Road to better paste extrusion, episode 1: Recap

Hello all, an awful lot of time past since the last meaningful post at Unfold Fab but fear not we have continued experimenting and printing lots of ceramics.

In a series of posts I will try to recap and document all the experiments we've done in the last 2 years in order to try and get a reliable and usable paste extruder. I just never found the time to put the notes, sources and thoughts on virtual paper, you know that feeling don't you? Ed. took me again couple of weeks to wrap this one up :)

A year ago I started documenting this process also on the RepRap wiki (after a friendly push from Adrian B.) so I will try to update that page also as much as possible. Some content from that page will also be recycled here. You can find the Ceramic_Extrusion page on the RepRap wiki. But anyone, feel free to also jump in and edit that page.

So why 'better' paste extrusion? Whats wrong with the method we used here in our studio (Unfold) since day one and in fact are still using most of the time today?

It might be good to revisit the beginning...

When I started researching methods to print clay with DIY 3d printers in late 2008 there where basically two printer options, Fab@Home or RepRap. The Fab@Home was ready for my intended use since its default extrusion method used a syringe to extrude silicone. But it was, and still is, a rather expensive machine (close to 3000$ in early 2009). The choice for RepRap was based on price and maybe more importantly on its community. RepRap had a vibrant community, that exploded exponentially over time while Fab@Home didn't (well, doesn't) feel like moving a lot. I got into contact with Erik de Bruijn (now Ultimaker) and he kindly introduced me into all things Reprap at the fantastic Protospace Fablab, there he showed me his darwin machine made from cast parts manufactured by Bits from Bytes. I was never interested in building a machine from scratch because I wanted to work WITH, not ON a 3d printer and so decided to go for a kit. Bits from Bytes had just announced their Rapman kit, to my knowledge the first complete RepRap derived kit which included everything to start immediately, so I instantly pre-ordered one of the first handful.

And because clay doesn't come in 3mm filament, so the quest for an extruder started.

Claystruder 0 (Stepper Driven Plunger)
This extruder is based on the principle of a plunger being driven down a syringe barrel using a (stepper) motor. Since Fab@Home used this principle it sounded smart to start here. This can be done either with an expensive linear stepper motor like on the Fab@Home Model 1 Syringe Tool or with a more standard motor and gears. Examples of the later are the Fab@Home Model 2 Syringe Tool, a very early Syringe Pump Prototype by Adrian Bowyer, Zach Hoeken's Frostruder MK1 or Viktor's (VMX) Syringe Tool.
This design never left my drawing and cardboard mockup phase because around that time I met with Bre Pettis in New York right after they launched Makerbot. He described that the Frostruder MK1 was a dead end for Makerbot and that they did some experiments with air pressure to frost cupcakes which looked rather promising. Also around that time Unfold got a commission from Art Centre Z33 to create an installation (L'Artisan Electronique) in which a ceramic printer would play a major part.
So I skipped the Stepper Driven Plunger and jumped straight onto the air pressure wagon. So we actually never had any experience using this 'direct drive' type of extruder on clay paste. Something I feel I need to revisit, even just for the sake of comparison. But more on that in a later post I am sure.

Advantages
-This system is compatible with most software, firmware and electronics in use on Rep(st)Raps due to the use of a stepper motor. With some calibration and fiddling with Skeinforge settings this could be a drop in replacement for the plastic extruder.
-It extrudes a fixed and predictable quantity of paste with each revolution of the stepper.
Disadvantages
-The mechanical bulk and size of the system, assembly height is at least double of syringe length making it rather impractical for larger volumes especially if your printhead is on a moving XY carriage.
-Rather inflexible in syringe sizes.
-According to some sources who tried this system, issues with start/stops and oozing I believe.
-Generally its also not really a good idea to control your extrusion by pushing the whole stock of material from behind, this becomes especially hard when trying to scale this system up to the >100cc syringe range. Also if you go to larger syringes the diameter of your plunger gets larger and it becomes harder to extrude the same small amount as precisely as in a system with a small diameter plunger.
-A rather large force is needed when extruding really viscous clay.


