

http://www.imaging.org/ist/publications/reporter/articles/Rep23_1_NIP2007_OKUMURA.pdf

I'm sure there is more out there if you know how to dig for this stuff. The second article is much more interesting from an engineering point of view.

[Eric Gulbransen]

Chaddro - the google mercenary of hidden resources

[clic]

Thanks Chaddro.


It seems that it would be conceivable that if some ink dried in the upper chamber, it could then flake and go obstruct the path of each nozzle's chamber, possibly following the non random pattern we have witnessed all across the known incidents.  The problem that is stubbornly defiant with that theory is why then would that occur predominantly in LLK channels?  Especially when LLK seems to be thinner and less prone to clog than most other colors.

Of course even if some ink dried in the upper chamber, then why does it dry there?  It is not supposed to, air is not supposed to be present there.

[clic]

How many times do I need to repeat the same thing?  Out of all the post on the Epson LF forum regarding clogs on x900 printers, and going back on Lula to June 2011, plus various other networks, I have found:

* 11 issues of "clogging" resulting in head needing to be changed.

* 9 of them were about problems with LLK.  Eric is one of the two others.  And for the record there a couple of other cases presumably caused by LLK that I do not count, as the users never contacted me back.


That's what pollsters would qualify as a "valid sample," and while there are undoubtedly many more users who have experienced the issue than the ones I have found, there is no credible reason that would explain that my sample is "biased" toward people not "liking" LLK.  Given the relative and unhelpful stubbornness, let me chew it down for you: I understand that people with issues (I am not talking of people with intellectual issues that would prompt them to want to appear as more scientific than their neighbors for instance.  I am talking of people having technical issues) are more prone to post or just be present on the forums, but nothing would explain that out of all the people who need a head replacement, the LLK ones would be overwhelmingly more present than the other ones combined.

If after this you are still not convinced, you might want to check the definition of "denial" in your favorite dictionary.

[Ernst Dinkla]

Of course even if some ink dried in the upper chamber, then why does it dry there?  It is not supposed to, air is not supposed to be present there.

The ink feeding system is no longer based on a siphoning effect like available in the 9000 - 9600 generation. Ink is delivered from carts that are below the head(s) in most of Epsons wide formats. When switched off there is no pressure on the air around the cartridge ink pouch and the ink flow backwards into the carts is prevented in 3 ways (based on my old 10000):
1/ a small backflow preventing valve in the cart (makes refilling difficult for the same reason)
2/ an electronically activated valve just after the cart slot that is closed when the printer is idle or off power
3/ the capping station sealed against the head nozzle surface and the waste ink tube pump seals the other side

Of the three I have the least confidence in the capping station seal, the cart backflow valve comes second.

While the electronically activated valve may be enough to withstand the ink pressure down in the channel there are other physical phenomena that could retract ink from the nozzles into the channel when the printer is off power:
1/ changed atmospheric air pressure could already have that effect
2/ changed temperature affecting a different thermal expansion of the channel length of ink to the channel length of tubing

The damper membrane could prevent part of the mentioned effects but it becomes a delicate balance.

When on power I can imagine that there is some prevention with sensors and/or scheduled ink flows to let this not happen but off power and worse on transport it will be much more difficult. Taking the carts out does not change that, it is about the inks already in the channels.

On transport, a pallet forklifter in a warehouse can do a lot to the ink gravity that valves ot seals will not control, up or down. Depends how macho the operator is and his coffee break near or not. That box does not have to drop off the pallet for the effect.


On statistics: I would like to see an independent poll with unbiased questions on a randomly selected group of 3000 users, throwing in some odd questions too, before I would make any assumption on what the most affected ink channel could be.


met vriendelijke groeten, Ernst
Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/

[clic]

Ernst,


And I'd like to be able to go on week ends on the Moon!  At least once in a while.

The fact is that polls on say elections, are currently done on panel samplings sometimes as small as 500, rarely much larger than 1000, and they can still predict the result quite accurately on average.  I have no idea what's the size of the sampling I am pooling data from.  I would have to guess how many people visit the forums I have researched, deduct from that mass how many would not post their problems unless threatened of examination by the McCarthy commission, add how many people are reached by the networks I pool from, deduct there again the "McCarthy" factor etc...  But my barely educated guess is that altogether, that "pool" is anywhere between 500 and 1000.  IF that is true, then my stats are perfectly revealing.

Other than that, your analysis of how air could get in the channel in pretty convincing, the question is then, would that be enough to thicken the ink so that head failure occur?  I am not convinced of that, because in my testing, which have been verified by somebody testing independently, LLK thickens, but does not harden just by contact with the air.  It is therefore beyond unlikely that it would flake.  Then the cleaners that I have used inside my head, and probably the cleaner that Eric used, would definitely have cleared that.  It therefore does not seem very probable that dried ink is at stake here, but of course, all I just wrote is not a scientific demonstration, just an educated guess.

[Randy Carone]

clic,
Do you have 1000 respondents or are you drawing from a pool of 1000 possible respondents? You are also limited to 1. folks who read forums and 2. folks who read forums AND are willing to answer a poll. I also suppose (unscientifically) that those with problems are much more likely to respond to a poll about problems. I think your poll is dreadfully skewed to the negative.

[Mark D Segal]

Actually over the years, polling has become a highly scientific undertaking based on the professional application of rigorous statistical methodology. Unfortunately, with the advent of web forums and the internet, it has become all too easy to cull or solicit information and then cite it is a "poll". No one trained in this field or familiar with its methodology would consider much of what passes for polling on web forums to be of any real value.

[John R Smith]

Folks, Epson would just not still be in business if the generality of its customer's experience was as black as it is being painted here. Especially with these big, very expensive professional printers. So I am inclined to think that we have a pretty small sample.

John

[Mark D Segal]

I am more than *inclined* to think.........

[designpartners]

I am more than *inclined* to think.........

  funny.. Agreed!

[clic]

Randy,


I don't now how many people there is in the pool of people qualified in my previous post, but as I wrote, my guess, for what it is worth, is that we are talking of probably 500-1000 folks.  Out of that number who could and probably would have posted or manifested themselves, had they had any head failure on a x900, I have identified 11 cases.  I have also heard of three other cases, but have more than one degree of separation with them, and therefore I do not retain them as valid enough.

So for the nth time, yes, we are talking of a small fraction of users.  It is difficult to know how small a fraction given the bias of the methodology as well as the non negligible amount of people experiencing the issue under warranty and therefore non reporting to me, that's the other bias of this study.  My estimate, once again for what it is worth, is that it seems that we are talking of something around 1% of users, maybe slightly less.  That is not a whole lot, but given the severity of the situation when that happens to you, that is no joking matter.

A little bit like if it became known that you have a 1% chance to die tomorrow, randomly, from an unknown killer.  1% then start to be far too many, and you'd want to get the guy before he gets you.

[Jim Coda]

Clic,

As I recall, you got most of your data on which channels clogged on the Yahoo wide format group.  Why don't you take the opportunity to do a poll here, using a new thread so people can't miss it, asking people who have (or had) 4900, 7900 and 9900 printers that clogged (despite cleanings) to specify which channel or channels they were? 

I don't know how many people will respond, but any additional data you get would seem worthwhile whether it supports your hypothesis or not.