How to print and use purge files


A goodly number of members here and elsewhere suggest doing fewer head cleans (as in the Epson software) and printing more surge patterns. I haven’t seen much detail on just how to do that, but here’s what I do. Please comment, correct, as you see fit. Am using a 1400 printer and an inkset with Eboni in the K position and various dilutions of that ink in the other five positions.

I’m using the tiff files as listed in this Article: How To Flush Individual Channel using QTR Calibration Mode I print these files in the Calibration Mode because, as I understand this, they are “pure”, using only the ink in the cartridge in question. Despite the outward appearance of the color, these files print according to a code in the file.

First I print the 6 channel chart at 720 ppi to get a rough idea of trouble spots, Then I print only the file having issues at 1440 in unidirection, sometimes in bidirection. I may make more than one copy. This more intensive print serves to free up clogs and air bubbles.

Am I on the right track here?


Sounds ok to me. It’s not overly complicated.


Well, you’ll remember me from a previous thread. I have a genius for making simple things complicated.

BTW, you’ll also remember getting me to use the syringe push-pull cartridge cleaning technique. I now use it with no qualms and have a good feel for it. I’ll be forever grateful to you.

As an aside… do you know why most of these these tiff files show on my desktop monitor as some form of magenta - regardless of the internal code telling the printer otherwise? It’s very curious.


Yes, I remember.

I need to amplify something I said earlier. When you print these TIFF files in calibration mode, you are not printing those colours. The colours are just codes that QTR interprets when in that mode, and the RGB values specify which channel and how much ink. Some of them may look identical, but in fact they are every so slightly different.

If you examine them in Photoshop, B is always 255 and G is zero, and the red value varies. The R value is the code which determines which channel is printed in calibration mode. The R value doesn’t vary by much in some cases - for Y it is 239, R is 251, LLK is 254, PK is 253 - but by enough for QTR in calibration mode to take these values as a signal to fire a particular channel. QTR can do this, because being a RIP, it interacts with the printer directly bypassing the driver, or most of it.

If you look at the ink separation page that you would normally print in calibration mode, you will see that the green value in the RGB number is used to determine density. Zero green = 100% and 255 Green = 0%, and linearly between these two end points. Again, you are not actually printing these RGB values, QTR only interprets them as a control code. In the TIFFs in the above link, G=0 so those TIFFs generate printing at 100% density.

If it’s any consolation, it took me a while to realise all this. I was always puzzled by the colours in the ink separation page, and it was only when those TIFFs became available that I realised what was going on.

Better agitation of ink in desktop printers

I don’t use QTR (at least not yet), but one method I’ve adopted for reducing head cleanings on my 4900 is to use a macro through an application called Keyboard Maestro (excellent for automating certain functions) to run a relatively small (but using most of the width of my 17" printer) print with the full range of colors. Since I generally print only on weekends, and sometime am away for a few weeks at a time, without this I know I would have had may more clogs. I was turned on to this through Martin Bailey’s blog. You can see his full instructions here:

Also review the comments at the end of the post which, if you adopt this, may help in trouble-shooting the process so it runs error free. I’ve been away for two weeks and this has run flawlessly in my absence. If you try this for any time you are away, be sure to turn off any auto-notices from Adobe (including asking if you want to back up your catalog), since those will cause the macro to fail.



Thanks for the in-depth understanding of the code in those tiff files. It is indeed a consolation!

Being able to print purge files has unearthed a few more questions:

-> I’m seeing some banding at the beginning and the end of the C pattern, about an inch long. It’s the sort of banding the 1440 creates, but my 1400 has firmware that supposedly doesn’t do this. I’m referring to the purge files on my 1400… is it possible this 1400, a refurb, has the print head from a 1440?

-> when I use the syringe push-pull declogging technique, I’m using Piezoflush. Does that have the effect of thinning the ink in that cartridge? It’s the Cyan cart, and is about 30% black and 70% base fluid.

