It's Dead, Jim

Short version: after nearly a decade of service, my 7900 has reached the end of its life. Printing had become increasingly challenging in recent months; clearly some of the machine's components were no long operating up to par. But I was always able, with various levels of effort and wasted time and ink, to get good  nozzle check on all channels. That is no longer true. The longer story follows.

On 3 May I prepared a couple of image files for printing. One would be and black & white image on an 11 x 17 inch sheet of Canson IFA Baryta, using one of my last remain sheets of this discontinued paper. It's been a favorite for years. The second would be printed larger, on a 17 x 22 inch sheet of Epson Exhibition Fibre. I prep'd the files in the usual way, and then began what I knew would be an ordeal getting the printer ready. I'd last printed in mid-March, about three weeks earlier.

The first nozzle check print showed showed all channels at 100% except VM, which had a small missing nit in a corner, and minute "displacements" or deflections on the sides of the LK channel. I first noticed this displacement back in December, 2020, but I could see no obvious issue with the prints, so I'd come to ignore it on later nozzle checks. If I could get all nozzles firing in all channels, I'd continue with printing and get the usual excellent results. I ran a standard cleaning of the C/VM pair, did a second nozzle check, which showed problems with O, LC, LK, and LLK missing entirely.

Before I finished I'd run a total of 15 nozzle checks and print purge files (which prints a business-card-sized patch of a single ink) for several channels, over the space of three days, in total consuming about six hours and leaving me very cranky (don't ask my wife). I removed and cleaned the wiper (which needed it, as it turns out). I had to replace the maintenance (waste ink) tank and two ink cartridges; fortunately I had a new ones on-hand. When I'd start to feel like I might be a danger to myself or things around me, I'd shut off the lights and walk away, hence the three days. My normally low blood pressure probably wasn't for a lot of this period.

In the end I had a good nozzle check. Except for those LK displacements, as shown below, with the second scan enlarged for detail.

  

As mentioned, I'd been seeing this for a while, and based on past results didn't consider it a problem. As it turns out, I was quite wrong. Below is the first print I made, of the B&W image.

Those bands pointed out with the tiny red arrows are about .75 inches (2cm) from the printed edges, with the image being close to the edge of the paper. This sent me down the hours-long gopher hole of troubleshooting the problem.

A nozzle check looked fine (except for the LK displacements). I made a few partial test prints on junk paper and learned little, with one interesting exception: If I made a small print in the center of a larger sheet, such that the unprinted borders were several inches wide, the print looked fine. I've thought long and hard about this and have come up with no reasonable explanation, especially given the root cause of the problem (keep reading--I'm almost done).

Finally, I printed some purge files for the black and gray channels, PK, LK, and LLK. PK and LLK looked fine, but LK showed a wide band, as shown below.

 

It seemed pretty clear that the displacements in the LK channel had progressed to the point at which quality prints were no longer achievable. Looking back through older nozzle check prints I found the first hint of these displacements in late 2020. Comparing that to a current check it's clear the deflection has increased. My theory is that when tiny, whatever impact the displacements had on print quality was covered by later passes of the printhead laying down ink over the area where banding happened. Just a WAG, but I can't think of another explanation, and in the end it's irrelevant anyway.

Since I had nothing to loose (but a little time) I ran through the head alignment procedure one normally does when setting up a new machine (or when banding issues arise). This had no impact.

Just a sampling of head alignments, failed prints, and other tests.

So, that's it, the end of Life With A 7900. I'm not happy about it, but I'm feeling surprisingly serene, maybe even a little bit relieved. The machine produced first-rate prints, and a lot of them. It paid for itself many times over, although it often wasted more of my time than perhaps it was worth.

I have mixed feelings about Epson; the 7900 punished me for many of the prints I made. There were buggy drivers (which got fixed), a somewhat weird UI (which I got used to pretty quickly), more wasted ink than I care to recall. I understand some, perhaps most, of the hardware issues (clogging or other ink delivery problems) never happened with later generations of the machines: a friend with a P7000, which does not see frequent use, doesn't even bother to run nozzle checks; he just prints, and the results are excellent every time. With later models Epson clearly fixed many of the x900 series's issues, proving that nothing is so screwed up that it can't serve as a bad example.