Claystruder 1.x (Time-Pressure Valve)
Based on Zach Hoeken's Frostruder MK2, the time-pressure valve based Claystruder 1 (and 1.5) is the printhead that we used extensively for almost two years to successfully print earthenware and porcelain ceramic objects and is still the tool for no fuss printing but it has major drawbacks, especially one… But first the basic of this system. Instead of a mechanical plunger, you use timed pulses of air pressure to drive the material out of the syringe hence in the industry this is called a Time-Pressure Valve. Mechanically its a dead simple system (apart from the needed source of compressed air). With the use of one double action 3/2 solenoid valve or two 2/2 single action ones you can switch the air pressure on and off from your controller if it has a free port for it. Our Rapman controller has two AUX ports switchable via Gcode but on the software side there is no real support for it in Skeinforge.

Because we where initially a little to lazy to figure out how to add all the M-codes in the Gcode file automatically we evolved into continuous single line printers (more poetically: 'one liners'). I added an ON Gcode in the beginning and an OFF at the end, for the rest of the print it's actually continuously extruding. On later extruders we completely omitted the solenoid and just plug-in the air at the right moment. KISS all the way :). For someone handy with scripting this should be easily solvable and I think it should be rather easy to customize Makerbot's frostruding scripts that post-process Gcode but we just found interesting ways to design around the issue and work with continuous prints and somehow it feels more natural to do this for me and actually design for the process (works nicely for plastics too when printing at 0,5mm). Will post more on that in the future. This also makes sense because clay prints are much more sensitive towards start/stop actions and the speed/direction of non printing moves. The print stays highly plastic during the whole print job and some ooze on your nozzle can easily disrupt a print when the head travels over already printed lines. Switching air pressure behind a body of clay does also not result in reliable repeatable material flow rates.

But there are also designs for objects on the drawing board at the moment that really need an extruder that can be turned on/off reliably so therefore we keep on searching for one.

In the meanwhile I made some improvements to the system since the Claystruder 1.1 version I posted on Thingiverse. The main one being the decoupling of the nozzle (a polyprop tapered tip) and the syringe. This simple change solves the issues with swapping syringes on long prints. If you have the nozzle attached to the syringe directly, like on the Frostruder, each time you want to swap a syringe you also remove and refit the nozzle with it and its very hard to get that syringe+nozzle back in the exact same spot again. Your nozzle is not often straight so even a slight rotation can put it a millimeter off. So if you continue your print (on Rapman its fairly easy to pause and restart a print) it will continue in a different spot. The solution is easy, make sure your nozzle stays in its place on the print head carriage when you remove the syringe. I use a small luer-lock male-female extension bit (from my favorite source) that is glued in the printhead (a simple mount). The nozzle is fixed to one end and the syringe screws in the back. This feature should be part of any paste extruder that uses syringes since it solves a lot of trouble with multi-syringe prints.

Luer-Lock Syringe + Female/Male adapter + Tapered Tip

Claystruder 1.5, bottom and top halves 

There is also a set of windows in the barrel holder that allow you to guard the level better.

Claystruder 1.5, paste level window 

Claystruder 1.5, tip mount

For this extruder I also designed a syringe adapter head that can be easily twisted on syringes and can withstand (depending on print quality) pressure up to 6 bar by adding bolts and washers as reinforcement. This file can be found on thingiverse here and is usable for many applications, it's also a much better replacement for the awkward system with screws on the Frostruder. Warning! I have operated this part and standard medical syringes at pressures up to 6 bar without issues but I guess this is close to the limit. Your millage may vary and I am not responsible if stuff explodes and harms you, your family, your dog or anything else.

Syringe Adapter Head on Makerbot

Syringe Adapter Head

This Claystruder is simpler than the quick and dirty first version and also more modular so that the parts like the adapter head can be used in other applications/ extruders. You can find the files here: http://www.thingiverse.com/thing:21788. At the moment there is no place for a solenoid because we don't use them but maybe I find time to add it in the same modular fashion.

