I’ve always been interested in the tradeoffs that designers make when defining topologies and selecting components for a project. After all, making tradeoffs is at the core of the design process, and it involves a long list of objectives and constraints related to size, weight, power, cost, reliability, production, risk, and more. Each design has different “weightings” of these attributes depending on the overall objectives and must-haves.
That’s the reason we often see “ancient parts” — perhaps 5, 15, and even 15 years old — in many designs, especially in the analog and power sections. This makes sense when the risk of unpleasant surprises when using new parts may outweigh any incremental improvement they provide, while the older parts are known and stable. As a senior designer explained to me, it’s sort of a 90/10 rule, where you want to use only a few newer “brighter” parts (10%) and the unknowns they bring, and stick with time-tested ones for the 90%.
Older components have a track record with respect to performance, idiosyncrasies, application support, and availability, so they are often a much safer bet. In contrast, newer ones may offer more features and better performance, but there may be hidden aspects that are apparent only after many units have been “in the field,” and they may have early-stage production problems as well.
The riskiest option, of course, is to go with a custom-designed part for the highest performance and best match to the application objectives, assuming the up-front engineering costs and time aren’t excessive. However, this option comes with many well-known uncertainties and risks. It’s a version of the classic “make or buy” decision.
What about the JWST?
That’s why I was fascinated to see some of the component details that led to the spectacular success of the James Webb Space Telescope (JWST), launched in December 2021 in a project partnership among NASA, ESA (European Space Agency), and the Canadian Space Agency (CSA). There’s lots of good information available via these partners and other sources, including photos of its design and assembly as well as high-level block diagrams highlighting it many subsections, as seen in Figure 1.
Looking at the systems, subsystem, boards, and individual electrical and mechanical components – most of them fully or partially customized, of course – makes me marvel at how the project was “pulled together” to achieve its dramatic success. Due to its siting in space near the Sun-Earth L2 Lagrange point, about 1.5 million kilometers (1 million miles) from Earth, very unlike the situation with the Earth-orbiting Hubble Space Telescope, there would be no second chance to fix or adjust anything on the JWST if things went badly.
Yes, it was billions of dollars over budget and years late, but given all the scientific and technical unknowns, I’m not sure how you even begin to work out a budget and schedule for a project of this type, except as a very rough number with at least +100% uncertainty.
I was especially interested in learning more about the analog/digital conversion subsystem, which takes the outputs of the various image sensors and digitizes them. Due to the number of new parts that had to be designed and validated for the JWST, I assumed that the designers would try to reduce that number by using space-rated versions of an available ADC for this function. It seemed to me to make sense to select one that had a “track record” in production and design-in idiosyncrasies, and so minimize risk and uncertainty.
I was totally wrong. Not only did they use a custom converter that was absolutely optimized for the mission, but its design was also part of a larger IC that performed many other functions. It is not a standalone ADC but is part of a larger IC dubbed SIDECAR ASIC (System for Image Digitization, Enhancement, Control, and Retrieval Application Specific Integrated Circuit), seen in Figure 2.
It is located next to the detector like a sidecar on a motorcycle, to minimize the distance the analog signal travels and thus reduce the system noise. The three instruments that use the SIDECAR are the Near Infrared Camera (NIRCam), Near Infrared Spectrograph (NIRSpec), and the Fine Guidance Sensors (FGS).
The custom ADC — and more
There are many people and organizations involved in the SIDECAR ASIC. The ADC subsystem was designed by Dr. Lanny Lewyn, founder of Lewyn Consulting Incorporated (LCI), who was asked to create a 36-channel 19-bit ADC array that is embedded in the imager ASIC. The array of 36 A/D converters was required to fit within an allocated IC width limit. The power limit was 1.5 mW per converter at 100 kilobits/sec, with 16-bit ±2.5 LSB integral nonlinearity (INL) accuracy. It also needed to meet a special imaging requirement of ±0.3 LSB differential nonlinearity (DNL). A 10× speed mode was also required, but at a lower resolution.
