Unclogging epson inkjet heads.
Desperate times call for desperate measures.

More printer dramas at http://nerdipedia.com/

(2 of the 6 rows of nozzles in a 1160 print head)


Photography page

Inkchip hack

An almost random collection of thoughts on getting the buggers to print again.

Last month I posted this in a couple of newsgroups.

(14'th sept 2002)
When I bought my first photo quality printer - an epson stylus 760 - I was quite convinced I'd killed the head on the first night.
This has happened many times since then and happily both the 760 (in a new home) and my 1160 are still working.
This is despite much abuse (read that however you like), several ink sets and much experimentation building many versions of my home-brew CIS.
Last night the 1160 looked like it had finally died.
Some background.....
This printer is about 9 months old. It has done hundreds of photo-quality prints. It has had about 250ml of black, 150 of yellow (I use a lot for some reason) 80ml cyan and 100ml of magenta pigment thru it.
I converted it to my CIS when it was brand new, I didn't even finish the OEM cartridges that came with it.
I use the "enhanced generations" G4 pigment inkset. Pigment is going to cause a lot more trouble than dye.
I've had a lot of trouble with bubbles in the cyan. I've rebuilt the CIS last week to get a better handle on it. It really looks like the ink is "gassing" rather than air is leaking in - it is hard to tell exactly. I have a theory that the silicone tube flicking around is cavitating the ink. If it persists I will go back to using plastic tubing (not yet - one change at a time).
With the old CIS still in place I re primed the cyan and had a bad test pattern of cyan. I got sick of it and ripped out the CIS and built one similar to my original syringe based system.
Primed it and had a lot of trouble with cyan. I found cleaning the head with tissue helped and cleaning cycles actually made it worse - eventually I found out I could get the rubber wiper which wipes the head into a position where I could clean it - this helped a lot. I think I got it working that night but I'm actually not sure.
Anyhow cyan continued to be a problem and eventually I did the windex of the landing pad trick and after leaving it an hour one cleaning cycle had all jets working.
I thought all was well, I printed a photo and turned it off.
Next day - no cyan.
I decided to give it a longer soak. I windexed the pad and left it about 18 hours.
I could not clear it. Having read some of the drastic measures recommended of the web - like forcing windex thru the head with compressed air. I figured it was time for desperate measures. I forced 2ml of windex into the cyan head with syringe, reconnected to the ink and did a zillion cleaning cycles, dummy prints, head wipes and only got a few spots of cyan ink thru.
I had pretty much given up - I windex the pad parked the head and cooked some dinner (mussels in their shells – yummie).
I ate dinner, did one cleaning cycle and all jets worked!!!!!
I did two prints and turned it off.
A nozzle check this morning had all jets ok. This is not conclusive but it's a good sign.
So when I hear of epsons being junked I wonder if the owner gave up a little to soon.
I also wonder whether the 2X00P series print heads are any different.
I also wonder what tomorrows nozzle check will do.
If the printers just cleared their nozzles a few times a day when idle it could make a big difference.
Cheers Eddie, aka inkyfingers.

This had a few responses, the most interesting was....

I just cleaned a Canon 8200 with a strong HOT solution of Ammonia and water. I had used windex before, but it didn't touch this clog. I also used a piece of rubber tubing to press onto the outlet and suck the solution up into the capillaries. Since the cleaning I have printed about 20 pages of color photos. Not a clinker in the bunch. {except where the picture was bad to start with ;-(( }


Getting these printers working can be a very frustrating task. What works one time doesn't another. Sometimes it just needs patience, other times chemistry and sometimes (though I've managed to avoid it) the printer needs dumping.
After a long unsuccessful cleaning session one tends to become desperate and a little irrational.
The full moon is in Libra/Aries tomorrow - I'd better have the windex ready.....

Some of the suggestions I've found online are not relevant to me because I run pigment and because I run CIS.

My latest project “Project light black”. PLB has caused the worst clogging ever, whether I can save the printer remains to be seen (it died). I knew this could happen which is why I did the tests in my 1160 and not my new 2100p.