In older BfB firmware one was able to on-the-fly adjust the print speed (not extruder RPM), this was very handy to adjust for material flow changes but unfortunately that feature has gone in recent BfB FW and my pleas to add it again are not heard. The old firmwares were too buggy in SD card reading etc, some machines refused to run on old FW without dreadful resets. So now the only way to compensate is to adjust pressure which is not as easy, especially when lowering the pressure it can take a few minutes for the pressure to lower in the system. I bought high quality pressure regulators which are much better than the ones on most (cheap) compressors. You can also place them much closer to your machine and it allows (the reason I got them in the first place) to run multiple printers from one source of compressed air. By the way, you use so few air that we managed to do a whole one week workshop with 15 students on a single charge of a large compressor.

Pressure Regulators

Two machines printing (nr 3 visible on the left)

Large print

So the good and bad points of the time-pressure method:

Advantages
-Simple straight forward design, the ease of construction of the extruder.
-Nearly instantaneous start/stop capability.
-Easy to clean.
-The extrusion is pulsation free in contrast with many other potential systems that use a pump.
Disadvantages
-Incompatibility with most RepRap electronics, Gcode processors etc which is a big issue but not impossible to solve.
-Air compressor or other source of compressed air needed.
-On/off control of the extruder by switching the air pressure is unreliable, a solution here would be to instead of switching the air pressure behind the material to control the material flow at the nozzle and leave pressure constant. In industry various valves are available that do just this and these could be easy to replicate. F@H's valve tool has a simple method to do this by using an off the shelf valve between the syringe and the nozzle (added bonus is the decoupling that I mentioned earlier), also the vintage RepRap Support Extruder 1.0 uses a similar method.
-Non-metered, the big issue. The problem with a Time-Pressure Valve system is that it depends on many variables to keep a repeatable and predictable flowrate. The main variables are pressure and material viscosity and the combination of both (in addition to friction of plunger, changing material level in the syringe etc etc. Read here for example) gives you certain flow rate. Flow rate = Material Viscosity + Air Pressure. So if your materials viscosity changes only slightly you need to compensate that with higher or lower pressure. We tried many things and found many ways to improve it one way or another and one could even program some of the parameters in the system to compensate for some known effects. But this would also mean that you need virgin syringes each time because a plunger acts differently in a used syringe, that the consistency has to be exactly the same each time and throughout the entire batch etc etc. Conditions you can get in an industrial setup but not really RepRap style. Other solution I thought of could be to meter your flow rate at the nozzle and adjust the air pressure based on that, one could use various types of flow meters but digital air pressure regulators that would need to act on those readings are rather expensive parts. You could also alter the print speed based on the flow rate within a certain 'workable' range, not to fast/slow. But basically we never found a way to get metered flow rates and from all my reading I think that it is impossible to solve this elegantly in an air pressure controlled system. Therefore unguarded operation is no option and one needs a trained eye and hand to get to the results that we have here, this is a serious drawback when you want to do production like we do.

Next episode: Auger Valves, learning from industrial solutions... I'll promise to make it shorter than this one :)

ps. you can also follow our design studio Unfold on Facebook: http://www.facebook.com/pages/Unfold/138586236204562

Virtual Pottery Wheel

We recently made a video from the Virtual Pottery Wheel that is part of Unfold's L'Artisan Electronique installation. We have already a slightly better version where you need to manually spin the wheel to spin the virtual model vs. the automatic rotation in this video but no video yet of that.

Video - Jonathan Keep

Excuse the quality of the video but I hope it can give some idea of the processes I am going through to print pots. To ensure a good flow of clay ooze during a print I remix, blend just enough clay to fill a syringe for each print. The video does not show it but the syringe is attached to an air compressor, you can see the air hose, and it is set at about 3 Bars of pressure.