Dr. Lewyn used the dimensionless design methodology called Gamma rules that he developed for another ADC project. These dimensionless layout design rules can be ported to multiple technology scales and foundries and represent a significant improvement over an earlier dimensionless circuit design and layout approach (called Lambda rules) created by his advisor at Caltech, Carver Mead. (Professor Mead’s classic 1978 textbook “Introduction to VLSI Systems,” co-authored with Lynn Conway, is widely credited for starting the VLSI revolution). The Tanner EDA design suite from Siemens was the primary design tool, with additional design and fabrication done at Rockwell International Corporation (now part of Teledyne Scientific Imaging).
The design uses a successive-approximation register (SAR) architecture with a precision capacitor array for the most-significant-bits (MSBs) and a precision resistive divider for the least-significant-bits (LSBs), detailed in Figure 3. To comply with the DNL requirements, an algorithmic method was used for perfectly matching the voltage-switching boundaries of the MSB capacitors to those of the lower LSB resistors. Further, the DNL requirement specifies a random error of only one part in 218,453.
Using conventional autocalibration techniques to correct for small A/D linearity errors was not an option because of the requirement for low-power operation and the non-random nature of the signals. The INL requirements were met by using a combination of centroiding algorithms that had been previously used by Dr. Lewyn. One of those algorithms, a perfect centroiding algorithm, involved duplication and rotation of elements rather than the side-by-side transistor method.
The SIDECAR ASIC was evaluated for basic performance at various ground-based optical telescopes in Chile, Hawaii, and other sites, but that’s only the first step. It was also checked out in the cryogenic vacuum chamber at NASA to assess performance in the harsh environment of space.
Beyond JWST
As the JWST is not a classified project, there are many good references and resources available, ranging from broad overviews to detailed technical discussions. The systems engineers and designers working on this project obviously took a big chance on the SIDECAR approach rather than a standard ADC, but also decided it was the only way to meet the tight performance objectives, which were bounded by stringent constraints. That’s a difficult decision, but one that had to be made, tested, evaluated, and succeeded beyond its goals.
Meanwhile, I’m hoping for a well-written book about the project, its many problems and delays, reinvigoration, tradeoffs, and eventual, incredible success. I’d be thrilled to see a quality work similar to these books:
- “Roving Mars: spirit, opportunity, and the exploration of the red planet” by Steven W. Squyres, the principal investigator on the Mars missions that landed the rovers Spirit and Opportunity in 2004
- “Voyager: seeking newer worlds in the third great age of discovery” by Stephen J. Pyne, a fascinating look at dual missions launched in 1977, which is still sending back data from beyond the edge of our solar system, and beyond
- “The Right Kind of Crazy: A True Story of Teamwork, Leadership, and High-Stakes Innovation” by Adam Steltzner (with William Patrick), who led the Entry, Descent, and Landing team in landing the Curiosity rover on the surface of Mars.
Having that book to capture the project would give the thousands of people who worked on the JWST project the wider recognition they deserve.
References
The Mid-Infrared Instrument for the James Webb Space Telescope, II: Design and Build, Publications of the Astronomical Society of the Pacific
Analog-to-digital conversion is key for deep space exploration with the James Webb Space Telescope, Siemens [design methodology and tools]
Cryogenic Data Acquisition ASIC, NASA
The SIDECAR ASIC — Focal Plane Electronics on a Single Chip, SPIE
Amazing Miniaturized ‘SIDECAR’ Drives Webb Telescope’s Signal, European Space Agency
Related EE World content
Can I park in one of your Lagrange points?: Part 1
Can I park in one of your Lagrange points?: Part 2
Can I park in one of your Lagrange points?: Part 3
The Hubble Space Telescope: concept, delay, embarrassment, despair, and finally – jubilation, Part 1
The Hubble Space Telescope: concept, delay, embarrassment, despair, and finally – jubilation, Part 2
The Hubble Space Telescope: Part 3
The Hubble Space Telescope: Part 4