In the (ongoing) process of trying to save the printer I tried a few things I haven't seen elsewhere.
Before proceeding readwww.inkjetart.com/tips/cleaning.html if you haven't already.
You probably should also see my 1160 CIS page

My experimenting with PLB hints at a possible problem with one of the suggestions.
The one that says “Repeat step 1 and step 2 with denatured alcohol to avoid any residue which might cause problems with the ink to follow.”
The thing is my “clog from hell” happened from running pure(ish) methanol/ink mix. I've also found pure water/ink to be bad. My best results have been from a 30:70 alcohol water mix and mentholated spirit seem a little better than methanol. I doubt dye based inks have this problem.

If the “denatured alcohol” is blown away with canned air then it probably isn't an issue. If the windex blown away you probably don't need the “denatured alcohol”.

I'm speculating along with most people. I haven't pulled apart a print head (if that is even possible) so I don't know the exact geometry. I presume there is a largish ink channel feeding the small piezos. The head appears to have a fairly small volume micro/litres not Milli-litres. When changing ink colors they seem to flush out easily so they don't seem to have pockets to trap old ink.

Knowing the real geometry could be handy when diagnosing problems. One school of thought says that randomly clogged jets are caused by gas bubbles. This may be true but could large lumps of foreign matter floating around also do this? I don't know.

If there is a large bit of muck in there no amount of cleaning/blowing/windexing is going to help.
I don't use compressed air, I use a syringe. With a syringe you can see if there is a flow and feel the force required, I'm not sure how much pressure is safe. I've found that when I can't get anything into the head I can always suck back and then I can.

Here are some of my tools. On top is a 10ml syringe with a glass tube extender. The joint is silicone tube. The other piece of tube is multipurpose. One function is to fit directly over the ink spikes. The cut down syringe can be attached to the spike as a reservoir to feed solution into the head during cleaning cycles etc. A cover over the top would be good to prevent stuff splashing out.

The second use for the multipurpose tube is to go inside the reservoir to allow stuff to be sucked out without removing the reservoir.

People using cartridges sometime make cleaning carts by filling MT ones with windex – sounds like a good idea to me. Having a trace of ink left lets you see which jet are clear. However if like myself you use pigment you probably don't won't this mixed with cleaner. So I mixed some old dye based ink with my windex instead.

The cleaning.htmlhas already mentioned that not all cleaning cycles are equal. You should also be aware that a “charging” cycles is different as well. A charging cycle happens when the printer thinks it has had a cartridge change. In the 1160 it is triggered by the lid switches on the head. In some it senses the “inkchips” are different. On the 2100p unplugging/replugging it triggers this - expect when there is a chip reading MT. If you just want the printer to pump ink the charge cycle is probable better than a cleaning cycle (some cleaning cycles do pump a lot of ink but most don't).

My other perhaps silly innovation is to give the head a bath.
I do NOT recommend this!!
I did this for 2 reasons.
Firstly I wanted to know if the air I was sucking out of the nipple was coming thru the nozzles and not a leak somewhere.
Secondly I thought it might be good the suck solution thru in the reverse direction and possibly dislodge crude which wouldn't fit thru the nozzles.
Making a bath turned out to be very simple (for the 1160 at least). What I did is moved the head out manually and then I pulled some plastic sheet (a sandwich bag actually) under the head till it reach the metal rod the head slides on. I raised the sides by packing sponge under the bag and poured windex into it. This worked and I sucked windex up thru the head. The windex was very black from the ink it had dissolved of the bottom of the head. If I do it again I'll cleaned the head first with tissue.
I won't swear that this did anything useful – it is hard to tell. It may even be dangerous – there could be wires down there for all I know. I didn't seem to do any harm though.

The patient died on the 22'nd of October– cause of death unknown.

