STM32L-based wireless board « LeafLabs Garden

LeafLabs Garden » Community » The Garden

STM32L-based wireless board

(13 posts) (4 voices)

Started 2 years ago by jpnorair
Latest reply from jpnorair

jpnorair
Member

I've just finished designing a wireless board for Maple or Arduino, and you can see the PCB design here.

I'm sending it to fab now. Here are the basics:
- STM32L151C8: 48 pin, 64KB Flash, 10KB SRAM. This footprint is compatible with many devices in the L15x range, up to 384/32.
- SPIRIT1 RF transceiver and antenna for 433 MHz.
- "Half Shield" pin-out plus a 30-pin edge connector if you want access to everything.
- By default it runs the OpenTag/DASH7 stack, which is an ultra-low-power wireless networking stack. It can be configured in many ways. The "basic" configuration is set-up to establish an IP-like "socket based" network with autodetect and stuff like that.

At the moment, this is either a module that attaches to something like a Maple, or it is a hardware kit that you write applications for (typically in C), compile, and upload via SWD. It is capable of being a standalone Maple board, though. I would like to port libmaple to STM32L, and then to run Maple programs as tasks via the OpenTag RTOS. This latter part is a bit esoteric, but I can manage that. I can also manage the porting to STM32L, but if anyone in the community is interested, I will gladly accept any help. It will certainly speed-up things. Send me a message if you want to assist, and if I get the impression you know what you're doing I'll send you some boards :)

Posted 2 years ago #
gbulmer
Moderator

jpnorair - In reality, I am probably too busy to help. However, I might be interested in buying some when the work is far enough along to be useful. Right now I am into STM32W or Atmel 'Raven' and Contiki. For reference, what is the target retail price?

Posted 2 years ago #
bnewbold
Ex-Leaflabs

@jpnorair, from the rest of your site i'd guess the design files would be under an open license, is that so? I'm pretty interested in low power wireless and might have time to try getting libmaple running smoothly.

@gbulmer, looks like the BOM could be around 6 bucks even in small quantities, so after fab and assembly maybe $20 (USD) retail? Don't know if this is intended to be a retail product though.

Posted 2 years ago #
jpnorair
Member

Target price is probably more like $35, considering that there's more on here than there is on a Maple, apart from board-IO. I wouldn't be able to cover my non-material costs at $20.

I haven't thought about making it open source or creative commons, or what. The firmware is all open source. I'm not certain there is a big enough market at the moment for me to get benefit out of open sourcing the hardware. That might seem counter-intuitive, but the truth is that people seem to expect your support whether they paid for it or not.

@gbulmer: Contiki is nice, but it is not a really a low-power (or ultra-low-power) system. The products I do go into applications that have three requirements: ultra-low-power, long range, and low latency.

On low power, if this board is running the default firmware configuration, it should average about 60uA. That's not a tiny amount, but it's still low enough to run it on 2 AA batteries for 3-4 years. It is possible to configure it to about 7.5uA, although accepting some increased latency -- fine for low-duty apps. Wireless range should be in the 1km range, or more if you live outside the USA.

Posted 2 years ago #
jpnorair
Member

Proto boards are in. Final proto design is slightly different than the original image, but you probably won't notice.

Image Link

Posted 2 years ago #
gbulmer
Moderator

jpnorair - That looks lovely :-)

Approximately how much does it weigh? I am wondering, might it make a light payload for a Helium balloon?

What might its range be "outside the USA"? Is its range mostly a 'broadcast power' constraint?

Posted 2 years ago #
jpnorair
Member

@gbulmer: The weight is less than I can really measure with the scales I have. I suspect a few grams. If you want to reduce weight further, you can remove the USB connector. You can connect a coin cell to the LVS input.

This is actually a module/dev-kit we built, alongside a product we are making. That product is smaller, lighter, and it includes a 10 mAh solar+LiPoly charging system. Incidentally, 10 mAh is enough to drive that product for weeks.

