Dev Kit Weekly: QuickLogic QuickFeather Development Kit
November 11, 2022
It wasn’t too long ago that FPGAs were really only used for pre-silicon prototyping. Blink and they’re in embedded systems ranging from industrial control platforms to network RF infrastructure. Blink again and they’re at the bleeding edge of edge AI.
A perfect example is the QuickLogic QuickFeather Development Kit, an open-source platform based on the 0.9” by 2” Adafruit Feather form factor and designed around QuickLogic’s EOS S3 Arm Cortex-M4F microcontroller with integrated eFPGA technology. The eFPGA provides 2400 logic cells, 64 Kb of RAM, and eight FIFO controllers, and is joined by an on-chip sensor manager as well as a Flexible Fusion Engine (FFE) that contains its own memory resources and delivers microDSP-like functionality for efficiently executing mathematical computations like always-on, real-time sensor fusion algorithms.
Those last two blocks are put to good use on the QuickFeather kit, where a pressure sensor from Infineon and mCube accelerometer interface with the sensor processing subsystem via the board’s I2C or SPI interfaces. Audio captured by an Infineon PDM digital microphone, also onboard, leverage the QuickFeather’s I2S interface to transmit signals to logic blocks dedicated to Pulse Code Modulation (PCM) conversion or low-power sound detect (LPSD) that enables local voice recognition.
All this is a great segue into the kit’s AI capabilities, which take shape in the form of TinyML applications built using TensorFlow Lite or the AI Analytics Toolkit from SensiML, a QuickLogic subsidiary. Developers can take advantage of the SensiML Analytics Toolkit’s AutoML workflow, for example, to generate code based on advanced AI and ML algorithms that’s optimized to run on the QuickFeather kit and be updated over time.
According to the company, the SensiML analytics toolkit enables 5x faster time to market over hand-coded algorithms deployed into industrial, commercial, agriculture, and consumer use cases.
All of that powered by a single Lithium Polymer battery. And there’s even an onboard Li-Po charging circuit.
At this point I bet many of you are saying, “But there’s still the whole FPGA programming thing…” That’s why the company has made open source tools like SymbiFlow and Renode available as part of the QuickFeather software stack. SymbiFlow makes FPGA development more accessible to the engineering community by optimizing the FPGA design flow from Verilog to bitstream generation, while Renode provides a system-level simulation framework that accelerates platform development.
Among all of this, there’s no requirement for a multi-gigabyte software install that clogs up the 16 Mbit SPI NOR flash from GigaDevice. And, since the programmable logic is tied to USB data signals and a Serial Wire Debug connector can be used with a USB-TTL converter (sold separately), the path to efficient hardware programming has been made pretty straightforward.
These combine with other well-known tools in the QuickFeather toolchain such as support for the Zephyr RTOS and FreeRTOS SDKs, Android compatibility, and MicroPython-friendly to give users the potential of a completely free and open-source software stack.
And bringing us back full circle while still keeping it open, the QuickLogic QuickFeather Development Kit can be extended via 20 GPIO header pins defined in the Feather specification.
So, now you’re ready to get a jump on FPGA-based edge AI with the QuickLogic QuickFeather Development Kit. If you’re looking to get your hands on one, you can do so from Crowd Supply, Mouser, or directly from QuickLogic for just $59. Not bad. Of course, you can also enter the raffle below for a chance to win this QuickFeather here for absolutely free — we’ll even pay the shipping.
And once you’ve got one, you can take advantage of the QuickFeather Dev Kit resources page on the QuickLogic site that provides software, videos, resources, and even a forum.