Introduction
I am curious whether every sample prep that a robot in a DNA sequencing machine does--using pipettes and mixing plate--can be done in a credit card sized "consumable card" made of PDMS (poly dimethyl siloxane). This "card" should contain every reagent that normally goes into a DNA sequencing machine. Compared to robot based workflow, this route may potentially be faster, smaller, cheaper, and more reliable. Quake and Fluidigm have proven that such a custom made uFluidic chip can control fluid flow inside the chip, and I am sure that a straight forward application of the technology will work inside the DNA sequencing machines as well. But The only thing I don't like about the Quake/Fluidigm setup is the external air/liquid pumped into the control channel. While the control channel inside the PDMS chip itself is small and reliable, the external infrastructure necessary to actuate the control membrane do not seem to be. When I asked a co-worker who is from Fluidigm about why they are still using the pnematic/hydraulic based control, he said that someone in Fluidigm looked into piezos a long time about, but primarily because all the control workflow is done outside the Fluidigm instrument, and the uFluidic chip is a passive constant flow device when inside their instrument, such compact control methodology is unnecessary. This does not jibe with my understanding of the Quake/Fluidigm value proposition, so I must have misunderstood him--but I can't imagine how, given how clearly he explained the reasons against piezo actuators inside Fluidigm instrument; something to revisit later if an opportunity arises.In this picture (originally copied from LBNL website) of the LBNL uFluidics lab setup, the pumps, pressure regulators, tubing, and the solenoid take up the entire right half of the bench. The solenoids that open and shut the pressurized control line has not yet been miniaturized to the scale of the uFluidic chips and presents a challenge in true miniaturization and very large scale chip design. |  |
Because my master's thesis was in controlling piezo stacks, I wondered if the piezos can push down on the control lines directly and replace the bulky pneumatic/hydraulic setup like those shown above, and it seems that others believe this as well. According to a paper by Shuichi Takayama of University of Michigan, programmable braille cells can open and close the fluid channel in uFluidic chips. It seems that the braille cells from the following two manufacturers would be good candidates to evaluate:
| KGS Corporation, Saitama Japan | SC6 and SC4 are braille cells that can be driven by an external electronics. I contacted their US office and asked for a quote, but did not hear back yet. | ||||||||||||||||||||
| Metec AG, Stuttgart Germany | B11 and P16 are programmatic braille cells that can be driven by their USB controlle, which can be controlled from a Windows PC. B11 has a feedback unit built-in (I imagine for the user to press the button or something) and is self powered, while P16 is just piezo dots, and not self-powered. The schematic and physical dimension of Metec P16 is as follows:
2.45 mm dot spacing and 0.7 mm dot stroke seems rather large for uFluidics application, but as you can see from the picture, Metec's engineering expertise is in producing the piezo bimorph actuator that will apply force against pins. The support person (Ingrid) said she is eager to discuss my requirements, and the mechanism seems like it can be easily shrunk by a factor or two or three at the braille pin end (where the interface to the uFluidic chip takes place). If the vertical range of motion requirement of the braille pin is reduced, the length of the bimorph could conceivably be reduced--although this would probably entail a higher NRE.
Because they offered this convenient prototyping environment, and because they replied to me promptly, I shelled out my own money for the following parts:
|
Interfacing with the prototyping unit
How would one control the Braille cells? Ultimately, piezos deform under high voltage. So you can make a circuit that will drive the individual piezo biomorphs; in theory, you should only need 9 pins to drive an 8 piezo Braille cell (one pin being dedicated to the ground). With your custom electronics, you supply variable voltage to the piezos to control the Braille height--although the voltage vs. height deformation would have to be characterized. Ability to control the Braille stick height may be useful in the following cases:- Braille cell height achieved for the same voltage varies too much for the application. If we just want to open/close uFluidic gates, the height variability might not matter all that much.
- Peristaltic pump that has to remain at the same layer as the rest of the uFluidic channel. For reasons I don't understand yet, some uFluidic peristaltic pumps do not close the gate all the way. I will have to see whether a peristaltic pump can still work if I close the gates all the way.
Programming Metec's prototype system
As soon as Metec acknowledged receipt of the wire transfer, I asked for and got a code example to interface with the USB prototyping board. The Metec USB board implements a custom USB driver which is only supported Windows version up to XP (thank God I still have an XP laptop), so I can use the DeviceIoControl() WIN API, with Metec specified values to control it. The provided example was in C, so my first attempt will be in (Visual Studio) C++, but in theory I should be able to use any dotnet language. Perhaps my second attempt will be in C#. Roughly, a program has to go through a one-time (per application invocation) setup before sending the bitmap for a Braille cell line (where line consist of 8 Braille cells--although I will only have 2 Braille cells).| For the first pass, I only want to control the individual piezos (of the 2 Braille cells I have), a crude GUI like this is adequate. Having the "Simulate" option is a common sense approach to any HW/SW integration (my specialty), where early testing of the SW cuts down the total development time. For example, using the "Simulate" feature, I can test the "Set" button click handler even before I receive the actual Metec USB board. The intention is for the two Braille cell controls to be enabled only when the GUI is connected to the device. |  |
2014 Update
I moved over to Fluidigm (practically the only company with pressure based programmable uFluidic control in the market) in 2013, and learned my the cantilevered pushrods cannot actuate pressure gates: the high pressure required to actuate those gates. I studied the fundamentals of the Fluidigm fluidic control, and derived equations from fluidic models (sorry cannot put this out to the public domain). When I punch in the material parameters of the Fluidigm PDMS (Fluidigm has a custom formula for the PDMS) and water, and the channel dimensions of the current Fluidigm chip, the pressure required to actuate the gates and associated timescales are roughly in line with Fluidigm's seminal paper: Pressure driven digital logic in PDMS based microfluidic devices fabricated by multilayer soft lithography, Naga Sai Gopi K Devaraju and Marc A Unger, Lab Chip 2012 Nov;12(22):4809-15. A directly driven piezo stack can generate enough force to overcome the fluid pressure at the gate, but then one runs into the problem of squeezing in enough piezo stacks in the relatively small area of the instrument-to-chip interface. Why do I want to keep the piezo stacks in the instrument rather than integrating directly into the chip, as is the case in semi-conductors? 1 word answer: COGS.
