Since the launch of the Raspberry Pi Pico back in January the little board with its newly-designed RP2040 microcontroller has really caught the imagination of makers everywhere, and we have seen an extremely impressive array of projects using it. So far the RP2040 has only been available on a ready-made PCB module, but we have news today direct from Eben Upton himself that with around 600k units already shipped, single-unit sales of the chip are commencing via the network of Raspberry Pi Approved Resellers.
This news will doubtless result in a fresh explosion of clever projects using the chip, but perhaps more intriguingly it will inevitably result in its appearance at the heart of a new crop of niche products that go beyond simple clones of the Pico in different form factors. The special ingredient of those two PIO programmable state machines to take the load of repetitive tasks away from the cores raises it above being merely yet another microcontroller chip, and we look forward to that feature being at their heart.
The Broadcom systems-on-chip that power Raspberry Pi’s existing range of Linux-capable boards have famously remained unavailable on their own, meaning that this move to being a chip vendor breaks further new ground for the Cambridge-based company. It’s best not to think of it in terms of their entering into competition with the giants of the microcontroller market though, because a relative minnow such as the RP2040 will be of little immediate concern to the likes of Microchip, ST, or TI. A better comparison when evaluating the RP2040’s chances in the market is probably Parallax with their Propeller chip, in that here is a company with a very solid existing presence in the education and maker markets seeking to capitalise on that experience by providing a microcontroller with that niche in mind. We look forward to seeing where this will take them, and we’d hope to eventually see a family of RP2040-like chips with different package and on-board peripheral options.
Shouldn’t that link be “two pio programmable state machines”? Or did they made a new version?
There are two PIO blocks with four state machines each.
And article says:
> The special ingredient of those three PIO programmable state machines to take the load of repetitive tasks […]
Yup. There are not. There are two PIO blocks, four state machines each, for a total of 8 PIO state machine on RP2040.
oops!
I found the source: https://www.raspberrypi.org/blog/raspberry-pi-rp2040-on-sale/
As with all Raspberry Pi products, it seems like getting them in sufficient quantity to build into a product is somewhat difficult, with vendors already sold out, or limiting quantities to 10 per person.
Also, unless there is a source in China, it’s gonna be annoying to get parts to the factory to include on boards.
Until these things are solved, it seems like a hobbiest only part.
This is single unit availability for prototyping and testing. Reel quantities will be available in the autumn, along with reel quantity pricing.
As expected, the Japanese “authorized” resellers are pulling scams compared to what your press releases all say. They’re only selling in 10 packs, no single units available, and they charge 15% more than you claim. Can’t you hold them to a pricing contract or something?
$1/£0.85 in single units is a pretty incredible price, almost a third of what Mouser is quoting for an ATSAMD21E18. It’s a bit over-powered for most of my projects but at that price I might as well use it anyway.
If this is a direction for the future, my hope is that they will make a chip comparable with the BMC2835, so it will be possible to buy a Pi Zero (alike’ish) in quantities, for around 5$, from RS and Farnell.
Hopefully they are just clearing out the stores of the silicon revision with the faulty ADC & they have a new rev with this error fixed.
Details?
does anyone realy use the RP2040 for project because its the best chip for the problem? or is everyone just using it for the raspberry hype.
I bought one the first day and messed arround with it some time – at this point i didnt want to use low level code because i am lazy and not that good at programming.
first thing i tried was circuit python which was far form a compatible port – worse i2c and pule io (one wire sensors) were not yet ported and so for my project it was not use able. after some month features kept growing but sleepmodes and other hardware features are not yet supported.
then the inofficial and official arduino ports were released – both very poor documented and barely useable.
i learned my lesson: dont jump on hype trains! there are similar micros with better support and same or lower price.
i should have used an stm oder esp for my project.
While I’ve not yet used a RP2040 it can do things most other micro’s can’t with the PIO bits, which is very exciting to those who understand this, and makes it look like exactly the right tool for many tasks – the other options when you need PIO like functionality are not trivial to operate either, FPGA, propeller etc are a whole new world from the basic python/Arduino worlds you seem to somewhat know…
The documentation might not be as polished and rookie friendly as other things out there yet, but its the Pi foundation, so it will be, eventually, and it looks to me like the important details are out there.
