When we advance high-speed data conversion and synchronicity, engineers and scientists can create the world’s largest radio telescope array, giving the world access to more deep-space data than ever before.
In 2018, Dr. Krzysztof Caputa ran into a serious challenge: The astronomy instrumentation project he was leading for the National Research Council of Canada (NRC) was struggling to reliably sync the timing between key electronic components separated by hundreds of miles.
Without syncing the distant signal timing to a quarter-billionth of a second, astronomers would lose the ability to see further into the universe, and with more detail, than ever before. Complex engineering projects run into delays all the time. But the stakes on this pioneering project were high, and solving the challenge would give Caputa’s team a chance to break new ground.
"The timing problem was a significant hurdle for the most ambitious radio-telescope project ever undertaken," says Philip Pratt, a business development manager at our company.
From many dishes, one
The Square Kilometer Array (SKA) radio-telescope project is a collaboration between scientists and engineers in more than a dozen countries around the world, designed to combine the signals from thousands of relatively small radio antennas into one or more large signals.
Together, the many SKA antennas will provide a signal-detection ability equivalent to that of a single massive radio dish with a one-square kilometer collecting area.
When the project comes online in 2027, astronomers will have access to more deep-space data than has ever been available in the history of radio astronomy, thanks to breakthrough advancements in high-speed data transmission and the world’s largest radio telescope array. The combined antennas will have the game-changing sensitivity needed to pick up new details about a range of faint, distant astronomical objects and phenomena, including the first black holes and stars born after the Big Bang, clues into how galaxies form, the nature of dark matter and dark energy, and the molecular building blocks of life.
But back on Earth, a quarter billionth of a second stood in the way.
The quest for speed
In the first phase of the SKA project, which began in 2019 and will continue through 2027, some 130 radio dishes will be located in the Karoo desert of South Africa. The signals from each of those dishes will be whisked by fiber-optic cable to a global central processing bunker thousands of miles away. There, the signals will be electronically combined.
But before the signal from an antenna can be sent on its way to the bunker, it must first be converted from an analog signal into digital data. That initial signal processing takes place inside the antenna, and designing the electronics to handle it is a critical piece of the project. In 2014 that responsibility fell to the NRC, under Caputa and the NRC engineering team in Victoria, British Columbia.
By 2016, Caputa and his team – with the help of our company’s high-speed data converter team – had come up with a solution, based initially on what was then one of our new, state-of-the-art analog-to-digital converter (ADC) chips, the ADC12J4000.
“We have a long history of designing products that help astronomers look into deep space," Pratt said. “The project needed a certain amount of speed and performance and the new chip exceeded the initial requirements. We wanted to do everything we could to help the NRC team succeed."
Timing is everything
But in 2017, an SKA design review turned up a new challenge.
SKA antennas will hop between different frequencies to catch as many deep-space signals as possible. Each hop introduces a brief, tiny wobble in the rate of data flowing through the ADCs. “Brief" is really an understatement – the wobble was typically in the vicinity of a quarter of a billionth of a second. But because the SKA is combining high-frequency signals from many antennas, that infinitesimal mis-timing is enough to throw off the synchronization with the central processing bunker.
“We have to stay in perfect sync for hours on end," says Caputa. “If we miss a single clock cycle in that time, we lose coherence."
The problem seemed to require two separate ADC systems in each antenna, so that while one system is passing the signal from the frequency of immediate interest through to the central bunker, the other one can be tuned to the next frequency of interest. Hopping to the next frequency would then be a matter of switching to the other converter, which could be accomplished without losing timing. But a two-ADC solution would also require doubling all the connections and digital-signal-processing components, which would have necessitated a complete redesign of the system, potentially setting the team back years.
“We were really trying to avoid having to double everything," says Caputa. "We were trying many different ideas, but it didn’t look as if there was any way around it. That’s when I reached out to TI, and we worked together on what turned out to be the solution."
Blazing speeds and unprecedented accuracy
Pratt and engineers at our company sketched out to Caputa what might be the key to solving the problem: A new, even faster chip that could hit a nearly impossible-sounding 10 giga-samples per second.
"We had given them more speed than they first asked for, but it turned out they needed even more," Pratt said. A 10 giga-sample chip, Pratt explained, could be run in a dual-channel mode, with each half-speed channel tuned to a different frequency signal – and still have each channel operating at the SKA’s demanding data rates.
“If this worked, it meant we wouldn’t need additional communications lines," Caputa said. “We could just process two signals through the same equipment."
Hoping that the chip would be able to produce the needed blazing speeds, Caputa in 2018 proposed that solution to the SKA leadership and got a thumbs up. The TI team pulled out all the stops to deliver, Pratt said.
In addition to reworking the chip throughout in order to handle the extra heat put out by the power requirements that go along with doubling the speed, the team also came out with new tricks to pin down clock timing on the chip to ensure it could nail each fractional billionth-of-a-second cycle with unprecedented accuracy.
Four months after the initial discussions – including countless phone calls and emails between Caputa and the TI group to address Caputa’s needs – Pratt brought Caputa a prototype of the new chip, the ADC12DJ5200RF. The chip passed all of Caputa’s tests.
“He proved it could run at the target rate of 5.2 giga-samples per second through each channel without ever losing a timing cycle," says Pratt.
Back on track
In 2019, a second detailed design review of the SKA system validated the new components.
“We passed with flying colors," Caputa said. “Instead of needing a complete redesign, we just swapped chips."
The new chip not only keeps the timing in perfect sync, but it has also improved the system’s frequency response, which means clearer signals for the SKA. His team is now ready to start building the first devices that will go into the antennas.
Pratt said he and the TI design team take a lot of pride in what they were able to do for Caputa and the SKA.
With the project back on track, astronomers are getting excited. "The SKA is amazing," Caputa said. "It’s really satisfying that we were able to work with TI to find a solution."
Living our passion to create a better world
Advancing high-speed data conversion and synchronicity is one example of how our company is living its passion to create a better world by making electronics more affordable through semiconductors. This passion is alive today as we continue to pioneer advances in integrated circuits. Each generation of innovation builds upon the last to make technology smaller, more efficient, more reliable and more affordable – opening new markets and making it possible for semiconductors to go into electronics everywhere. We think of this as Engineering Progress. It’s what we do and have been doing for decades.
Images provided by the SKA Organisation and adapted according to the Creative Commons Attribution 3.0 Unported License.