Preliminary Detailed Design
Team Vision for Preliminary Detailed Design Phase
During the detailed design phase, the team planned to continue proof of concept and prototyping studies with both the optical and the conductive ink systems. In particular, we planned to find additional ways to test the conductive ink system and establish its feasibility in the absence of an MRI. For the optical system, the team intended to develop a prototype board that incorporated a photodiode with superior noise equivalent power to that in the lab currently. Additionally, we hoped to establish the feasibility of removing the metal packaging of a commercial electro-optic modulator and repackaging it in a new, MRI-compatible housing without damaging its fragile components.
After receiving the parts for the prototype board, the team encountered numerous challenges in interfacing the photodiode with the fiber optic that were unable to be overcome within the time constraints of the design phase. However, by optimizing amplification factor, phase plate orientations, and laser power with the existing electro-optic modulator, we were able to discern a 15 uV signal from noise - a significant improvement over the ~80 uV obtained in the prior phase. Additionally, the team established a relationship between laser power and baseline noise, and this will ultimately allow us to purchase a less expensive laser for our system.
MRI-compatibility testing was performed by taking measurements in an NMR on campus, Unfortunately, the team soon discovered that the absence of magnetic field gradients in an NMR prevents us from testing the signal distortion induced by an MRI. The team was able to collect data for NMR peak distortion as various materials were introduced into the sample tube, which will help inform our material selection in the final design. Finally, extensive foam electrode testing was performed in an attempt to verify the team’s previous finding that foam interfaces on the electrode increased the signal to noise ratio of alpha waves during EEG collection.
Throughout this phase, the team began to better understand the challenges associated with each proposed system path, and the testing results are expected to aid in our final concept recommendation.
System Architectures, Engineering Analysis, Simulation
Universally Fit & Movable Electrode Cap Design:
- ¼” diameter PVC tubing
- Small zip ties
- Allows for 10-20 system to be utilized
- Allows for electrodes to be clipped in place
- Comfortable to wear for long periods of time
- Needs better method to tighten tubing around ears
- Needs to be designed for used with a chin strap
The Cap Design Document can be viewed here.
Testing Data and Analysis
The Testing Document can be viewed here.
The goal of these tests was to observe the effect that a surrounding magnetic field would have on various samples of materials. The materials used were tinned copper wire, conductive foam, conductive ink, single-mode fiber optics, and shielded tinned copper wire. Each sample was placed into a glass tube, and the tube was then filled with water to a certain line. This line designated the end of the active region of the NMR.
Each sample tube was placed into the NMR, and a single rf pulse was applied to read the interference caused by the material placed in the water. The samples that were tested, and their corresponding experiment number, are listed below:
Experiment 13: Only water
Experiment 16: Gray Wire w/ Insulation
Experiment 17: Multi-Stranded Red Wire w/ Insulation
Experiment 18: Gray Wire w/o Insulation
Experiment 19: Gray Wire w/o Insulation: Take 2
Experiment 20: Gray Wire w/o Insulation: Take 3
Experiment 21: Gray Wire w/ Insulation: Take 2
Experiment 22: Conductive Foam
Experiment 23: Conductive Ink
Experiment 24: Fiber Optic Wire
Experiment 25: Conductive Foam in ⅓ of Active Area
Experiment 26: Conductive Foam: Take 2
Experiment 27: Invalid Data
Experiment 28: Single-Strand Red Wire w/o Insulation
Experiment 29: Black Wire w/ Shield and Insulation
Experiment 30: Gray Wire w/ Insulation: Take 3
Experiment 31: Gray Wire w/o Insulation: Take 4
Experiment 32: Gray Wire w/o Insulation: 0 degree rotation
Experiment 33: Gray Wire w/o insulation: 45 degree rotation
Experiment 34: Gray Wire w/o insulation: 90 degree rotation
Experiment 35: Gray Wire w/o insulation: 180 degree rotation
Experiment 36: Gray Wire w/o insulation: 270 degree rotation
Experiment 37: Fiber Optic: 0 degree rotation
Experiment 38: Fiber Optic: 90 degree rotation
Experiment 39: Fiber Optic: 180 degree rotation
Experiment 40: Fiber Optic: 270 degree rotation
Experiment 41: Gray and Red Wire, both w/ Insulation
Experiment 42: Black Wire w/ Shield, w/o Insulation
Experiment 43: Black Wire w/o Shield or Insulation
The following are observations from comparing Experiments 13, 16, 17, 23, 24, 28, 35:
- Water is used as a control measurement, where a similar amplitude and similar or thinner width is desirable.