As with all things electronic the best chip selection its down to your desired use, cost, and what you understand already – the cheapest end of an FPGA family might do the job better than a even a very beefy expensive CPU (which is pretty much true for any task you want working in ‘realtime’), but if you already know how to program for the CPU or want to leverage existing code, then it can be a valid choice – just a less efficient one. (though there seem to be alot of FPGA blocks you can add into your design as well – so maybe there is something you can use – I’ve still barely dipped my toes in the FPGA world)
I did just have a project where the rp2040 proved to be the only real solution, specifically because of the pio. my parents bought an rv a year ago, and have been looking to upgrade it to support home automation through Home Assistant, but the airconditioners, made by Dometic, used a super propritary protocol to talk to the thermostat. This protocol not only had some funky voltages going on (solved through a level shifter), but also some absolutly absurd start and stop timings, mesured down to the microsecond. One fellow RVer had managed an arduino project to convert that protocol to your standard 5 wire thermostat, which used an arduino to bitbang the protocol, but for our case, that wouldn’t work, as we had two units, and we needed to send a serial signal back to the raspberry pi that handled the home automation. And with a bitbanged protocol that was timing sensitive, we couldn’t get the dual serial lines going. So instead, I converted the timings into something the PIO could handle, added DHT22 support on the second (surprisingly easy), and used the two hardware serial ports to talk between the two ac units and the pi. Took less time than trying to force an arduino to do it, and at $5 ultimatly cost less. It’s kinda funny the chip just came out, as we just ordered pcb’s for the final install next week. Would have much rather used the chips directly instead of the pico… but oh well.
The QFN format is hard for hand soldering and would be great if they provide another package format but I guess it’s not going to happend.
So the best option to make your own 2040 dev board is they provide stock to chinese pcbs factories.
QFN isn’t all that bad, at least compared to QFP (which is what I assume you mean by “another package format”). It doesn’t have the same issues with solder constantly bridging between pins – just put a ball of solder on your iron, run it over the side of the chip, and you’re done.
Exactly. If its mostly going to be a maker/hobbyist microcontroller, why not release a more friendly package like a QFP, or even a PLCC or DIP version.
As a hobbyist, there are various form factors (adafruit, RPI, Sparkfun) that will probably meet anything I would want to do, but to work with the chip directly PLCC would be an easier ‘hobbyist’ format for hand soldering the socket. A bit bulkier but easier to use. I know we used them on are company boards for 68332s and never had a problem with them. Thing is the market probably isn’t there…. My guess in large outfits the current format will work well for ‘minimum’ foot print.
I’m surprised that no one mentions the lack of internal Flash and that it _requires_ an external Spi Flash chip to run.
That’s good to know. Thanks.
External spi flash is actually good. First – if you debug with it – if you wear it out – you can swap flash, not the microcontroller itself, which will be cheaper (and easier).
Also – SRAM/Cache is definitely faster than internal flash, unless there are some measures taken (like ART Accelerator on STM32)
OK, I’ll admit it. I’m an RP2040 fan. For me, the PIO is the reason. It’s a useful waystation between bit banging and an FPGA, particularly when you need to deal with oddball (and not so oddball) protocols. Of course it can be improved, but it’s useful now and points the way for derivative devices in the future.
try
https://github.com/Wiz-IO
I don’t understand the attraction to this. Is it just the PIO feature? What is an example where PIO would help over a microcontroller? Some MCUs, PSoC in particular can do many FPGA like things.
The use cases explained in the comments sound like they could be achieve by other regular microcontroller-level means.
Its cheap, powerful, pretty easy to use, and versatile. Not really just the PIO, even without using that feature its cheap and capable, though of course that is a large part of what makes it so versatile.
There are many other options that can do similar PIO like things, and more powerful CPU elements that can do more that way, so a RP2040 may not be the right option for you but its a capable and very affordable option (at least on paper, I’ve not played with one yet).
looks like raspberry is using atmel IPs. I hope it s an actual atmel die.