Hardware Reference
In-Depth Information
Feedback from
sensors
f
T
Y
e
s
Input from the
signal generator
T phase?
Frequency divider
No
f
D
/M
f
Fig. 4.2
A tunable frequency-divider controlled by the feedback from sensors
•
Method I—Software implementation:
Since the output frequency of the signal
generator can be controlled by the software, the frequency can be adjusted
dynamically during the execution of the bioassay. Recent work in a different
context, viz. to understand the reliability impact of multiple frequencies, has
demonstrated the feasibility of such a hardware setup [
16
].
•
Method II—Hardware implementation:
The tunable frequency-divider con-
trolled by the feedback from sensors can generate control signals with more than
two distinct frequencies. A tunable frequency-divider can be implemented on an
FPGA,asshowninFig.
4.2
. It will not lead to any extra cost when handling more
than two frequencies.
•
Method III—By adjusting the actuation signals for electrodes
: For an electrode
that is driven by a signal generator with a fixed pulse frequency, we can adjust
the frequency of switching on/off by changing the actuation sequences of the
electrodes. For example, we assume that the actuation signal “1” means switching
the electrode on, and signal “0” means switching the electrode off. Then, by
applying a signal sequence such as “11110000”, the frequency with which the
electrode is switched on/off will be one-fourth of the frequency associated with
the signal sequence “10101010”.
In the D/M phase, the biochip operates at a nominal frequency (for example,
f
D/M
D
8 Hz), while in the T phase, higher-frequency signals (for example, f
T
D
16 Hz) are applied to the electrodes. The time spent on dispensing and transporting
droplets can be significantly reduced with no additional degradation of electrodes or
any additional cost in hardware.
It is important to note that, the above methods can be used to generate two
or more working frequencies for biochips. For simplicity, two distinct working
frequencies are used in the simulation for D/M and T phases of the bioassay in
this paper.
Since the time spent on each dilution/mixing operation is determined by sensor
feedback rather than a pre-determined module library, we can derive a “semi-
deterministic” design for biochips when considering timing uncertainties. The
design includes the synthesis result (Sect.
4.3
), and droplet transportation paths in
each T phase (Sect.
4.4
).
