For his Hackaday Prize entry, [Ted Yapo] is building a Robot Radar Module breakout board. His design uses the A111 60 GHz pulsed coherent radar (PCR) sensor from Acconeer AB (New Part alert!) .
The A111 is a low power, high precision sensor ideal for use in object detection or gesture sensing applications. The BGA package is tiny – 5.5 mm x 5.2 mm, but it does not appear very difficult for a hacker to assemble. The sensor includes an integrated baseband, RF front-end and Antenna in Package so you don’t have to mess with RF layout headaches. Acconeer claims the sensor performance is not affected with interference from noise, dust, color and direct or indirect light. Sensing range is about 2 m with a +/- 2 mm accuracy. And at just under $10 a pop for 10 units or more, it would make a nice addition to augment the sensor package on a Robot.
To get started, [Ted] is keeping his design simple and small – the break out board measures just 32 mm x 32 mm. The radar sensor itself doesn’t require any parts other than a crystal and its loading capacitors. A LDO takes care of the 1.8 V required by the A111. Three 74LVC2T45 chips translate the SPI digital interface from 1.8 V to external logic levels between 1.8 V to 5 V. The three level translation chips could possible be replaced by a single six or eight channel translator – such as one from the TXB series from TI. For his first PCB iteration, [Ted] is expecting to run in to some layout or performance issues, so if you have any feedback to give him on his design, check out his hardware repository on Github.
Acconeer provides a Getting Started guide for their Evaluation Kits, which includes a detailed Raspberry-Pi / Raspbian installation and an accompanying video (embedded after the break) targeted at hackers. We are eagerly looking forward to the progress that [Ted] makes with this sensor breakout. Combined with LiDAR ToF sensor breakout boards, such as the MappyDot, it would be a great addition to your robot’s sensing capabilities.
Was just this afternoon looking at TI’s AWR1642 and friends. The A111 looks like it might be a viable alternative – cheaper, but probably less accurate.
Looking at this chip’s datasheet, it looks like it only has a single antenna – so it can do ranging and range rate, but can’t provide any angle info. The AWRs can, and at greater range. (I’ve played with the base AWR, I just haven’t had time to tune it for my applications. In general we’ve been so far paying the extra money to go to scanning LIDAR for applications at work.)
Wonder how many of these “new parts” exist because of AVs?
PS. Odd choice of packaging given the number of unused pins? From a skim of the datasheet, it wasn’t clear whether the FFT pipeline is on the client side or the host side, given the price point, presumably the latter? Why oh why are manufacturers so coy about accuracy and resolution? A vague graph with 1SD (68%!) at short distance doesn’t cut it – sharp, TI & the rest are just as bad.
Well, I can imagine a QFN or similar would work as well… but would end up being larger even at the same pitch as it coculdn’t have two rows of pins all the way round. They probably don’t have pins in the middle as it would interfere with the antenna etc…
It’s a flip-chip CSP, i think that means the chip is facing up, than bonded on the edge to the package, so they use only two rows on the edge, mainly for grounding
packaging choice might be due to some kind of RF witchcraft. Hey, how can I complain–radar right on a computer chip. It’s amazing!
This is great but I am really excited for a tiny sensor that can provide points for a 3D map, like a 3D camera, Kinect, etc.
Maybe a MKII of this chip could have provisions to build a phased array out of them? :D
I am currently in love with a similar part, the vl53l0x, which is a 5-10€ IR laser rangefinder with 1-2.5meters range and -/+1mm precision.
Vl5310x is nice. But its IR and limited by lightning conditions and field of view. This is a radar module and can do lot of more fancy stuff like measure heart rate of someone on the other side of a wall or simular.
Even then 2m feels a bit short range for a radar module. Would be much more usefull for navigation etc if it was 5m or more. Looking at the data sheet it looks like you could get some usable data för distance > 2m as well.
“2m ranging is guaranteed for an object size, shape and dielectric properties corresponding to a spherical corner reflector of 5 cm radius.”
– 60GHz
– “see” through walls
You can only have one, not both :P
Were you able to get the A111 working? I used a pi 3B+ and following all the instructions, the output was just a list of errors
Same for me but the 1.3.0 software solved those problems.
Yeah, that was a nice waste of about 6 hours for me :/
This sounds like the Google radar gesture sensing project. I never heard anything else about that. Is this the result of it?
No, you’re referring to project Soli for the Google radar gesture project (which I don’t believe is available). This is a separate Swedish company
yay. binary blobs.
I lol’ed
+1
The part is cheap at $12, the development board is reasonable at $48 but the connector board is just silly at $198.
Yes, exactly. The idea is to integrate the development board with the interface electronics from the overpriced connector board into a single module that will cost much less.
Can we just blue wire the development board to the Raspi? looks like it just need power, ground, and SPI signals?
Is there an SPI register list anywhere? they show they can detect materials and distance but inly provide an API to measure distance in their documentation.. If the SPI nterface is nothing too fancy speed-wise it should be doable to use an stm32 or a PIC to interface the chip!
Could this be used to track positions of individual bees in a swarm? The summary says a few mm precision over a couple of meters, but it’s not clear what is the smallest object it could detect nor the number of objects simultaneously.
Could this be used to track the positions of individual bees in a swarm? What are the limits on the smallest size and numbers of objects which can be tracked simultaneously?