Like many of you in the tech industry, I am lucky to be able to work from home while the Covid-19 pandemic takes its course. To be effective and efficient, I adopted a disciplined schedule to manage internal meetings and discussions with customers. The adjacent benefit of maintaining needed social links was welcome at a time of deep isolation. More importantly, I was able to reinstate long-lost time for open-ended thinking and learning, accompanied by old fashioned pen, paper and sneers from my college-aged children.
Over the years, I maintained a list of “topics to clarify in the future”, which never seemed to shrink. Today, I can proudly state that I checked-off items for the first time in memory! In practice, it is turning out that answered questions have spawned newer ones, but I can at least claim forward momentum and progress. I will be occasionally sharing some of these topics, among others, with the hope of stimulating discussions with the interested community. For me, hearing different technical perspectives has always made the difference in building intuition around a topic.
I was hired into ADI many years ago by Barrie Gilbert (pictured to the left). Yes, the Barrie Gilbert who baptized the famous mixer. Sadly, he passed away unexpectedly earlier this year. Barrie was best known for his mastery of IC design; bordering on a fanatical attention to detail. He was an inspiration to many inside ADI, as well as outside, with his innovative topologies above and beyond his famous mixer. These include Doublets, Triplets and Squaring cells based on Translinear principles, a host of variable-gain amplifier structures known as X-Amp and Z-amp, and fascinating bandgap bias cells that challenged circuit simulators with their convergence and the reference community with their low noise and precision. I could go on and on with the other topologies that only a select few of us were privileged to see.
Another side of Barrie that went unnoticed, simply because his circuits were so spectacular, was his ability to anticipate customer needs and craft a solution based on that starting point. Before he ever put down a single transistor on a schematic, he would define multiple ways the part could be used, what the scaling and function of each pin should be, what the pin-out and package should be, and so on. He called this “Imagineering” and insisted that we all “start at the end” by writing the concept data sheet with all trimmings; including the specifications page with target specs and the applications section, brainstorming the use cases. The process of standing in the customers’ shoes would work out fundamental application impact issues early on and reveal critical aspects for the transistor-level design phase that was to come. This top-down approach put holistic product concept ahead of detailed circuit design. A philosophy that is even more critical now than in years past as solutions have become more complex.
ADI’s front-end solution for active electronically steered phased arrays (AESA) exemplifies this anticipatory approach to problem solving. For those new to phased arrays, they enable electronic steering of spatially directed beams. These could be radar pulses or communication signals, like 5G and satellite links. The trend is to replace single-function, bulky, mechanically steered antennas with sleek, multi-function electronically steered phased arrays. The ADAR1000 4-channel X-Ku band analog beamformer and the ADTR1107 X-Ku band transmit-receive (T/R) module demonstrate the application level principles embodied in Barrie’s philosophy.
Based on multiple conversations with customers and partners combined with an understanding of the application challenges, we anchored our design goals to the following starting principles:
In order to fit the 15mm pitch for 10 GHz, the ADAR1000 integrates 4 channels of independent gain and phase control, all programmable via SPI. It offers multiple options for loading states, including lightning access from on-chip memory for fast beam hopping. It also packs in all the control, biasing and monitoring of the matching set of four T/R modules for pulsed radar operation. The ADTR1107 incorporates LNA, PA and switch, fitting neatly and efficiently around the periphery of the ADAR1000, which was designed to avoid any crossovers or awkward routing. The overall footprint is reduced to 5 QFNs, a few decoupling capacitors and power rails.
AESAs can suffer from multiple sources of gain and phase errors introduced between channels that need to be calibrated globally to maintain beam shape integrity. By centering the ADAR1000 symmetrically between elements, the routing between the chips themselves and the antenna is straightforward and direct. The 4 channels of the ADAR1000 are identical and designed to minimize variations from part to part. The absolute accuracy and linearity of the gain and phase functions was emphasized during the design, with guaranteed monotonicity. Additionally, interactions between the gain and phase control (change in gain causing a phase change and vice-versa) were minimized to an unprecedented level. Finally, temperature variation was consistent and repeatable for all channels and states. The expectation is that each element can be calibrated one-time at a fixed gain-phase state and temperature to achieve array level performance, as opposed to relying on extensive tables.
Power dissipation was minimized by using the optimal technology for each function. The silicon-based ADAR1000 provided the RF signal processing and control functionality. The ADTR1107 combines GaAs amplifier and SOI switch technologies for optimal performance and efficiency. Future T/R modules will include GaN technology for higher power handling. The ADAR1000 is able to directly pulse the ADTR1107 PA and switch on and off or simply throttle down the power, all through SPI control.
The above example demonstrates that starting with the customer’s “end" in mind drives critical decisions that turn a good solution into a great solution. To be effective suppliers, it is more important than ever that we work in partnership with our customers to truly understand their bottlenecks and challenges. When there is such clarity of purpose, projects run more smoothly, and time-to-market is accelerated.
Until next time.
I have to admit that it's sort of unbelievable that you were hired by the famous creator of the well-known Gilbert Cell! And I do love the word he coined—Imagineering.