Good news – it seems that the fab@home fabber can indeed create tiny little drops. One has to push the syringe tool about 0.1 – 0.2 units in, and then a little back so that the next drop doesn’t start to form immediately. The drops seem sufficiently small to be used for the installation, so it seems no special pumps are necessary.

The bad news is that the control software doesn’t want to take my simple hand-made STL files to create a lot of these small droplets. But everything in due time…

[Show as slideshow]

On Saturday, I demonstrated the fab@home 3D fabber to the participants of the Happy Chaos symposium, using Nutella to print shapes on bread – something inherently sweet The idea of using Nutella was actually pioneered by Frank Oxener. Though the material doesn’t allow to draw complex 3D shapes, as it melts away fast, it’s a nice way to raise awareness towards the FabLab itself.

[Show as slideshow]

The fabber kit comes in the form of a white box, a bit too big to transport manually alone, weighs in at about 10kg. It includes almost everything you need for assembly, basically all the custom-made plexiglas parts, motors, cables, etc. The fab@home site provides a list of additional equipment that you need, see here. You can follow the documented assembly process, it’s put together pretty much, though navigating between the assembly pages is quite difficult, as the navigation menu is only provided at the bottom of some of these pages, and it’s not obvious which is the first one. The process is surprisingly easy, but of course you can never get everything right at the first time.

Below I listed some issues we’ve found, and also some improvement suggestions.

Some issues with the kit

We’ve encountered the issues below when trying to get the fabber work.

Some motor orientations were incorrect

Though we’ve followed the wiring color convention suggested in the assembly tips page, at the end it turned out that actually 3 of the 4 motors had opposite orientation then expected. Don’t know how / why this happened. By reversing the cable connections from the motor controller to the motors, this issue got mitigated. We’ve had to reverse the limit switch connections as well accordingly.

We only found out about this issue when we were controlling it from the software, as the 3D representation of the kit and its actual movements contradicted each other. For example, the plate was moving up in reality while moving down in the 3D view. Coincidently, the deposition tool’s motor was also reversed, so it was actually pulling back instead of pushing the material out. No wonder nothing got printed at first

The controller support doesn’t hold the controller

It seems that the small parts that are supposed to hold the microcontroller on the fabber chassis were added as an afterthought, and they just don’t lock in properly, and thus they don’t support the controller. Maybe a better design would help in the future, so that the controller is held properly on the side of the fabber box.

A Windows-only solution

It’s not mentioned in the required bill of materials, but actually you need Windows to run the fabber software, or even just to connect to the fabber. Seeing that it’s an open source project, actually I didn’t check in advance, but I supposed it most be running on Linux, and possibly cross-platform – well, it’s not.

What’s even more surprising is that the controller is not a regular USB-Serial device that would be generally recognized by operating systems, so one can’t even just connect it to any OS and play around with it via a serial protocol. It needs a special USB driver that is available for Windows only. This also means that if you’re running Windows from a virtual machine within another OS, you need one that enables for USB pass-through from the host OS to the guest OS. (For example, the closed source version of VirtualBox does this, but not the open source version.)

It would be nicer to have the controller as a generally recognized USB-Serial device that is picked up by all major operating systems. It would be even nicer if the whole solution would run on a range of operating systems, especially Linux, so that one can create a complete fabber solution running on nothing but open source software.

Possible improvements

The following would improve the kit building experience.

Include more of the easy-to-break parts

While I know that professionals never make mistakes – me as an amateur, I sure do. And as more and more people try to get into 3D fabbing, more amateurs are entering the scene. It would be nice thus to include more of the easy-to-brake-but-cheap parts in the kit. This is especially true since internationally some are very difficult to replenish, just because of the different markets around.

For example, I ruined all limit switch connectors when putting them together, and took me about 3 days to find replacements. Unfortunately the kind of connectors used are not available here in the Netherlands, so I had to buy replacement limit switches. Fair enough – but these are actually of different shape (the connectors on it are not on the side but on the bottom of the switch), which means they don’t fit into all the locations they are supposed to on the fabber. We worked around it by creating adapters for them using a laser cutter at the Waag FabLab – but not a lot of people have this luxury.

All this could have been prevented by supplying say twice the amount of the small white connectors and the crimp contacts used, which I guess would have cost about $5 more at most, less than 0.2% of the kits overall price. And would have saved me 3 days.

Use fewer kinds of screws

It’s not apparent on the assembly tools page, but putting the fabber together requires a huge range of different screwdrivers. Moreover, it requires both metric and imperial scale screwdrivers. Thus no one is having a good time: actually you have to buy sets of different norms, it’s not enough to have everything from one norm. It’s also annoying to always look for the other screwdriver as the one in your hand is just not the one from the six kinds used.

I’m sure this all stands to reason, but maybe lowering the number of different screws in the construction would make for easier assembly and maintenance.

Provide a JTAG adapter with the kit

A JTAG adapter for the board used in the kit is not easy to get internationally. Took me about 2.5 weeks to finally get one shipped here in the Netherlands, and about €85 including shipping, even though the list price is about half at that at $55. This being about 1.9% of the overall price of the fabber kit, it might be wise just to include the JTAG adapter in it.