At the moment I'm tuning the antenna design for the module shown above. I'm expecting to achieve 40-50% efficiency at 433 MHz, which is pretty good. If I were to us ~900 MHz, efficiency might be about 1.5dB better in this form factor, but instantly I would lose 6-9 dB in propagation, so 433 MHz is the "goldlocks zone" for this form factor.

Assuming the low-end, 40% efficiency, I expect that the surface-to-surface max range with this hardware is about 2km, maybe 3km, using the built-in PHY. If you put it on a balloon, then the propagation becomes line of sight at a certain elevation, and the range will be a lot better. Additionally, it is possible to drop the data rate or use a block code (also drops effective data rate) to reduce the decodable SNR threshold, thus increasing range. The max-range setup that is built-in to the firmware uses a 1/2 convolutional code on a 28kbps data rate (56kHz symbol rate). If you layer a 16bit code on top of that, data rate would be low, but you would pick up a ton of processing gain. The STM32L @ 16MHz is more than capable of eating-through coding at that rate, as the MAC instructions are great for doing correlation.

Posted 2 years ago #
gbulmer
Moderator

jpnorair - Thanks for the extra information. The requirement is for as light as practical, and as high a data rate as practical. However the flight time is well under a day (likely under 10 hours).

People let Helium balloons rise to about 30km, so sacrificing battery life for distance is okay.

Posted 2 years ago #
jpnorair
Member

This board isn't designed to transmit beyond 8dBm EIRP, which is not really legal in USA, but it is fine almost anywhere else. Some countries allow up to 15 dBm EIRP (25mW) at 433 MHz, which is quite a lot. Even with 8dBm, if you build a nice directional antenna at the ground station (such as a circular Yagi about the size of a traffic cone ... maybe I'll upload plans at some point), it may be possible to reach 30km line-of-sight. Certainly, in outer space it would be no problem. I don't know how much the air will impact the signal, but it might actually work at the built-in 28kbps.

Posted 2 years ago #
gbulmer
Moderator

jpnorair - Thanks again. When do you expect your boards will be available?

Posted 2 years ago #
jpnorair
Member

Available in a month or so. I might run a kickstarter campaign to fund a mid volume run, although I am manufacturing a bunch right-away for distribution to partners. I'll reserve a handful for Maple community, since I am interested in porting Maple to the board. The MCU is the STM32L151C8U6, which is 64K/10K, but there are pin compatible variants of 128/16, 256/16, and 384/32. So, there's room to grow if necessary.

Posted 2 years ago #
slibert
Member

I've been looking into a STM32 L1-based design; have you found libmaple to work on your L1 design? Could you share any thoughts/insight you may have on any issues getting libmaple to work w/the L1 MCU?

Like you it appears, I'm working on a low-power board w/wireless and I think the L1's a great fit - but am trying to determine how much effort it would be to get the L1 up and running in my software development environment.

I'm only planning on using the USART and ADCs, but am hoping these would work pretty much as is.

Thanks for any insight you may have,

- scott

Posted 2 years ago #
jpnorair
Member

Hi Scott: I don't get to this forum too often, sorry for 4-weeks delay.

Some of the STM32L and F peripherals are the same. I have not done an audit, but it is easy to check by comparing registers in the respective Reference Manual documents from ST.

The main trick in getting STM32L to run in deep-low-power is to leverage STOP mode. This can be tricky. It's actually much easier on the STM32F. If the STM32F STOP mode is good enough for your low-power needs, just figure out how to enable that. Most Wiring platforms are not very good at having low-power. If all you do is trade "delay" or "millis" for STOP mode with RTC interrupt, you will see an enormous difference even with F.

If you need to get the last-drop of energy savings, the L can get you there. It is more efficient than MSP430F5 if you make the effort to push it into STOP mode. STOP mode is much more difficult to do right on L than F, though, because ST decided to make an RTC peripheral that is more clock than timer -- that was f-ing stupid, but it can be worked-around with some clever code (I have it on GitHub).

Anyway, I think the STM32L has more in common with STM32F2 than F1, in terms of peripherals. So, if LeafLabs stabilizes the F2 port, that's probably the one to use.

Posted 2 years ago #

RSS feed for this topic

Reply

You must log in to post.