I am interested in hearing more about your knowledge of the printer design that leads you to believe it could be improved upon in respect of the cleaning process.
Hi Mark,
To me, it's quite an interesting question: "What would I do if Epson were to retain me to address the inkjet clogging problem?". Let me start off with the disclaimer that I don't have any special knowledge of how Epson makes their printers, except for reading the field service manual for the 9900.
To my mind, corrective design changes are best made on the basis of observed behaviour, rather than on conjecture – although there's certainly a role for imagination in the process. I'd start by bringing in at least a dozen machines that had exhibited clogging during their service life, and I'd do a forensic teardown.
While the machines were still operable, I'd want to measure head and flow across the ink lines, and then measure delivery capacity, nozzle by nozzle, at the head. I'd want to quantify damper function too.
After that, I'd probably section the ink lines to quantify any pigment sedimentation, and I'd want to do some very careful cross sections of the heads, to determine the exact location of any clogs on a per-nozzle basis (for instance, is the entrance to the piezo pump clogged, is the clog at the exit portal, or is cavitation present?).
Without having done that, I'm in the land of conjecture. Here, I see two areas for potential changes: physical changes to the design, and firmware changes to the existing design.
Since it's easy to distribute firmware, and hard to distribute physical changes to the existing customer base, I'll just talk about potential firmware changes in this post (it will be plenty long enough already!).
Let's assume that we have a functioning ANC circuit. I don't know how sensitive it is, so let's assume that it's sensitive enough to detect a failure of x nozzles, where x lies between 1 and a dozen or so. I also don't know if the ANC circuit works by firing all of the nozzles in a channel at once, or if it works on a group of nozzles at a time. I also don't know the degree to which ANC operation can be modified by firmware.
My thought here is that once we have identified the presence of a clog, whether by using ANC or by the user reporting a gap in a nozzle check pattern, the first thing we want to know is exactly which nozzles have clogged.
As far as I know, all of the current cleaning cycles are based on firing every nozzle in a channel pair. This has a potential issue, which is identified in the service manual, which is that the nozzles can pump out more ink than the ink lines can deliver to the head. This can lead to negative head in the ink lines, which can suck air back into the nozzles, and is the reason that Epson suggests that you do some printing between cleaning cycles.
At this point, I don't know of any reason to fire the nozzles that are known to be working properly during a cleaning cycle – to my mind, it makes more sense to only fire the nozzles which exhibit clogs. Firing all of the nozzles is a waste of ink, and taxes the ink delivery system. There might be a good reason for firing all of the nozzles, but I can't think of one at the moment.
I would probably create a cleaning cycle which only fired the clogged nozzles, and only fired them until ANC detected that they were delivering sufficient ink. If there were a lot of clogged nozzles, I would probably clean them in groups, to avoid ink waste.
If ANC doesn't currently have the ability to identify which nozzles are clogged, I'd be interested to know if a successive bisection approach could be used to isolate specific nozzles - for instance, dividing the head into a top half and a bottom half, and then successively halving the section which delivered less link. If the sensitivity of the ANC circuit didn't allow for specific nozzle identification, I'd be interested to know if following the first bisection with a different bisection, like firing all of the even-numbered nozzles, and comparing with all of the odd-numbered nozzles, could further isolate which specific nozzles were clogged.
This might be a time-consuming process, but it could be done during idle time, and be interruptible to accommodate new print jobs. Here I'm thinking that the machine might be able to, with minimal ink usage, keep the heads unclogged all of the time, including periods of disuse.
One more firmware feature would be helpful, if it's not already present – a diagnostic log which identifies which nozzles have exhibited clogs during the service life, and how many nozzle firings it took to clear them. With the user's permission, this log could be transmitted back to Epson, where it could inform further design changes.