Crucible

2016

After long months of silence I am finally able to talk about the research I was involved at FATHOM. Before I go into details however, let me give some background information. FATHOM is an advance manufacturing studio based in beautiful Oakland Ca. They do a lot of things very well, from CNC machining, to injection molding and a lot of 3D printing. They also have an outstanding team of engineers, designers and fabricators to bring anything you can imagine to life. 

What makes Crucible? 

Crucible is an important partnership between UC Berkeley, NASA Ames Research Center, Autodesk and FATHOM. The entire project is the effort of many bright people, among them Aaron Berliner and Carlo Quiñonez who together really are the minds and hearts behind Crucible. Aaron is research scientist in the Bio/Nano group at Autodesk, also a visiting research at the Arkin Lab at UC Berkeley, check out his full bio here. Carlo is the Director of research at FATHOM, I write more about him later in the post.

So what is Crucible you ask?

Crucible is a 3D printed incubator capable or simulating extreme atmospheric conditions. Similar to the ones found in... lets say... Mars! "Mars in a jar". 

Ok ok, but why?

Why is a project like Crucible even worth making? It all starts with a little science fiction, the idea of one day living in another planet. One of the biggest challenges in Synthetic Biology is the terraforming of other planets, Mars being the place to start. There are probably a handful of way of doing so, in reality the solution might be the combination of multiple methods. In this particular case, Crucible seeks to explore terraforming through life. 

This means trying to engineer biology into surviving the harsh conditions of Mars, then have those living bacteria transform the planet into something where life as we know it can thrive. 

Crucible is a step in the process since it helps emulate these harsh conditions in Earth, allowing scientists to conduct experiments at an accelerated pace with no need to leave Earth.

Goals and Functional Requirements

• The incubator shall withstand vacuum of 0.5kPa and will be maintained using a closed loop control system.
• The temperature of the incubator shall be variable between -100C° and +150C° and will be maintained using a closed loop control system. Furthermore, the external temperature will be recorded.
• The interior of the chamber shall be lit with variable intensity from 1000 𝑊/m^2 to 100 𝑊/m^2 and will be maintained using a closed loop control system.
• The chamber shall allow the use of custom pre-blended gases.
• The chamber shall support a web API allowing for remote operation. 

Why 3D printed?

If you think about it, crucible is somewhat similar to a fridge, or even an oven, nonetheless, you don't usually see ovens or fridges made out of plastic, or 3D printed for that matter. So why did we wanted to 3D print it?

• Inexpensive, allowing for future farms of incubators conducting parallel experimentation
• Easy to share with scientific community, print files could be attached to a research paper
• Possibility to adapt to specific experiment needs and requirements
• Wider reach in open source community
• Accelerated design process through testing and fast iterations
• Steps closer to earth independent manufacturing

How does it work?

Lets over-simplify things for a second. Basically you have a very cold fluid going through 3D printed channels wrapping around the chamber. After the fluid has circulated around the chamber it exists back to the cooling unit to be re-cooled and recirculated. 

Sensors are constantly measuring the temperature, pressure and light intensity inside the chamber. A Raspberry Pi 3 paired with an arduino can change the temperature of the fluid, control the amount of pre mixed air (or vacuum) inside the chamber, and finally, adjust the current to the LED to control the light intensity inside.

In an effort to keep the temperature inside the chamber as controlled as possible, we printed a thick insulating cover that seals all around the chamber.

Air Tight, water tight  and Vacuum Tight

If you have ever worked with FDM machines (like a reprap, a ultimaker, a makerbot or even a fancy Fortus), you probably know the parts often have tiny pores by nature, especially at the corners, at locations where the raster joins the contours, or where details are too small for a contour to fit. 

A lot of my time was spent developing ways to print sealed parts right out of the machine or with minimal post processing. The image below shows the top half of the chamber. The channels going around would carry the temperature fluid while the cone structure would ultimately be capable of holding vacuum.

Evolution

The chamber didn't always looked liked this. In fact, the form is the result of the evolution and iteration 3D printing offers. The form purely follows function and the process took perhaps a little over 20 different designs. Many of the samples may look very similar but are distinct in the parameters used to print them. Perhaps the contour thickness changes, perhaps the overlap between contours, or simply the number of contours, etc.