- The Fiber Optic response is seen to be very similar to the water response, only shifted on the x-axis. This points to the Fiber Optic having little interference with the NMR reading.
- The quality of the response then degrades in the following order: Gray Wire w/o Insulation, Red Wires w/ Insulation, Gray Wire w/ Insulation, and Conductive Ink.
- The Single-Strand Red Wire w/o Insulation appears to be overall more desirable than the base water sample. While it would be understandable that this sample would be better than the Multi-Stranded Red Wire, as there is only one strand in this sample, it is rather odd that it appears to be better than water.
- The Gray Wire turned at 180 degrees was used as it yielded the best results for non-insulated gray wire. This does bring up the possibility that other conductive materials might benefit from a rotation. The rotation is further discussed below.
The following are observations from comparing Experiments 13, 22, 25:
- Observing Experiments 22 and 25, it is concluded that the response of foam is quite unpredictable, and unstable. This note will be taken into account within the project decisions.
The following are observations from comparing Experiments 32, 34, 35, and 36:
- Observing the results from these, it can be seen that the rotation of the sample does make a significant difference in the response of the NMR readings. Referencing this back to the results seen in the initial Gray Wire w/o Insulation response, that response can be discounted.
- These results do bring up the question on the success of previous Experiments, and how much the rotation might change their response. Experiments 37-40 were run to check if the rotation of the apparatus holding the sample was the cause of the distortion, but it was concluded that this was not the case.
The following are observations from comparing Experiments 29, 42, and 43:
- The wire used was a couple of thin wires coiled in the center, surrounded by a layer of insulation, then surrounded by a shielding layer (likely normally connected to ground), and then by another layer of insulation.
- These plots show that this type of wire is absolutely unacceptable for use in this type of application, as the response is nothing close to what was expected.
- It is hypothesized that one reason for the bad response could be the width of the wire not allowing for much water to exist within the active area of the sample.
The Optical Calculations can be viewed here.
The Optical Results can be viewed here.
The first conclusion drawn from testing modulators was that the power of the recorded noise signal is linearly dependent on the laser power fed into the device. As a result, increasing the power of the laser beam will not improve system performance as much as originally believed. This linear trend in noise power (RMS) vs laser power is shown below in Figure 1:
The best-fit line for noise versus laser power has been incorporated into the mathematical model to allow better prediction of system behavior in response to potential lasers under consideration for purchase.
Next, system performance was studied in response to laser power at various input signal strengths. The signal to noise ratio (SNR), defined by the RMS^2 Signal / RMS^2 Noise, was found to have a power-law relationship with input signal strength, as depicted in Figure 2:
By the Rose Criterion, the minimum detectable signal to noise ratio should be roughly 5. This was born out by our testing. Figure 3 compares two signal responses, one for a 15uV signal and the other for a 10uV signal. The 15uV signal has an SNR of 5.05, while the 10uV signal has an SNR of 3.74. While the weaker signal does make an impact on the system, the stronger can be seen with markedly greater confidence.
To get a sense of the impact of laser power on system sensitivity, a variety of laser powers were tested and their signal to noise ratios compared. The results of this comparison are given in Figure 4.
As Figure 4 demonstrates, there was very little difference in performance between any of the laser powers 2mW-6mW. The 1mW laser (and to a greater extent the 0.5mW laser which is not pictured) had noticeably poorer performance. For the laser powers which performed well, all had a signal to noise ratio of approximately 5 at 15uV, with the 6mW laser taking a slight lead with SNR=5.05. Hopefully when the modulator is swapped out for our final device (which has a lower Vpi), the sensitivity will drop even lower. This, coupled with a more precise laser as well as more sensitive detectors, may drive our performance down to the desired 1-5uV sensitivity range.
