In case you missed it: STM32WLEx - Wireless MCUs with LoRa support - STMicroelectronics
It’s a true SoC, as in everything is on one die. This makes it different than the Atmel LoRa parts, as far as I know. ST has also done a few nice things that Semtech hasn’t – notably documenting some features that Semtech was reticent to document (Semtech and ST have slightly different marketing objectives).
Naturally, it gets announced two weeks after I submit a STM32L4+SX127x design for prototype. 
Ooh, very interesting, thanks for sharing. The combination would definitely simplify a couple of my designs. No pricing or availability info yet that I saw. Wonder if they’ll have non-BGA packages at some point, like the …WB line.
STM32WLE5J8 is $2.94 @ 10K units apparently, STM32WLE5JCI6 is $3.45 (they list the prices if you click Open product page and scroll to the bottom.
Which means it will be in the $2 range in volumes, maybe lower if it gets popular. That’s a similar price point as the SX1262 alone – maybe even slightly better.
That’s definitely a cool looking chip and I’m interested in trying it out. I always dislike the waiting game after these early announcements until the point when a dev kit is available.
STM32L4 + SX1262 kits are readily available, and for FW purposes there’s 99% congruence. That’s what I’m using. Actually, I built my own because I’m cheap… using two cheap kits… and mounting tape… wires to come shortly.
This is how the complementary Nucleo board will look like.
Enthusiasts or professionals looking to prepare for the mass market availability of the STM32WL can start working on an STM32L4 Nucleo board and a Semtech expansion board such as the SX1262DVK1DAS, the SX1262DVK1CAS or the SX1262DVK1BAS.
Probably a good idea, especially because this tests just how serious I would be about jumping to this platform. If I’m just waiting for the shiny new thing to come out…well, I’m not really serious about it.
ST Microelectronics have an area at Embedded World which is happening Feb 25-27 Nuremberg for anyone nearby or intending to go. They usually have boards to give away. The catch is you usually have to register for them. Last year registration was at the event.
Appears that samples are available. I put in a request and hope to get a few ASAP.
But now I need to make a new PCB. The proto I made with discrete LoRa + STM32L4 is shipping (assembled) tomorrow. 
More news: there will be at least one vendor marketing a compact LGA module that includes this SoC plus companion circuits.
And something else: the clearances of the BGA are such that through-hole PCB fab with 0.76mm space/trace and 0.15mm drill will be sufficient. These days, that’s not exotic. You don’t need to go to Blind/Buried.
I will be building a PCB module for this part with integrated everything and header out. I’ll keep you informed.
I’d be interested in that module. I have been fantasizing about making my own board in my free time and then I remember I don’t have any of that 
Samples are in. Time to go into the basement for a week and crank out some PCB designs 
I want to (again?) make the point that this SoC is cheaper to buy than the SX1262 itself, so it’s kind of a no-brainer if you’re using LoRa for anything. A lot of folks have been using a Nordic BLE SoC + LoRa transceiver as a BLE+LoRa solution, but now I’d say that the optimal BLE+LoRa solution might be a bog-simple CSR 102x plus STM32WL. Just my two cents.
I hadn’t heard of the CSR 102x before, is it hard to get ahold of? I don’t usually hear about Qualcomm being super easy to work with
The short answer is that if you’re building in volumes where the low price of the CSR102x matters enough to select it over one of the entry-level Nordic SoCs (e.g. 52810), then yes, it’s easy enough to get Qualcomm’s attention. For anything lower than these volumes, having a minimal 52810 dedicated just to the BLE jobs is also a good solution.
I had worked with CSR dual mode chip (supporting classic Bluetooth and BLE), so I know what is like to work with CSR or Qualcomm tech support. Basically there are lot of technical documentations for their chip are not publicly available, if you come to a technical question, you have to discuss to CSR or Qualcomm tech support personal, and then you may receive some useful technical documents to help.
I’m not sure that I understand the design goal for this hardware platform. Is this for LoRa + BLE coexisting? ST has solution provided for BLE and LoRa. For example the BlueNRG-2 is a SoC for BLE. Why pick CSR 102x for BLE while take LoRa solution from ST? CSR 102x include a 16 bit core for BLE stack and incorporate an extra 51 core to help on I/Os, seems much more difficult to manage, while BlueNRG-2 using a Cortex-M0 core to handle all BLE stack as well as user application. My 2 cents.