-> Again, in Calibration Mode, I have several options for printing. I ran across this very helpful thread on this topic: Solved: printing at 1440 and bi directional or 2880?? But I can’t ask there because it’s listed as solved. My question is what does “super” mean in the list of resolutions? If you’d rather I ask Roy, that would be fine.


It’s been a long time since I had my R1410, like nearly six years. At that point I knew nothing about micro-banding in the first and last inch. I only discovered it when I upgraded to my R1900. I read reports from others who claimed no micro-banding on the R1400/1410 and I went back and examined my prints. I couldn’t see any. In one or two instances there may have been the slightest hint of it, but I couldn’t be sure. It’s possible that in fact it’s there but at a much reduced level, and perhaps you only really see it at the 720 setting and not at 2880, which is what you’d normally print at with QTR.

As per the article in my blog, my understanding is that what determines whether a printer has this first and last inch micro-banding is the printer firmware, and whether certain code to deal with the first and last inch is in there or the driver. The smaller printers have in the driver and not in the firmware and so QTR can’t access it. Why the 1400/1410 is an exception is a mystery. But I’m fairly certain that the head won’t make any difference, and given the cost of Epson heads I’d be stunned if they replaced it as part of a refurb on such a cheap printer.

I’m not sure I understand. Thinning the ink in which cartridge? The syringe is normally used in conjunction with the dummy cartridge. You’re using Roark inks, are you not? When you reinsert your ink cartridges, with a mix of black and diluter, the resulting head clean should pull the solution in the cartridge through and the piezoflush should be evacuated as a result. You may need two head cleans for some of the lighter shades, and the odd purge pattern at 2880 might also help.

I have seen the differences between the various 1440 printing resolutions described somewhere but I don’t recall. You’ll have to search. I always print via QTR at 2880 and unidirectional, even on matte paper. If printing via the Epson driver (ABW (heaven forbid) included) on matte paper only then would I drop to 1440.

Regarding the suggestions from JDRamsay, scheduling some sort of regular print job is a useful way to prevent printer problems. This is particularly true on larger printers with ink lines and dampers where you can’t readily (i.e. cheaply) hibernate them. It’s doubly true on the notoriously troublesome x900 printers. On the desktop printers I think it’s better to hibernate them with flush carts for extended absences, as we discussed in another thread.

I think JDRamsey’s implied point is that you can’t automate a QTR calibration mode purge pattern, at least not on Windows. So if you want to use this technique for preventative maintenance, rather than diagnostic and for clearing air, then you have to be present and trigger it manually. For automated purge patterns, you will have to use a different technique such as the one described. These other techniques should be fine for exercising the printer, but the resulting prints will not be as useful as a calibration mode print for diagnostic purposes, because the printouts will be a mix of channels rather than one column per print head channel.


Very extensive reply, Brian, thank you.

(Just to be clear, this thread has to do with my 1400, not my 1430.)

All the surge prints and syringe cleanings are paying off. My Cyan surge file, at 2880 unidirectional, printed flawlessly, I had the top and bottom margins set at around 1/2 inch, and there was not even a hint of banding. And even at 720, no banding. For critical work, I’ll print at the max, ie 2880 unidirectional. Then I printed the six ink surge file, with the same parameters, and it too was flawless.

My Cyan position nozzle check has greatly improved. I now have about 4-5 dashes missing in the top line and 2 in the bottom line. Previously, there were more dashes missing, some were a very rough dash, hard to describe. I am hopeful this pattern will improve. (And yes, these are Roark dilutions - I hope I don’t get kicked off the island. I am an IJM customer, Piezoflush, cartridges, EZFill color carts for my 1430, etc.) However, these missing dashes do not seem to affect the print quality of the Cyan surge file - strange.

Thanks to you also, JDRamsay, but I think I’ll pass on implementing your suggestions for now. I’ll keep that in mind for possible future reference.