I've got a very busy year ahead of me and I'm not feeling any immediate rush to replace the 7900 (although a couple of my regular clients have expressed a hope that I do). I have a lot of research to do on the path to choosing my next printer. As a former U.S. President and "reality" show host said all too often, "We'll see what happens." But one thing is certain: I won't be blogging about it!

If you've followed my increasingly infrequent (and irrelevant) postings here, thank you. I hope you found them useful, and failing that, maybe at least a little bit amusing.

  --Jay

The Nasty Nines (Years, That Is)

 It's been a slow summer and fall here, during which I've done very little printing. The 7900 sits idle most of the time. Such conditions made the printer unhappy even when it was new. As things deteriorated with age, my now nine year old machine has become increasingly cranky. This ever more irrelevant record of my experiences with the 7900 is full of examples of this behavior. Here's the most recent:

Last week I had a job to make 14 small prints on mat paper for a client I'd not heard from since spring. I had no trouble with the prep work in Photoshop, just typical adjustments and then resampling and sharpening for printing. But I knew a nightmare of nozzle checks, cartridge swaps, and cleanings awaited.

I last printed on 31 October, but that was a bit of a cheat: I needed to print some CAD drawings of house plans on 24 x 36 inch sheets, using only black ink. I didn't bother with a nozzle check, just opened the drawing files, fed the paper, and printed, with a fine result. But I knew a nozzle check would show LLK, a constant problem channel, completely missing, and very likely some of the color channels in bad shape. I had no need to deal with any of that, so I didn't. Bottom line: the last time I needed to make photo prints was in September, which means it'd had been nearly three months since I had done the work to get a 100% clean nozzle check print. Oh, boy....

And of course, the Epson punished me for my lack of devotion and attention. As usual, LLK was completely absent, as was LC. Other channels had plenty of missing nozzles. Also as usual, as I worked through cleaning pairs, some channels outside those pairs disappeared. In the end, after about an hour of cleanings, powerful cleanings, and ink cartridge swaps (replacing low-percentage ones with new ones for the cleaning process) I finally had the machine ready to print. It's still possible to get all channels 100% "clean."*

After all that, the printing went as expected, the prints are lovely, and the client is happy. And I got there without bringing out the dynamite, although it was a close call.

*They don't stay that way more than a day or so. I suspect the 7900's capping station no longer seals properly when the print head parks, but that's only a guess, and I don't plan to tear down the machine to find out.

  --Jay

A Couple of Weeks Later

Almost two weeks after the nasty experience described in my last posting I powered up the 7900, and then set up as usual to make a single, small print. The nozzle check showed most of the LLK channel missing. I did a standard cleaning of the Y/LLK pair, after which a nozzle check showed 100%. The previously-problematic LC pattern was 100% both before and after that cleaning.

The machine printed normally, with none of the slow-down/delay described in that 16 February posting. The print looks as it should.

Just another Stylus Pro mystery that seems to have solved itself. These recent experiences are why it's hard to have much confidence in the printer, especially given its age. But life with a 7900 goes on, A couple of my inks had reached < 5% capacity, so I ordered replacements and will hope I can continue to print well into (and beyond!) the capacity of those inks.

  --Jay

The Worst Printing Day Ever

Welcome to another chapter of this increasingly irrelevant blog. My Epson Stylus Pro 7900 is now over eight years old, and is two generations behind the current models. The blog certainly won't be helping buyers of new printers, its original intent, but I plan to continue postings here whenever there's something I feel worth noting, and today's attempt at printing certainly warrants.

I've just spent four hours making six small prints, which required making eight prints, and has frustrated me to the point where I can barely type this. But I'm doing it now to get this behind me. Another day like this one will surely result in a YouTube video of what dynamiting a old 7900 looks like.