Dr. Preble’s old modulator was partially depackaged in the machine shop to determine the layout and specifications of the inner components. The device was left partially packaged until a plastic case could be 3D printed and the device could be transferred to its new, MRI-compatible housing.
Foam Electrode Testing
Goal: To test the use of foam-electrode and foam-fabric-electrode interfaces in EEG collection.
- Disposable electrodes were used for data collection. The ground electrode was placed on the subject’s left temple, positive on the right temple, and negative behind the right ear. For each test condition, approximately 15 seconds of EEG data was collected while the subject’s eyes were open, followed by an additional 15 seconds of eyes-closed data collection. Five of these 30-second trials were conducted per condition, and the best three of five datasets were used in the final signal-to-noise calculations. The test conditions were as follows:
- Electrode cups
- Electrode cups + fabric
- Electrode cups + conductive foam
- Electrode cups + conductive foam + fabric
- Electrode cups
- Electrode cups + fabric
- Electrode cups + foam
- Electrode cups + foam + fabric
Full Prep (abrasive with gel)
- Electrode cups
- Electrode cups + fabric
- Electrode cups + foam
- Electrode cups + foam + fabric
Data was analyzed for SNR by taking the RMS while the eyes were closed, the RMS while the eyes were open, and dividing the two. The average and standard deviations for each set of three SNRs were computed and compared to the other test conditions.
The procedure document can also be viewed here.
The Foam Electrode Results document can be viewed here.
Initially, a test was performed with the disposable electrodes on both hairy and non-hairy sites, and three datasets were collected for both the conductive foam and the disposable electrodes at each site. Results are displayed below:
The team ultimately switched to testing with the reusable gold-cup electrodes to save resources and because they are MRI-compatible, unlike the disposable electrodes. Here, both the foam and the gold cup electrodes performed poorly with no skin prep, but the performance of the foam was equivalent to or better than the gold cup electrodes after the skin was prepped.
A larger-scale study was then planned to test different variables (no prep, alcohol prep, full prep, and electrode only, fabric, foam, and foam + fabric). Unfortunately, the gold cup electrodes were not working, so the study was instead run on a non-hairy site with the disposable electrodes to test the different interfaces between the electrode and the skin. The results are displayed below. In contrast with the previous experiments, the foam and foam + fabric systems consistently had the worst signal-to-noise ratios. Further testing is needed to understand the discrepancies between experiments.
Bill of Materials (BOM)
The Bill of Materials can be viewed here.
The Engineering Requirements Document can be viewed here.
The Test Plan can be viewed here.
The Risk Assessment can be viewed here.
Plans for next phase
The main goal for the next phase is to hone in on the major and minor risks associated with each design concept and to make a final decision to move onto MSD II with. To do this, more testing will be necessary, specifically on the new electro-optic modulator to obtain the minimum discernable voltage for an unbiased modulator. We will also compare the laser transmission to that of Dr. Preble’s modulator with the goal of determining if laser splicing created too many losses in the new modulator. A wavelength sweep will be performed on this new modulator to determine the optimal operating laser wavelength. Moreover, designing and 3D printing a new case for the de-packaged modulator will also be required to test prior to making a decision on the optical design. Testing to verify functionality in the new packaging will be ideal.
The group also plans to work with the REMCOM SAR modeling software to establish its capabilities and see if it can be used to estimate the SAR of the modulator electrode plates and the available conductive ink options. Work on photodiode interfacing will also be completed, attempting to temporarily align a fiber optic with the active area. Different materials for the cap design will be investigated with the hopes of finding a better method to keep the cap on the head. Additional work on the conductive ink wires will be completed, testing the RTS theory and different methods of shielding the ink. The group will also begin the IRB approval process by completing the required online training and submitting the initial documents.