The testing performed were usually simple. Prints were submerged under water with pressurized air running through them. This simple method is effective in pinpointing medium size leaks. Observing prints under the microscope helps to see how regular the contours and paths are, not so much to find leaks. Finally, I would run a more quantitative test by pulling vacuum through the channels and chambers and measuring the pressure rise over time. Some images of the tests can be seen bellow. 

RASTERS VS CONTOUR

Pretty fast we realized that the most successful prints had a smaller percentage of rasters vs contours. In fact the most successful prints had very little rasters. But what are rasters and a contours anyways?

Rester refers to a toolpath designed to cover an area by going back and forth in straight lines (above image, left). Contours are lines that follow the same shape as the outside perimeter but constantly offsetting until the area is fully covered (above image, right). The circle to the right is also randomizing the connections with between contours, this prevents creating a single path for fluid (gas or liquid) to escape through. The concentric rings of the contour fill also helps retain any leakage that may exist, making the part superior at holding a seal.

Contours Only

If you have ever worked with 3D printing software you probably have been a little frustrated with the little control you have over the toolpaths, same goes for the infill but that is topic for a different section. Insight, the software form Stratasys for slicing and creating cmb (print) files, has a relatively high level of customization, at the end however it wasn't enough for our needs. 

One of the biggest advantages of going through this process is that we can verify the toolpath not only layer by layer but more importantly, in a 3D context. We can see how layers contours will align with the previous layers or the ones after. We can add contours where we need more or remove them if they will harm the print. 

Once in Insight, we just match the settings used in the plugin to slice our model. In other words, very little work is left before we can hit that print button. Still, it is good practice to go layer by layer to make sure everything is how it is supposed to.  

INSULATION

As you probably realized already, keeping the temperature inside the chamber very different form the temperature outside one can be quite tricky. To solve this we designed 3D printed geometries to prevent temperature flows. Plastic is a great insulator, air a much better one. Air by itself will create convection currents lowering performance of the insulator, to avoid this the insulation is composed closed cell chambers of trapped air. In future iterations, we might remove the air and cells inside the insulation creating a vacuum. This would vastly reduce thermal losses due to conduction and convection. 

Autodesk has an extraordinary suite of software, among them is Autodesk CFD, which can be used to predict product performance, optimize designs, and validate product behavior.  

Electronics, sensors and communication 

There are two boards that together make the electronics for Crucible. To avoid any confusions lets say there is a main board and a peripheral one. In the image above you can see the main board (top view and bottom view) as well as the peripheral one designed to go inside the chamber. 

The main board is equipped with an Arduino Micro (ATmega32U4) in charged of reading all sensors and controlling all valves as well as the LED. The main board is also equipped with a Raspberry Pi3 that constantly communicates with the arduino and the client.

Future iterations could get rid of the arduino and just use the Pi but it was easier to start this way, specially because we are using a bunch of libraries for which the arduino code already exists. 

The boards were design by Carlo Quiñonez using Circuit Maker, an (free) PCB design software that allows for 3D modeling and revision control similar to Github.

Piecing it together

In all, Crucible is made out of a couple of systems. The picture below helps visualize how things fit together as well as what parts are printed with what technology. One of the most geometrically complex parts is the gas manifold, it connects to the valves and channels the flow of gas, vacuum and atmospheric gas (air) in and out of the chamber. Do to these complex requirements it was printed using an SLA machine. The gas manifold was designed entirely by Carlo Quiñonez. This is a good moment to talk more about Carlo, he is the Director of Resarch at FATHOM. He is incredibly knowledgeable in additive manufacturing,  electrical and mechanical engineering, has a PhD and a post-doc in Biology and has a passion for hardware, specially in additive technologies for biology. He is without any doubt the reason I got involved in the project and the reason why I would do it again. If you ever have the chance to work with him do it! 

Back to Crucible. As I was mentioning before, Crucible is made of of a couple of systems. The following diagram serves to illustrate how everything is connected as well as what sensors and equipment we chose.

Once everything is assembled 

At this point we delivered the project to Aaron for him to continue his research at the Arkin Berkeley Lab.

Want more? You can watch this video of Aaron explaining things in more detail. Especially the science parts that I didn't really talked about.