As is nearly always the case, this started as a simple client job, to print five 8 x 10 inch portraits on letter-size sheets of Epson luster paper. These are portraits of tribal council members (I live on the Flathead Indian Reservation in west-central Montana). I've made these for years, each time there's an election and new members are voted in. The client provides high-quality files for these, the print jobs are easy, and the results very nice. For the same client I would also print one slightly larger scenic on a nicer paper.

I last printed 17 days ago. When I set up today to print I got a perfect nozzle check (#1), so I made the first print. This was the lone black & white image of the bunch, and the banding in the print couldn't be missed. I ran another nozzle check (#2) and found the LLK pattern almost entirely missing and VM with large empty sections. I did a powerful cleaning of LLK/Y pair and a standard cleaning of C/VM. The reason for the different types of cleanings is based on experience with the machine, and knowing which channels have historically required (or not) powerful cleanings. Experience is a weird teacher, providing its lessons after you need to know them, and often being pretty fickle regardless. After those cleanings a nozzle check (#3) showed LLK at 100%, but LC entirely missing and most of VM likewise.

The next attempt was a powerful cleaning of the LC/VLM pair and a powerful cleaning of C/VM. After those cleanings the nozzle check (#4) showed LC still entirely missing but the rest of the channels at 100%. Since I hadn't in a very long time, I cycled the printer's power off and back on. I then printed an LC purge file, which prints a business-card-sized patch of pure LC ink. That patch looked perfect, so I printed a nozzle check (#5) to see if anything else had failed. The LC patch still had some small, scattered divots.

With the morning quickly drawing to a close I decided to print the small portraits, and got a good result.

The last print for this client was a scenic with a lot of empty blue/cyan sky, to be printed on an 11 x 17 inch sheet. I'd print this on Canson Infinity Baryta Photographique 310 (I love these concise paper names). The result was about as you'd expect, some banding in the sky. I printed another nozzle check (#6) and found the LC channel to be a bit worse than it was before I made the print. Really annoyed now, I took a break for lunch. I thought it might help if both the printer and I cooled off a bit.

Half an hour later I did a powerful cleaning of the LC/VLM pair (#7) and finally got a perfect pattern in the LC channel and all the rest. I made a test print on Epson luster (cheap); it looked very nice. So, thinking I was nearing the end of this ordeal I made the final print on the Canson paper. This started out as usual, but about 1/3 of the way into the print the 7900 stopped, seemed to think about things for half a minute, and then started again. Behaving exactly as I described in a posting here waaay back in September, 2017, the machine took another 20 minutes to finish the print. The good news is, the print is perfect. That bad news is, I've no idea what's going on, and I certainly wasted a lot of ink making these six simple prints.

I've powered down the printer and have zero interest in messing with it any more today.

  --Jay

Photo Tex

In my last posting (9 November 2019) I mentioned getting an interesting commission. This may involve printing on materials I've never used, which is part of what makes it interesting.

A client, for whom I've done some scanning and printing work, is looking for a photo to be installed or mounted on the doors of a large cabinet. He showed me a rough sketch of the cabinet, but the design is still in the conceptual stage. I don't have any dimensions, but the cabinet will hide a large flat-screen TV, so the print, which will span a pair of sliding doors, will be several feet wide. The client is an experienced carpenter and craftsman; he's made some beautiful furniture and built a number of custom high-end homes. I expect the cabinet to be nothing short of gorgeous, and I hope his idea of mounting a photo on its front doesn't impact that.

I could make a print (or prints) and then laminate it, or print on a plastic substrate, or print on metal; these last two would be farmed out, not printed on my 7900. I'd also thought about printing on Photo Tex®, a material I'd never used, nor even seen up close. I've sold images to clients who printed them on Photo Tex and then applied the material to mural walls or outdoor signage, but I've not seen the final results. Those clients claimed to be happy with the finished product.

I ordered a sample roll of Photo Tex from the maker. This got me a 42-inch (107 cm) wide roll ten feet (3 m) long, which of course is too large for the 24-inch 7900. This is a lot of material for $30 USD, plenty for making test prints with different profiles. It's a much better "deal" than the Photo Tex sample packs sold by my usual paper vendors.

