Design Process

Proof of Concept:

The FlexiForce A201 sensor was attached to ~25 feet of shielded twisted pair cable, and plugged into our two-source 1st order active low pass filter with a cutoff frequency of 68 Hz. The active low pass filter could amplify the signal (by changing the resistance of the 1 kOhm trim potentiometer) which allowed us to vary the dynamic range of our device, and was designed per guidelines provided by Tekscan (the FlexiForce vendor). The two-source filter was initially chosen due to its ability to cater to a larger dynamic range.

As a proof of concept, we decided to apply weights to the device to see if the output was linear as per the sensor spec sheet and to see the affects of changing the trim potentiometer resistance on the dynamic range.

Figure 2: (a) The initial calibration test set-up. The "puck" was used to ensure the load was fully applied to the sensing area of sensor.

                (b) The breadboard with our two-source, first order, active low pass filter (68 Hz cutoff frequency setup pictured).

                             (c) The circuit diagram of the breadboard pictured in Fig. 2(b).

                                                          (a)                                                                                            (b)                                                                                                (c)

Figure 3: Video showing calibration of our device using the two-source filter, and the step-response as each weight was added.

With each weight, the Vmax (second value from the top on the oscilloscope display) increased, as expected.

From our proof of concept efforts, the range was found to be 0.1- 45 N for the 500 Ohm trim pot resistance and 0.1 - 25 N for the 1 and 2 kOhm trim pot resistances.

Figure 4: Plot of applied force v. voltage output for calibration using the two-source active low pass filter with a 68 Hz cutoff frequency.

Range for 500 Ohm = 0.1 - 45 N. Range for 1 & 2 kOhm = 0.1 - 25 N.

Further Calibration Efforts:

Per a conversation with our sponsor, we learned that the sampling rate would likely be 40 Hz, so we changed our circuit to feature a 32 kOhm feedback resistor, giving a cutoff frequency of 10 Hz. Calibration efforts with this circuit resulted in Figure 5. Around trim pot resistances of 600+ Ohms, the device began to max out the 5V before the last couple of masses were added which is why those points were omitted from the figure.

  Figure 5: Plot of applied force v. voltage output for calibration using the two-source active low pass filter with a 10 Hz cutoff frequency.

At this point, we realized that the two-source filter provided the extra complexity of needing two power sources, so we attempted the single source circuit to see if that would be sufficient. Initial calibration attempts proved that a single source set-up was adequate for our requirements so we proceeded forward with the single source circuit with a 10 Hz cutoff frequency. We modified the trim pot resistance so that the device could handle the full 0.1-100N range. For sanity checks, we performed calibrations comparing twisted pair v. coax cables (Figure 6a) and calibration with just the puck v. with the ergonomic grip (Figure 6b) to check for any differences between the two. The calibration comparing the puck and ergonomic grip was our first attempt at calibrating for the entire 0.1-100N range.

Figure 6: Plot of applied force v. voltage output for (a) Single source circuit comparing Twisted Pair & Coax cable (dynamic range of ~0.1 - 75 N)

(b) Single source circuit comparing calibration using just a puck v. the ergonomic grip (dynamic range of ~0.1 - 115 N).

                                                                                              (a)                                                                                    (b)                                

Figure 8: (a) CAD of the second prototype (b) Exploded view of second prototype (c) Group member holding second prototype

                                                                    (a)                                                                (b)                                                                            (c)

MRI Testing:

The true test...does it work in the MRI? The team got a chance to test out our device in the MRI while scanning a phantom (a sphere of water that's imaged instead of a brain during experiments like this) to find out the answer to that question.

Figure 8: Video of one group member applying known weights (textbooks that we weighed) to our device while the MRI is scanning.

Overall results:

• The single source, first order, active low pass filter successfully attenuates noise by a factor of 32 (see Fig. 9a)

• The calibration from Fig.6 works well to convert output voltage to force (see Fig. 9b). Discrepancies are likely due to the fact that we forgot to condition our sensor prior to use and because the textbooks were bulky and hard to balance on the small device so the load was unevenly applied.

• The twisted pair cable does a better job (2X better) of attenuating noise than the coax cable does (see Fig. 9c)         

   Figure 9: (a) Comparison of filtered v. unfiltered baseline noise for twisted pair cable (b) Comparison of output force calculated via calibration equation (blue) & actual weights applied (red)

(c) Comparison of filtered baseline noise for twisted pair and coax cable