Week 7: Interim Review & Decisions

This past week the Ultrafast Imaging team gave a Midterm Review featuring the LMU Seaver Electrical Engineering faculty. Information was presented in three stages starting with the basics of the system including the principle of interferometry and a block diagram. Secondly, the capstone objectives and limitations were discussed. Finally, we introduced our plans to alleviate the bottlenecks, implement a new PC system, and revamp the current code to utilize GPU-acceleration. In light of our presentation, we updated our website with new information to detail these limitations and our plans to solve them.

In terms of the system, we practiced aligning the laser specifically for the free-space collimator stage. To do so, we set the fixed position collimator to a certain height and measured it with a ruler. Then, we matched this height with the other track-bound collimator. We then used a spirit measure to ensure both were perfectly in line and level. Then, we used an Infrared laser and shot it through one end with the objective of seeing it, to some degree, come out the other end. Once this was found, the track-bound collimator’s fiber was fed to a power meter. By adjusting the knobs, we tried to produce a reading as high as possible. Then, since the collimators were relatively aligned, we attached fibers properly so that the system could be run and the interference could be seen on the oscilloscope. We then slowly adjusted the distance between the collimators, intermittently stopping to realign the collimators, until the signals overlapped and the amplitudes were visually summed.

It can be seen in Table 1 that the clock speeds are the most important aspect of choosing a CPU, followed by single-core benchmarks. This is because this is based on the notion that the GPU will be taking most of the workload. While with a CPU alone, more cores would be more beneficial than higher clock speeds, with a GPU the opposite is true. This is because the multi-core tasks are done by the GPU’s thousands of cores. Furthermore, the benchmarks provided by Geekbench tests the CPU’s computing performance on both single-core and multi-core tasks. With that, the AMD Ryzen 9 3900X was the best option. It not only has a competitive clock and boost clock speed, but also a strong benchmark score. In addition, with the 12 cores it possessed, we will get a high initial fps boost as a bonus. With this, we finalized components for the PC and submitted the list, a house of quality and a decision matrix for approval.

For the next week, we look to begin understanding the second probe that was implemented and hope to aid it in its full integration. Additionally, we will troubleshoot the oscilloscope and PC connection with new Ethernet and USB-C cables as well as a new oscilloscope. We also aim to do preliminary calculations with the 500 kHz max scanner and make comparisons in order to draw conclusions on which to use going forward. The 500 kHz scanner will drastically improve our data output speed, however, the field of view takes a significant hit. We plan to find out whether taking that drop is FOV is worth it or if we need to devise a workaround.