Just to be clear about my references to the R1410, this was a non-US (and probably non-Canada) version of the R1400. It was the same printer, but it wouldn’t take US cartridges, so it was branded differently. (Epson took a different tack with the R1430 and gave it the same branding in both markets, but international versions still don’t take US cartridges, so Australian users can’t use IJM carts, or at least not the chips.)

If you want to use one of these purge patterns in calibration mode for diagnostic purposes, in my experience it’s best to print at the 720 setting. Printing at 2880 lays down a lot of ink and if there are nozzle issues they won’t show, unless there are a lot of nozzles missing. The 2880 setting is good for actual purging, but less good for diagnostics.

For printing on paper (but not digital negatives), you can generally get away with one or two missing nozzles, although it’s better if they are not adjacent. 4-5 is starting to push your luck in an actual print.


Certainly not helpful to have many different editions of the 14x0 printers around the globe. Must be some marketing geniuses have a rationale for doing so.

Yes, I forgot to mention my reasons for doing a 720 print. That was my thinking. Both settings have their purpose. Correction: not my reasons. I “borrowed” that strategy from somewhere here.

Anyway, things are definitely looking up.


I suspect that the thing about 720 vs 2880 is that on plain (uncoated) paper, there’s a lot of dot gain or bleed. So I suspect that the dot gain when printing at 2880 may hide small problems. If you did the same purge print on coated paper, especially a PK paper, then you may see signs of gaps, but who wants to use good paper for that?

Somebody, who shall remain nameless to protect the guilty, once said to me that only a sad little puppy would take more pleasure in a good nozzle check than a good print. But it is a risk when you’ve struggled with a recalcitrant printer for a long time. When the nameless guilty party later fell into this very trap himself, I was quick to remind him of what he said. :innocent:


I needed to read that last paragraph more than once for it to sink into my addled brain. :confused:

This morning before email check I was taking pleasure in yet another advance in my nozzle checks. Down from 4-5 to 3 (adjacent) in the top line of the cyan pattern and only two (non-adjacent) in the last line.

And here are some bits of wisdom from an on-line colleague, taped to the side of my computer:

“People spend too much time chasing a higher level of precision than they really need.”


“I find that I end up chasing too many rabbits down too many rabbit holes.”

Do you think I heed his advice?

Next question.


There’s an element of truth there. In terms of printing with QTR, before Roy Harrington released his relinearisation droplet we used to accept and adapt to a range of variations from linearity. Custom curves where the only answer for Piezo users, and they weren’t cheap and, for those outside the US, were a hassle. Since Roy did so, we (or at least some of us) no longer accept even small variations from linearity, even when their impact on the print would be barely noticeable. But then relinearising the curve is not a major exercise, and since the droplet there is also now the IJM Pro tools, so little harm is done by pursuing a high level of precision.

I think that quote is actually more true of colour printing. I’m often told that I need more patches and better profile software than I currently use. So once I created a series of ICCs using different numbers of patches, one of which wasn’t all that many (256?), and printed out one of those colour test images using each of them, and the differences were pretty minor and not all that noticeable. You couldn’t say that the smallest patch number produced a bad print. I am sceptical of those who seem retentive about high levels of precision for colour printing, other than in some critical situations. But then I do prefer B&W.


By that you mean the Stepwedge Tool? It is pretty slick all right. I was pleased to discover it even worked with my humble PhotoShop Elements.


The re-linearization droplet for QuadtoneRIP is called QTR-Linearize-Quad and allows for direct linearization/calibration of .quad files (QuadtoneRIP BW media type profiles).

We built a set of tools to extend and add to this droplet which correct for measurement errors when reading targets (something that happens with all targets) + even more extended tools like L* characteristic matching, curve blending, and more:



Just confirming that Walker understood exactly what I was referring to. My blog article on using the relinearisation droplet:


P.s. as noted early in the blog article, the droplet is only really relevant to Piezo curves. If, purely hypothetically, someone was using Roark inks and had the matching .qidf file, perhaps because they created it, then they wouldn’t need the droplet. The point about excessive precision still applies.