I added some lines to the Arduino Sketch, so that the first four bits of the i2c message are used for Drop on Demand / shooting single drops for dot matrix printing and the last four bits are used for shooting drops at set frequency for printing along paths.
Maybe this could be useful because you can create the gcode for it very easily with every CAM software and it can also be a bit faster than dot matrix printing.
Here are some pictures of my latest printer design. It's not so different from the last design. The only changes are:
- New printhead which has a vacuum duct around the nozzles to suck in the excess ink.
- New driver electronics for switching the piezo discs - four IRFP460 that switch around 170VDC.
- Less 3D printed parts and cleaner look, because I placed the electronics in two boxes.
That's about it.
I tested it out and it works reliable even after not being in use for a day or two. After a short priming it is ejecting droplets again.
I'm currently using 0.2mm nozzles and the ejected droplets are quite large. Maybe I can change that by changing the switching time or by using even smaller nozzles.
But at least it's working now and should be usable for some test prints.
I uploaded the .stl files for it and will write building instructions for it as soon as possible.
I got the new electronics and all four pumps running, did some testing with it and besides one thing everything seems to work ok.
The remaining problem is, that by mounting the nozzles in a horizontal arrangement, they often eject not only one droplet, but rather one lage and multiple small satellite droplets, which is a bad thing.
If the nozzles would be mounted in a vertical arrangement, all droplets and satellite droplets would hit the paper at roughly the same spot so that they would merge together, but in the horizontal arrangement they hit the paper at the entire flight path of the main droplet.
So mounting the nozzles in a vertical arrangement would be good, but by doing so there would be forming of ink drops on the nozzle again which would block the nozzle orifice until they fall down on the paper.
The satellite droplet appear more often when there is some air in the lines.
Another thing is using the 6mm OD / 4mm ID tube in the case that air can get inside the tube it can happen sometimes that the ink forms a channel inside the air bubble so that the ink can flow through the bubble instead of pushing it out of the line.
Maybe lines with a smaller inner diameter could help.
So, I have an idea how I could prevent drops from forming in the vertical orientation - will try it out later.
Here is another printhead prototype with a vacuum duct at the nozzle to prevent any drops from forming at the nozzle. This is the best drop preventing design that I built so far, because it does not just collect the drops, but can completely prevent them from forming at all.
I will design another one for all four colors for more testing. It looks like (I hope) I'm on a good way with this design :)
So, here is another printhead for all four colors.
It looks like I have finally build a working printhead, pump and driver setup.
Even though there are some unwanted satellite droplets, which I think are caused by a partially clogged nozzle or a bit unprecise alingment while assembling the printhead, I'm really happy with the result because the machine is basically working as expected.
The ink is drawn by the piezo pumps from ink containers through check valves to the printhead and can be ejected in a drop on demand method. That's what I wanted and now it works.
In terms of reliability it works a lot better than the design from two years ago, that I could not get to work with four colors at all and with one color only if everything was 100% tuned right in terms of air in the lines and water level.
The new design can work with some air in the lines and can also force larger bubbles out of the nozzle again in some cases.
Now I can write building instructions for it and upload all files and infos.
I also want to work on a software for it to do more testing in the future.
Today I tested out to print something with the new design, but it did not really work as expected.
The good things are:
- The pumps can pump the ink up to the printhead and can eject ink out of the nozzles.
- If the pumps have enough power (when using two piezo discs) there is no problem with making bubbles.
- It's reliable. As long as there is no air in the lines it works without problems even after days of being not in use.
The only bad, but unfortunately essential thing is:
- It works good for streams of ink but not good for drops unlike the last design which, however did not work reliable.
So I have to find a way to get it more suitable for ejecting drops, somehow...
My first idea would be using DC again instead of AC.
I think the relatively slow rise of the voltage in the 50Hz mains sine wave could be the problem. With a DC square wave the voltage rise and therefore the piezo movement and ink pressure change would happen almost instantaneously what should be better for creating droplets.
The problem with using DC would be, that the piezo needs AC to move in both directions to perform not only a change in ink pressure, but a pumping action.
So the best should be a high voltage AC square wave.
At the moment I only have the parts to switch high voltage (up to 500V) DC.
I will try that out later, but it will likely not work and so I have to find or build something that can create a high voltage AC square wave....
I just figured out that the DC SSRs are really bad for switching the piezo disc compared to the IRFP460, so I will rather use these.
I built another printhead that has a drain directly underneath the nozzles to suck up the excess ink drops. It is mounted at an angle, so that ink drops do not form at the tip of the nozzle - what would block the nozzle - but instead form on the side of the nozzle until they drip into the drain.