Clay - Jonathan Keep

While it is early days in my ventures with ceramic printing the impression I am getting is that the clay qualities desirable in more traditions pottery techniques hold true for printing with clay too. So far I have tried porcelain clay, then a blended buff stoneware clay, a ball clay, a red terracotta clay and a black firing clay. The general rule of thumb with clay is that the whiter it is the less plastic and malleable it is. What is known as ‘short’ clay. The darker the clay, the more sticky it is and often the more plastic and able to bend before breaking. The converse is, the whiter the clay the higher the temperature it can be fired to before distorting. So porcelain is higher fired and stronger than say red earthenware red clay. There is always a payoff and that is why most clay bodies are a blend of a number of materials.

Another factor I expect will influence our choice of clay for printing will be particle size. Short clay tends to have large particles, sticky clay small particles. But then fine sticky clay dries more slowly than white less plastic clay. To support the weight of the clay as it builds up the print must dry quickly but to unsure the layers stick together and bend both qualities are desirable. To further complicate the particle size issue a range of larger particles give structure to clay wall and gives rigidity to a soft clay structure. So often crushed, already fired clay, or what potters call ‘grog’ is added to a clay body. As the ceramic printing head/syringe I am using has a 1.5 mm nozzle I have used a 80# sieved grog, adding about 10% in dry weight. Whether grog is helpful in printing clay only more experience will tell. My gut feeling has been to include grog from the start to help give a bit more structure to the soft printed clay but also as clay powder is very fine the grog gives a bit of ‘tooth’ when mixing the powder with water. Different clays naturally contain slightly different proportions of clay to water for the same consistency. As a guide I have been printing with mixes of 2.5 – 3 parts dry clay mix to 1 part water.

‘Volcanoes’ – ave size 9 x 9 cm (Experiments in modelling with ‘sculpt mode’ in the 3d program Blender)

• Porcelain clay printed well but if it is too dry it goes ‘cheese’ and the extrusion breaks apart, and too wet it cannot support itself and collapses, so consistency is vital. Porcelain prints are the only objects I have glazed so far and the print layering texture looks ok through the glaze.
• The two buff clay were best to print with. The stoneware clay is gritty anyway so I did not add grog. The other buff clay is a ball clay that offers plasticity without being too sticky. The ball clay print quality is finer and more pleasant than the stoneware clay.
• The red clay printed ok as well but depending on what type of glazes you want to use and what temperature you are going to fire to a red earthenware clay might not be desirable.
• The black clay, although it looks great fired did not print well and was very difficult to work with. In preparation the surface dries easily and it is difficult to avoid small lumps in the mix that block the printing head. It was difficult to get the slurry consistency correct. It went from being to dry for the compressor to push it through the syringe to being too wet to stand up during printing with very little water added. I did get a reasonable dark brown ‘ant hill’ print out of a 50/50 red clay, black clay mix.

‘Ant hill’ forms – ave height 13 cm (Experiments in modelling with ‘metaballs’ in the 3d program Blender)

Working Method - Jonathan Keep

Having just got my first printed ceramics outof the glaze firing I thought it would be a good time to offer a post on my current working method. On and off for the last ten years I have been using 3D computer modelling programs, mainly as a tool to train my visualisations memory and explore new forms. Using numerical transformations the computer can generate forms I would not otherwise conceive of. Scaling, particularly uneven scaling or scaling only in one direction I have always find interesting. The human eyes aptitude for responding to symmetries is another area I have explored both in 2D and 3D software. Morphology and the evolution of form is a third technique I use. The 3D program can calculate the transition from one form into another offering the possibility to capture a new form at any point along that transformation.

Recently I have started using Blender, an open source 3D program that offers great possibilities and I can see myself spending a lot of time with it. Unfortunately the most recent version does not yet have .stl export, the file type used to convert to code that the ceramic printer understands. So saved as .obj files I do the conversion to .stl in Netfabb Studio. What is useful in Netfabb is that the surface area of the form is given so I can make the necessary adjustments to make sure I will be able to print the object from one 60 ml syringe of clay paste. The saved .stl file is then opened in the BfB Axon program and finally the form is cut up to make a g-code file that goes on a memory card ready for the printer. The BfB Axon program has a minefield of settings that are required when using a self made printing head as with ceramics. This is another posting in itself and while I am still trying different syringes ( I want to be able to hold a greater volume of clay for each print) it will be a while before I know what my settled ceramic print head settings will be.