The sample arrived quickly. I cut letter-size sheets from the roll, downloaded the canned profiles from Photo Tex. They offer two, one each for MK and PK. I had PK loaded in the printer so I printed my usual profile test page with the PK profile. It looked pretty bad, but I set it aside to view after 24 hours. I then swapped the black inks to MK and printed the same test image with the MK profile. A little better, but not much.

A day later I still found both test images to be wholly unacceptable. They are little better than a poorly made image printed on plain bond paper, or the lowest-gamut of mat papers. Washed out low contrast, faded-looking color, weak blacks.

Next I made a custom profile, something I don't often do in part because I so rarely need to, and in part because what I have for making profiles isn't very good. I use LaserSoft's Silverfast scanning software with both my Nikon V-ED 35mm film scanner and my Epson V850 Pro flatbed. Both are profiled with LaserSoft's "Advanced" targets. A print profile software package is available. The software is used to make a print with several hundred sample patches. After an adequate drying period the patch print is scanned on the flatbed, and the software generates a printer profile. My results have been quite variable and never really very good. In this case my print made with MK ink and the LaserSoft-generated profile was much worse than the prints made with Photo Tex's canned profiles.

Photo Tex is quite popular and I'm sure has its place. With the right printer and profile it may be possible to get a decent/accurate print on Photo Tex material. With what I have, and the kinds of images my client is likely to want, it's not.

Perhaps I expected too much. But it was an interesting experiment.

  --Jay

It's Been How Long??

I can't believe my last post here was 11 months ago. Those months have not been uneventful, and some of the events I could easily have done without. But more to the point of this poorly maintained blog:

My 7900 is now eight years old. Among the owners I've known of these things, that makes it the oldest still-functioning unit. And that is, I suppose, the reason my last posting was so long ago; there's been nothing to report. The machine is still running, still making gorgeous prints, and still only moderately painful to start up for each print job. Nothing has changed. Some days I can spend 40 minutes or more getting a good nozzle check print. Other days things "just work" such that I wake the printer, run a nozzle check, it's perfect, and I print. I love those days!

I've recently been commissioned to make some prints for a rather odd installation. There's no deadline (yet), and I'll need to do some experimentation with materials to determine what will both "work" and be acceptable to the client. More to come, and I promise the next posting will be sooner, rather than later!

  --Jay

LLK Problem Resolved?

My last two postings here mentioned the problem I've had with the LLK ink channel. To summarize, for several months, going back to the middle of summer (2018), each time I've awakened the 7900 and run a nozzle check the LLK channel's pattern has been partly to completely missing. Nearly every time a simple "standard" cleaning of the Y/LLK pair has cleared the problem and I've been able to print with the usual excellent results. I worried that this might be indicating my printer was nearing the end of its life, now that it's more than seven years old.

Several weeks ago I did an experiment that seemed to solve the problem. I've waited this long to report here so I'd have time to complete a number of print jobs and gain more confidence that the problem really is fixed. I now believe it is.

As mentioned in my last posting, I'd replaced the LLK ink cartridge, which had about 20% of its ink remaining, with a new one; that didn't help. A while back I stumbled across Marrutt USA's Web site and found a section about using "purge files" to clear stubborn clogs (which may or may not really be nozzle clogs). Each purge file is a jpeg of a single color patch corresponding to one of the printer's color channels. I downloaded their files for the 7900, but didn't have an immediate use for them. I revisited their site and found they'd changed the files a bit, making the color patch a bit larger, so I once again grabbed the files.

I then set up by waking the printer and sending the LLK purge file. I did not print a nozzle check prior to this. The file puts a LLK patch in the middle of a sheet of paper (I used plain bond paper). I then ran a nozzle check and it was perfect.

In the weeks since then I've printed a number of jobs both for clients and of my own work. Various papers, various print sizes. As usual prior to each job I've run a nozzle check, and each time LLK has been perfect. There have sometimes been little nits missing from other channels, requiring a cleaning to include that channel. Pretty normal behavior. But LLK has been perfect every time. I can probably remove the question mark from this posting's title!

You can get these files here, and I'm grateful to Marrutt USA for making these available.

  --Jay