Of course one can argue that this is generic, re-usable equipment. But it never hurts to have one, and amateur enthusiasts are not going to have this specific adapter anyway. And yes, you’ll want to have one to upgrade the firmware, even though KoBa Industries sent the controller already flashed with some version of it.

Provide easier adjustment for the Z-top limit switch

One has to adjust the Z axis top limit switch quite often, so as to accommodate the size of the syringe nozzle in the deposition tool. This has to be done by loosening a hexagonal screw, which is opposed by a nut on the other side. Unfortunately the nut is very hard to reach because of the vertical support rods of the plate. It would make life easier if one could loosen and tighten these screws easier.

As a workaround, we glued the nuts to the limit switch, so that they oppose the screws movements without having to manually hold on to them.

Today we’ve done some more fiddling with the fabber, after exchaning some e-mails with Frank Oxener from the Texel FabLab. First the tip of the syringe hat to be clean of dried-up Nutella. Frank’s suggestion was to use a lower pressure for pushing out the Nutella, so that the plug will not push through it. Using a video link to FabLab Norway, we also found out that they were using chocolate to do some fabbing themselves. I also thought – if one can create a syringe that one can keep at constant temperature, one could really use well tempered chocolate for fabbing.

So, we tried again, but it didn’t work. After some annoyance and investigation, it seemed that the 3D representation of the fabber inside the fabber software was quite contradictory. After some investigation, it turned out that 3 of the 4 axes were connected in reverse – even though we followed the documentation properly. Well, anyway, after connecting the stepper motor cables to the control electronics, at least the 3D representation seems to be OK.

And them editing and playing about with the tool description file, and it seems to be working…

Though the thing is, that the fabber stops in between the process, ‘waiting’ for something. Well, I guess there’s still some stuff to figure out…

[Show as slideshow]

We got the fab@home fabber working today. We’ve replaced all the limit switches which had their connectors soldered incorrectly. Of course one can’t find such connectors here in the Netherlands that were sent over from the US, so I had to buy new limit switches, which are of course different in their shape, and don’t immediately fit into the old one’s positions. But this is when a laser cutter comes in handy, so that Bastiaan Ekeler drew up an adapter and then cut it out right away – and off we go.

It’s nice and productive to be at the Waag FabLab

Other then the limit switches, we had to solve the issue with some loose wires that made the disposition tool of the fabber not work.

And then off – to try the Nutella with the thing. Unfortunately it doesn’t work out of the box. The biggest issue currently is that the ‘plug’ that should push down the nutella in the Syringe actually goes ‘through’ the nutella – and thus no nutella is coming out of the syrenge, it’s just that the plug swims in it deeper.

Ah well. Someone else has made it work before, se it should be possible…

[Show as slideshow]

• some edge detectors have to be replaced, as I blew their connectors when soldering
• for some reason the syringe motor is not working

It would be interesting to make it work and experiment with it like that already.

To modify it to work in the biodisplay installation, we need to:

• get the JTAG adapter that can be used to re-flash the ARM controller
• get a pump that does 5μl or smaller drops
• do some coding, so that all this is controlled after entering a bitmap

Evan from the fab@home list suggested LEE fluid valves to be used, and we’ve seen the Bukler valves earlier. Let’s see if either of them works out.

The overall plan for the remaining 3 weeks here are:

• verify bacteria sterilization via UV lamps
• get and build the precision valve that is capable of the arabinose drops 5μl or less
• start to reprogram the fab@home 3D fabber (get a JTAG adapter for it, etc.)

During my time not here, Bastiaan Ekeler at the Waag FabLab managed to finish the fab@home 3D fabber – mostly. That is, the soldering that I did has to be re-done, so I have to get some replacement parts for those connectors.

Vincent van Rixel and Joeri Verasdonck here at the Leiden University also made some experiments with the bacteria, for example to try to create smaller pixels within the bacteria by using smaller drops (less than 5 μl) on an agar surface full of bacteria. Unfortunately the experiments didn’t really seem to have worked out.

I’ll re-grow the bacteria tomorrow on some plates, so that we have individual colonies as needed. And I’ll also look for a suitable UV light that can be used for sterilization meanwhile.

Alex Schaub will contact the valve company, so that we can find out what is needed overall for the precision valve. Then we can get along with that.

I also ordered a JTAG adapter for the fab@home fabber’s control circuit, so that I can re-program it. Unfortunately I didn’t get it yet, there’s some issues with the ordering. It would be so much nicer if they shipped the JTAG adapter with the kit itself.

So there’s quite a handful to do in these 3 weeks, and in two cities: Leiden and Amsterdam, which makes it even more complicated…

It also seems that the USB-Serial interface programmed into the firmware, while based on the lpcusb project, is still somewhat custom-made. At least the fact that the fab@home people supply their on USB-Serial drivers for Windows suggest so. And of course they don’t supply USB-Serial drivers for Linux or Mac

Today, with the very kind help of Bastiaan et al. from the Waag FabLab, we’ve pretty much finished the physical construction of the fab@home 3D fabber. Tomorrow, we’ll try to make the electronics work.

Being around people who are so experienced with building physical stuff is very rewarding. And yes, my soldering needs a lot of improvement indeed…

[Show as slideshow]

Wow..