RISCV: Bouffalo Lab's BL602 and BL604

Posted on
  • https://www.hackster.io/news/bouffalo-la­b-s-bl602-and-bl604-aim-to-bring-risc-v-­wi-fi-socs-to-the-esp8266-price-point-46­b9d6cdba36

    Has anyone else seen these? Is there any interest?

    It looks like a very interesting part - very much like an ESP32, but RISCV, and extremely cheap...

  • Mentioned also here https://hackaday.com/2020/11/20/new-part­-day-bouffalo-labs-bl602-risc-v-wi-fi-bl­uetooth-soc/

    they have some SDK and docs on github https://github.com/bouffalolab/bl_iot_sd­k


    They do not mention RISCV anywhere in datasheets so initially I was not aware that it is RISCV (e.g. Telink chips used a lot in BLE devices are also described as risc 32 bit chips) so was not that much interested. Also I am into Espruino and nrf52 mostly because of BLE and was guessing getting that running well would be huge task and the quality of their SDK could be lacking when compared to Nordic. However I agree it looks interesting.

  • BTW when talking about RISCV there is also https://www.espressif.com/en/news/ESP32_­C3 but not sure how much they improved power management, ESP32 is not exactly low power as expected for BLE devices.

  • Yes... I think some testing would be required, my guess is Nordic probably still beat Espressif and Bouffalo Labs on power consumption, but you never know...

    ESP32_C3 looks promising too - as I understood it ESP32 had a tiny reasonably low power processor in it, so Espressif were able to reduce the power usage of the silicon down to a sensible amount - maybe the new processor will improve matters. I imagine the API and docs will likely be a lot more stable too.

    But as you say, getting Bluetooth and WiFi functionality to a usable level may just be a step too far. There's also the same problem I had with ESP8266/ESP32 - how to make any kind of business out of it when you can buy the hardware so cheap from other places :(

  • IIRC Pine64 was giving away these 5$ devboards to almost anybody who asked for it and had enough (don't know what's enough) open source contributions at the time of announcement.
    I think (this might sound bit negative, but) doing low-level stuff on Risc-V appeals to the hardware nerds. And to people who want a "blob free" wifi chip. To use it commercially I think you would need to certify both BLE and Wifi, so a company would need to dump lots of resources into software and certifications. But still certifying a fully open source driver could be hard, since how would you limit emission power, if everything is open and documented?

    As Gordon said, this chip's main selling point is it's price. A product with it probably could work, if the chip and software is good enough to do everything, BLE, Wifi, https with some certificate validation, usable ADC and regular peripherals, low enough power consumption. And a good enough software stack to do all these and run the main application at the same time?

    Oh, and let's not forget W600 (Manufacturer's site is at https://w600.chip.haus): A cortex-M3 wifi chip, ~1€ +shipping at Aliex in an ESP-01 like formfactor. Made some noise about a year ago, but I think it didn't really take off. The W600 Pico does have micropython, but it does't have enough flash to do https, so what's the point?

  • I got two Boufalo units! They are on my table now!
    Espruino on it should be too much interesting! I can test for you 👏

  • @Gordon
    Complex to port ?

  • I got sent one of these boards (BL602) as well over the holidays through a twitter giveaway they made. Haven't really played around with it that much yet, but this is a good reminder :)

  • Very interesting about the W600 - that's amazingly cheap, and looks pretty competent. I reckon you could get HTTPS on that, just maybe with a limited packet size (which wouldn't cause a problem for most use cases).

    I stupidly missed out on the BL602 from Pine64 - they're giving them away free with every order at the moment if you ask and I ordered a Pinecil! If there's no certified WiFi/BLE stack that definitely makes life tricky though.

    Actually porting to RISC-V is pretty easy I think. It has already been done by one person, and I believe involved changing just a single like of Espruino's code.

    The thing that'll take the time is supporting all the peripherals and WiFi/BLE though.

  • I am trying to resist such freebies as my pile of interesting junk is quite big already :-)

    BTW now also Telink joined risc-v with their newer TLSR9xxx https://www.telink-semi.com/news/telink-­risc-v-tlsr9/ which is great too. Older 8266,8269,8251 chips are now quite hackable - there is SDK, gcc (for their own TC32 architecture) and also the SWS flashing protocol was figured out. Was checking their BLE developer handbook and the limits can be seen, current SDK can run either as central or peripheral not both, only single connection (for multiple there is separate beta download), there is some info about timing limits or you start losing packets (bt stack runs cooperatively as part of your code) so I see it can be much worse than with Nordic softdevice. And that's Telink with their chips already running in tons of cheap chinese stuff for years.

    Was checking if the thermometer could in theory run Espruino - it has 32KB of SRAM (unlike 16KB TLSR8266) and flash is 512KB but the FW can be only 128KB now. Maybe it is just SW limitation in SDK but maybe it is in HW - it uses 2KB from SRAM as SPI flash XIP code cache and can run dual banked firmware from two locations - remaps addresses from configurable flash offset to same base - so not sure about maximum size this remapping can do.

  • Nice @Gordon!

    I cant wait to test the Espruino my 2 boufalos modules!

  • Post a reply
    • Bold
    • Italics
    • Link
    • Image
    • List
    • Quote
    • code
    • Preview

RISCV: Bouffalo Lab's BL602 and BL604

Posted by Avatar for Gordon @Gordon