UPDATE – The head was probably OK!!! The printers go into the “error” mode when it think the ink pad needs changing. This can be reset manually – I did not know this at the time.
See - http://www.mwords.co.uk/pages/printers/printersHowDoI.htm#ResetPrinter

I seemed to be making progress, I tried a few other cleaning chemicals and had most nozzles firing. Then the printer just stopped and flashed it's lights at me. It could be that the head failed some sort of self-test or it could be a totally different fault. I have no way of knowing. It is safest to assume I killed the head but this may not be the case. When it stopped two thoughts came to mind. The first was that liquid had splashed on the main circuit board and shorted something out. The electronics turn out to be quite well protected from the wet areas. The second thought was that all the liquid that disappeared into the bowels of the printer had filled a reservoir and tripped a level switch. Also not true the liquid is soaked up by a huge piece of felt which seems to fill the full length of the pan.

Off with it's head.

Had I know that the head comes out so relatively easily I would have been out long ago. I've always wonder how it works. The basic piezo design is known and a low-res image of the first page of a patent is athttp://www.delphion.com/cgi-bin/viewpat.cmd/US05513431__?MODE=fstv
What I didn't know is what the underside of the head looked like, how the plumbing worked and how it was primed.

The bottom looks like this, you can click thru for a better look. You can see the 6 rows of nozzle, 3 for black and one each C,M + Y. Six color printers might use the same head. On the right is a plastic ruler showing a mm scale. It is a big silicon chip.

A this is a closeup of two rows of nozzles. It would be interesting to compare it with a new head to see if the holes have eroded.

This is the front view, the four spikes which go into the ink carts are at the top and the tubes running down to the head are visible. I had expected more. I thought there might be some pump or bypass valve for cleaning/priming but not so (more on this below).

This is this interface board which take signals from the ribbon cables and routes them to the head. The 3 pin socket presumably goes to the cartridge change switches.

Now it all makes sense (well mostly).
In my CIS the ink is at reduced pressure, the same is probably true of cartridges because you have to suck the ink out of the foam. The 2100 actually has a pressure regulator built into the cartridges which the printer sucks against. Once filled with ink the capillary effect will suck the ink thru the system. What I didn't know was how you got the ink into the nozzle to start with. The answer is simple, the printer uses a vacuum pump.

This is a shot taken thru the hole where the head usually is when parked. It shows the rubber cup which comes up against the head. This is the foam block you windex to un-jam the jets. What isn't obvious is that the tube coming off the bottom is not just a drain, it goes to a device which I presume is a pump. This is driven by the paper advance motor.

On the inkjet art page it says.

“You must stay in the utility menu for the printer to use increased pressure, you must also run a nozzle check between cleaning cycles, or the printer won't actually clean its self, just spin the roller. Each time you repeat a cleaning cycle without exiting the menu, the printer will use increased pressure.”
I don't mean sounds over critical, I'm glad he put this info online for everyone. But I think it is wrong (for the 1160 at least). The printer does not “apply pressure” - it sucks. It does not “just spin the roller” it is spinning the pump. This means the print head is never pressurized and the maximum pressure difference across it is one bar. With a syringe you can apply much higher pressure than that. The head may not be designed to take it but it seems to handle it.

So was the head clogged?
No it wasn't. On the high res image I can count all 48 jets are squirting. I think there may be more to it than blocked jets and bubbles. I think it may have to do with surface coatings and surface tension but it will remain a mystery to me a while longer.

Some more thoughts.

On Powell's page he digs a little further and reveals a filter below the ink spikes. If this were clogged it could have symptoms like I was seeing. Also if clumps formed inside the head (say pigment particles clumping together) there is no way to get them out. You would have to find a way to break the clumps up, remove the filter or add an extra opening to the ink line for sucking them out. This is not out of the question once you've reached to “bin” stage.

The next Patient.


A bit of a non-event really. I was given a stylus 400. It had not been used for a few months. Some heads were clogged. Windex on the pad - 2 hours and several cleaning cycles later all jets were ok. I continued with PLB and promptly clogged one head completely. Next day I suck it out, put in a different ink mix and easily got all jets working again.


Powell Hargrave has a similar page herehttp://members.shaw.ca/hargravep/head.htm


Ciao Eddie, aka inky.