Baker Adaptive Optics
Experienced Remote and Extended On-Site Electro-Optical Engineering
Advanced Microcontroller Interfaced Photonic Sensors Design
Optical Component System Identification Services where we Measure/Model/Validate via MATLAB Tools & NiDAQ hardware
I'll be putting fast feet on the ground, sharp, cautious eyes on your specifications, steady hands on the optics, instruments and controllers to get your innovative E-O project specified/assembled/aligned/working. And If the past is any indicator, we'll be on-time with PPT slides with fresh data to prove it. It's my job to provide you with the sound advice and professional guidance that experience brings every step of the way. Take a look at my site and see all that I can do for you this FY and thereafter.
Affordable, Convenient and Optimal Success
Optical Sensors Division
Everyone agrees that sensors set the overall performance limit of many E-O systems. Is your photodiode & associated TIA more than a few years old? I'll bet that it's obsolete--this field is changing FAST by things such as AI and self-driving cars. Photodiodes are all going to surface mount, making everything faster/lower noise. The IC's are getting smaller, faster, and switching to lower voltages. Most users aren't aware that circuit topologies slightly more complex than typically used can yield 2-10x improved results, but no big supplier wants to bother with tuning. And if you're making your own, you've probably discovered that naive PCB layouts can quickly destroy any chance for high performance, or cause crippling parasitic instability. It is documented that even some types of basic components can have more noise than others. Power conditioning is getting better all of the time, often limits performance, and these can now be made to match your setup without clutter. Output format and connector type options such as single-ended, differential, coaxial, triaxial, SPI or I2C, or microcontroller digital help integration. Until recently there's been little/no ability to reconfigure parameters such as pre- or post gain, feedback network, or post filtering. Now it's quite doable and easily integrated into MATLAB, Python, Labview, C++, etc. To summarize, I can provide my clients with a wide range of modern photodiode/TIA configurations, or we can try one that you've researched online. Or, let me suggest one since I've probably already studied and/or tried an approach that would work for you. To find out more, get in touch.
On-Site Electro-Optics Services Division
The R&D cycle of life: Propose something, get funded, do something, collect data, present results, Repeat. I can help with that. Maybe you are over-funded and need a hand. Maybe you need some expert support to fill a gap. Know it that Electro-Optics is my life and there are no projects more challenging than those which I've accomplished. Give me your most difficult and delicate work. I've aligned multiple telescope segments to perfect nanometer accuracy for multiple customers. I've tuned laser systems requiring tens of stable servo-control loops. From coherent fiber laser beam combiners blasting through steel in a shower of sparks, to active bowtie optical cavities creating light to shoot into space, to the adaptive optical control of ultra-short pulse lasers--I've done it all. From coude' feed telescope systems to HEL optical trackers' AI training systems, my clients are my number one priority, and my service history proves just how critical I've been to their successes. I've pushed forward many a stalled project, solving problems using what is at hand, and performing alignments, working interfaces and code, control systems and mechanical devices, resulting in the collection of data sets and generation of professional reports upon completion.
Talk to me about the predictable, linear process: I'll come in, listen, study, and recommend a few parts, if needed. Then I'll come back and get it done, collect data, write it up, and get out. All within a season. You show the results, and get funded. This is the cycle of life.
Controls and Automation Division
I've always been intrigued with electro -optical and/or -mechanical control systems. Think Simulink or Xcos. Device transfer functions with their poles and zeroes, bode and root-locus plots, compensators and their phase margins, overshoots and undershoots, anti-windups and slew rate limiters, it's all important to understand. The overall interplay of compensators, sensors, and disturbances in the time and frequency domain make for a challenging simulation and even more challenging implementation. Automation, that is, making things move based on sensors and programming, is quite enjoyable to dive into. The required linear/rotary DC, servo, stepper motors, PZT and voice coils are all unique in their applications and quirks, but quite a snap to control with today's tools compared to yester-year. Integrating live video, IRU's, accelerometers, gyros, metrology, temperature, pressure, strain and optical gauges and processing makes for quite a high level of functionality.
Implementation? This is where it gets exciting. Think Analog + Microcontrollers. Seriously. I've built and tuned many an old-school analog PID, and pure, PC based controllers relying on PCI-e DAQ boards, and hybrid solutions using circuits and supervisory DAQ to help out with known control systems "situations". It works, but you need a big box PC, expensive DAQ boards from NI, C++ compiler suites, etc. And there's a big learning curve, and cumbersome cable interfaces. Advance to 2021: NOW there are tiny, professional grade, inexpensive microcontrollers (~$100) with extensive DAQ and sensor options that boot fast, can run loops at tens of kHz using high resolution A/D's and D/A's, and do it using a few mA of current in a package the size of your thumb. These products are well supported, free to program, and already user controllable via popular software such as MATLAB, LabView, Python, on the web, etc. And getting faster and more capable every day. Combine these with a custom PCB carrier, modern analog IC's, maybe embedded lock-in amplifiers and adjustable filters and you've got a dynamic and configurable winning solution. Some micro's are even AI-chip and FPGA capable using Edge and TinyML. With low cost machine vision (MV). Quite powerful.
Towards helping clients, I can help with all of the above and am current and experienced. But for any of this to work I've particularly invested in the hardware/software to perform Systems Identification as defined by MATLAB/Simulink and their toolboxes. Accurate mathematical models are crucial to the creation of control systems be they feedforward or feedback, neural net, etc. This is how it works: One applies inputs to a device, and records it's responses. (This is a little harder than it sounds.) Then the System ID Toolbox and App will help me fit the best model to the data, and also give a measure of confidence. Finally we validate the model with new data. Let me help you characterize your components using the tools which I've acquired and learned to use. The data can either be done here in my lab or on-site. After I'm done with the data collection and analysis, I'll formulate for you a model of your device and detailed report which you will use in creating your control system.