If I’m buying a BLE device for a product with very basic requirements (e.g. bog-standard GATT behavior that is supported by a binary from the manufacturer), I don’t really care who it comes from or what it’s like to program. I do care about cost and simplicity of integration.
I was doing Google searches on the topic of “LoRa” and somehow found this thread.
As many of you have expressed interest, I’m also interested in designing a LoRa board. Specifically, my end goal is to use the latest sx1262 LoRa chip in the design of an “energy efficient board” for long range communications.
In the datasheet of the Semtech LoRa sx1262 chip, there is a schematic that shows what additional passives will be needed to actually drive an antenna. The list of components are: a PE4259 RF-Switch IC, 6 inductors, 15 capacitors and a couple of resistors.
Do I really want to mess with all these components? Obviously not. So, I need to find a small module that incorporates the sx1262 chip, and all these antenna design passives. Of course, RF modules of all types are quite common…except when it comes to the SX1262 chip.
I want to find a good LoRa RF Module, then choose my own flavor of microcontroller. Then all I have to deal with is a SPI bus and a few General I/0 Lines to talk to the radio module…and then a single pin(pad) off the radio module to attach an antenna…again, an antenna type of my own choosing.
So when I read in this post that STMicro had merged the CPU and Semtech chip into a single SOC…I was intrigued. I really had never dealt with STM chips, so I started digging into the Datasheet and Reference Manual for this SOC.
I found out some possible “gotchas” for those who are thinking of incorporating this SOC into their future board designs.
I quickly became frustrated! Why did STM have two separate documents that rightly belong in a single datasheet? I kept having to go back and forth trying to find information.
Right off the bat, the BGA package is problematic, as I like simple 2-layer boards and easy to solder TQFP CPU Chips.
But the real deal breaker is the electrical current carrying capacity of the SOC. The datasheet says the SOC can carry a maximum of 130 milliamps total for both the CPU and the RF portions of the SOC.
Well, if the RF portion of the SOC is transmitting at a frequency of 915MHz with a power setting of 22 dBm…it will be drawing 120ma…leaving a grand total of 10 milliamps for the CPU.
Houston…we have a problem!
I want long distance, and at the same time, I want to drive some I/O in the form of LED indicator lights. Three or four lit LEDs can put me over the 130mA SOC limit. And no, I don’t want to scramble to incorporate transistors as LED drivers. There’s nothing in the datasheets, that I found, that makes any attempt to raise warning flags in this regard while you’re doing your design.
Then, my last assumption was that this SOC contained all the aforementioned antenna circuitry. But I quickly realized that this would be impossible in a 5x5mm footprint. So, if you use this SOC, you will still have to make room for that long list of antenna components.
And I didn’t see any schematics in either document that shows a typical antenna design in regards to the RF Switch IC, and all the inductors, capacitors, and resistors. I also did a search of the word “antenna” in both documents, and came up empty.
I know this is a new part, but the two datasheets are sorely lacking. At the least, there needs to be a schematic of a typical antenna design included somewhere.
I consider this new SOC as being a rather niche part. If miniaturization is a goal, for something like a wearable, this SOC may fill the bill.
Otherwise, I think I’ll stick to a design methodology that lets the designer pick a CPU to his/her exact requirements. Then marry that CPU to a sx1262 module that combines the sz1262 chip, and all the passive antenna components under one roof. Finally, from there, the antenna pin(pad) off the module is where you add an antenna of the specific type you want.
Or maybe I’m full of prunes…and have missed something.
This SoC is geared towards people like me, who are building the kind of modules you want to buy, where all HW is integrated as well as the networking FW. Building an RF board on a 2-Layer PCB with leaded packages is generally inadvisable for frequencies much higher than 400 MHz.
The current situation isn’t as bad as you think, because the limit is 14 dBm when using the IC’s power supply to drive RFTX. That’s the same as the case with the SX1262. Edit: even in this 14 dBm case, the datasheet doesn’t group the SMPS supply domain with the MCU supply domain, which is what has 130mA limit. The RFPA domain is open drain with an entirely separate ground path.
The module I’m laying-out right now will have a mostly compatible pinout with the Murata Type ABZ modules. If there’s demand, I may also build one with limited I/O but much wider pitch.