For this printhead I used 0.1mm nozzles. With two piezos in the pump the amount of ink per drop is a bit high what I think leads to the drops that are forming on the nozzle, but using only one piezo in turn is too less ink what leads to no droplets at all.
I think the new printhead in combination with a PCB with four IRFP460 and a rectifier should make the design ready for first printing tests.
New electronics for the piezos.
A transformer, rectifier, 12VDC power supply for the optocouplers and a board with four IRFP460 and 47k 1W resistors for switching and discharging the piezos.
Placed the electronics in two boxes for a cleaner look.
To improve the reliability of the printer I build another printer design that is based on the piezo pumps. The new design does no longer need a level block and ink cycle because the piezo pumps are powerful enough to pump the ink up to the printhead to eject ink in continuous streams or pulsed drops.
The pumps are connected to check valves and 6mm silicone tubing and they have a smaller gazoline filter at the tube input to replace the large inline gazoline filters.
There is also no longer a piezo disc required in the printhead to eject ink, so that the printhead can be build more compact.
The printhead can be rotated to change the drops' flight parth. It's no longer aligned straight to the bottom but rather aligned at an angle to collect and suck in excess ink with vacuum that in some cases could form a drop that would then drop on the paper.
Included in the SLA printed printhead are channels that guide the ink to the bottom and then again to the top, so that the nozzle is at the highest point on top of a small chamber to prevent it from sucking in air bubbles what would lead to ejecting bubbles from the nozzle instead of ink drops.
On top of the printhead is a SLA printed plate in which I drilled small holes per hand. This plate can also be laser cut or cnc milled to create even smaller nozzle holes.
In terms of the electronics I also changed some things. The printer now has a 230VAC IEC input
that powers an 12VDC power supply for the RAMPS and
a small step down converter
that powers a small vacuum pump with a low voltage (set the voltage that flow rate and noise level is comfortable) which guides excess ink to a container.
The 230VAC also powers a small transformer that transforms 230V to 2x 115V (In reality I measured 131VAC at the output without a load).
The transformer's output is connected to a 4 channel SSR which switches the piezo pumps.
wiring
The connections are distributed via WAGO terminals. The backside of the piezo pumps are also printed in transparent SLA resin to make sure that every part sits on the right position and there is no ink leaking what would short out and destroy the piezo disc.
A part of the printer that looks likely a bit weird are the lightbulbs on the right which are used as dummy load which is needed to guide the leakage current of the SSRs around the piezos, because otherwise they would not fully turn off even when the SSR is not active and so there would always be a bit of ink leaking out of the nozzle.
They could be replaced by power resistors, but I had not the right value resistors at hand and so I used 15W light bulps.
I think using SSRs for switching the piezo pumps should be the easiest solution, but probably not the most precise in terms of controlling the piezo motion and therefore the exact amount of ink, but if it would work with the SSRs I will likely keep them to keep the build simple and easy to rebuild.
So, here is the new design and I hope it will be more reliable, because I fixed some problems that I had with the old design.
Because the current design with hydrostatic pressure fed prindhead has some reliability problems - sometimes one of the four heads starts dripping or sucking in air before or during printing - I'm working on a possible solution right now.
My plan is to build SLA 3D printed Piezo Micropumps to either feed ink to the printhead according to ejected drop count or integrate the pump directly into the printhead, so that the printhead can draw ink form the ink container.
I designed and assembled a pump and later today I want to test it.
I also bought flexible resin for printing the check valves in the pump which are currently printed from the transparent resin that I used for the printhead.
Update .... Success
After a week of failing on building all piezo micropump designs that I found on the internet I finally got something to work. I tried out pumps with 3D printed valves and valveless diffuser nozzle and channel design, but they did not work. The last thing I tried was using the latest inner channel pump design, connecting two duckbill valves to it, priming it and - because of the lack of a 100V+ piezo driver or parts to build one - connecting a 115V AC transformer to it.
And out of a sudden after days of failing - It worked!
I guess it's actually really simple you could for example if you stay with the 50Hz from mains use a small transformer and some SSRs to control the flow by the ON time to precisely dose the ink. It could work by counting the ejected drops, like after some amount of drops let the pump run for some amount of time or pump run time per drop in milliseconds.
Until now the printheads were fed by hydrostatic pressure and I have not tried out if they would also work when they are feed by pump, so I need to test it first.
The former ink cycle could likely be replaced by the pumps if you don't need mixing or want to use a big ink container.
If you would use the same driver setup it would likely replace the four pumps and step down converter with a small transformer, four SSRs and four piezo pumps.