Three stages in preparation for printing – Blender, Netfabb and BFB Axon.

Morphology of glazed printed porcelain forms – height 9cm each.

Simple glazed printed porcelain scaled forms – 6 to 3 cm high.

Shipping robots

I unpacked one of our printers that came back from an exhibition in Abu Dhabi last november and found that two of the four corners where completely loose, thats four grub screws for each corner that popped out of their place so now the vertical bar slides freely in the corners.
Funny thing is that when I unpacked them in Abu Dhabi, the crate was damaged and not very rigid anymore but the printer wasn't. I fixed the crate before shipping back and now it is the opposite.

Looks like I will need to take this one apart and reassemble correctly again, he was a bit wobbly in the first place...

And I will need to look into ways to better pack the machines, I always fill the crate with foam. Also on the bottom there is 3 cm foam in which the feet are plugged so the foam sits flush with the horizontal bottom bars. Will need to look into ways to improve this.

So fellow robot travelers, how do you ship your machines?

Kilns part 1: OpenKilnController

Last Christmas I got a small, almost new 8 liter kiln from my uncle who owns a large dental lab. He got it when he bought a whole lot of material in a liquidation and it was to big for them. Small for us is large for their industry, they use tiny desktop kilns, maybe 250 cc that fit a few tooth in them. The only problem was that it was only the kiln with the heating elements and no controller, thermocouple or relays. Buying all the necessary gear through our local pottery supplier would have made us ± 600€ lighter! So having a little tinker ethos I figured that it would be easy to build something yourself using an Arduino as a controller and a Solid State Relay to switch the 2 Kw elements. Not being such a great programmer or electro-engineer I dived into google thinking that this must have been done before considering the prices of commercial controllers,  some code would probably be available to get you started. Well, google turned op close to 0 hits so apparently there are not enough potters with programming/engineering background or hackers that like clay? After doing lots of research, driving to holland to pickup some old analog industrial temperature controllers, thermocouples and relays, requesting manuals for gear from the 80s from japanese companies I finally threw in the towel and bought a 300€ controller kit that was a perfect fit for small kilns like ours. I found this kit through all the googling (so never a waste of time) and figured that it was not worth it to continue my search for a DIY solution since I really lack the skills to pull it off from scratch. The ST222 kit I bought included a controller, thermocouple and relays and was made buy the UK company Stafford Instruments. So here's our inferno in a box.

I still love the idea of a DIY kit controller, I calculated the costs of a DIY solution to be around 150€ (Thermocouple 60€, Arduino 30€, SSR 30€, Display for Arduino, buttons etc 20€) which was a quarter of the initial commercial solutions and would allow you to preprogram all curves from a computer instead of using the awkward interfaces of commercial solutions and even monitor the kiln with plotted curves on the computer. The solutions on the market are really basic in functionality. While the 300€ controller was a sweet enough deal for our small kiln I still think a DIY controller would be very interesting for larger kilns. So I hope someone wants to pick up my mission for the OpenKilnController.

Here is what I have found so far, kiln controllers are PID controllers (I never heard of them, remember my lack of skills) and they appear to be very common in industrial control systems.  There is already a pretty decent PID library for Arduino. (http://www.arduino.cc/playground/Code/PIDLibrary) that could be used as a starter. The hardware is straight forward as mentioned above, you need a thermocouple as analog sensor, the Arduino with PID library as controller and a SSR as switching mechanism for the kiln. The tricky part looks to be the 'tuning' of the P, I and D parameters.
I also found some rforum posts over at arduino about DIY Arduino PID controllers so at least some people tried like PWillard.


UPDATE: While typing this post I revisited the forum thread I mentioned above and apparently GlennD did build a kiln controller for a glass fusing Kiln. Unfortunately only after I bought the ST222.