I also tested out to run the pump with just one piezo instead of two. It did also work, but the flow rate was really low and so I think it would be better to use two piezos and limit their On time instead.
I tried out to switch the pump with a SSR because I had some concern that because of the piezos being a really small (as far as I know) capacitive load it would not work, but it seemed to work. Because of the SSRs leakage current they do not entirely turn off, but in the off state the pump has not enough power to lift the liquid so that it should not be a problem.
I also ordered an 8 channel SSR module to try the 115AC for the printheads, too.
New more compact pump design
I changed the pump design to be more compact and also replaced the 4mm PU tube + M6 fittings by 6mm silicone tube + tube connector integrated into the part.
I did not integrate the duckbill valves into the part, so that you don't have to find the right sized one and can instead buy check valves that fit the tube diameter if you want to build the pump.
I think for the connections it would be the best to secure them with a small zip tie or something similar to make them more seal and to prevent them from getting loose.
I tested the new pump and it worked very well. And I also ordered different filter which are placed at the end of the tube in the ink container - the ones for chainsaws and lawnmowers.
Update:
The pump in the picture had a height of 5mm and also a tube connector with 5mm diameter. Because of the printing process one side of the tube connector got a bit flat/not perfectly round and therefore the tube connector got a bit leaky.
So, I'm currently printing one with a height and tube connector diameter of 6mm to get a tighter fit and hopefully get a sealed connection.
I also thought about designing a pump which would have the in and outlet at the top to get it even more compact.
Two Chamber Piezo Pump:
Here is another pump design with two chambers, the in and outlet at the top and 6mm height.
The 6mm tube connector diameter and some filing to remove the ridge from printing had lead to a sealed and solid connection.
Inner channels to guide the fluid from the input to the bottem into the first and second piezo chamber and to the output.
I think this pump design should be good enough for using it on the project.
Short note: I had a small printing error in the printed part - a small hole on the back through which air could leak into the chamber. I think this was caused by an error with my printing setting or related to the resin and not an error in the design.
I fixed it by applying some resin on the part and curing it with an UV lamp to get it seal.
This pump design has two piezos but it can also be build with just one if the higher flow rate turns out to be not needed.
New Filter:
I got some chainsaw/lawnmower petrol filters to replace the larger gasoline filters.
Piezo Printhead no longer needed:
While testing out the piezo pump and piezo printhead I figured out that the pump on its own is able to eject dropplets from the printhead. That's a great thing, because with that the printhead can be designed a lot smaller because it only has to contain the nozzle and no longer the 27mm piezo disc.
It's been more than a year since I last worked on this project, but for writing building instructions I redesigned it to be a printer add on, so that you can fit it on your own 3D printer, plotter or CNC machine.
I again fitted it to an Ender 3 3D printer.
On the pictures you can see that the build is made of the tubing for four colors, two SLA printed parts, four pumps, four filters and some electronics.
The machine is powered by an external 36V power supply which is directly wired to the H bridges, stepped down to 4V for the pumps and stepped down to 12V for the printer controller.
The printer controller controls two stepper motors and sends i2c commands to the Arduino UNO. The UNO sends signals (and 5V power to power their logic circuits) to the H bridges to drive the piezo discs in the printhead which eject ink droplets.
Hope this pictures help you, if you are interested in the project and want to build it.
I will also write step by step building instructions soon™.
Here I want to write down some ideas for the printer software.
I think besides CMYK printing it should have a mode for assigning a color to a printhead. E.g. if you have a svg graphic with areas in RGB Colors, black areas are assigned to printhead 0, pure red areas are assigned to printhead 1, pure green areas are assigned to printhead 2 and blue areas are assigned to printhead 3.
It should also have an option to print black and white with all four printheads for faster printing like I planned before for the 10 nozzle printhead instead of CMYK.
The last few weeks I changed a few things on the printer so that I can try CMYK printing. It would be a dream if it works out.
I removed the 36V power supply, 3 L298Ns, the step down converter, the shift register, the WAGO clamps and the pump speed controller and added a 12V power supply, 3 more pumps, a much heavier speed controller and a step up converter.
The printhead is now a 4 nozzle, single piece, height adjustable printhead for 4 different colors.
The T fitting has been replaced by a SLA printed block for 4 different colors.
I ordered cheap inkjet ink which I would mix with water for testing, but today I saw there is even cheaper ink out there which I could use without mixing it with water.
The printer is now ready for testing, but I don't have any software to create the GCODE yet.
I will try to write one in Python with a GUI so that you can select a picture and the settings and it will create a preview and the GCODE.
This maybe take a while because I haven't done something like that before.
Dominik Meffert
