"the 50 per cent premium"
Yes, it's a 50% premium, but on a $4 base, frankly I think it's awesome that you can get a programmable, wirelessly accessible thochky for six bucks.
A year and a half after the debut of the $4 RP2040-powered Raspberry Pi Pico, the company is shipping a wireless-enabled version: the $6 Pico W. Pico W At first glance, things look similar. The 21mm x 51mm form factor is unchanged and the RP2040 microcontroller sits at the heart of things. The CPU? Still a dual core Arm …
Nice but more limited in terms of IO, memory and processing power, and a bit of a pain in the arse once you take it away from the carrier boards.
Presuming the Pico comes with the kind of support and quality control I've got used to for the larger Pi's, this device has suddenly got a whole lot more interesting.
"Nice but more limited in terms of IO, memory and processing power" - Correct, but you've ignored the most important point which is power consumption, you can get what you mention out of a million ICs currently available.
Assuming you're using these types of things for home automation/wearables/any other battery project it needs a different evaluation. The Zero 2 W is great, but has to be used as the gateway due to lack of deep sleep and power consumption. It's perfect as a gw.
As far as I can tell, the normal pico uses about 1mA in sleep, whilst the ESP manages about 0.3mA, which is a world of difference when it comes to battery sensors.
I'll need to wait for the release to see the actual measurements of course.
"Nice but more limited in terms of IO, memory and processing power"
With I/O, you're right, or at least the I/O is different with each having some features not supported by the other. You're not about the others; the ESP32 has at least 320k of RAM, and often more depending on which module is used. 520k is common. The Pico only has 264, which may not sound a lot less, but when you're doing things like using TLS encryption and caching web requests, it can be. The processor runs a different ISA, but it can be clocked at 240 MHz, providing more power than the Pico's. You may not need either of these, but characterizing them as more limited is incorrect.
A 50% premium on down-in-the-noise is still down in the noise. <Yawn>
More power than the Apollo flight computer (haven't verified this), for less than the cost of a deli sandwich. The cabling and wall wart will cost more.
We live in an age of marvels. If we can just manage to survive it...
> Indeed, and in the US, it would be $6 + sales tax
Which here in Oregon, at least, isn’t a thing. It’s nice to pay the price of an item, rather than California-style (price of item)+(state sales tax)+(city sales tax)+(tourist tax or whatever else local politicians see fit to add)
Not much advantage over STM32 and if you want to drive some circuits that have parallel interfaces, you'll lose performance by having to mulitplex the I/O.
Probably its only edge over the competition is that it is available whereas STM32 is not in stock.
Give me RP2xxx with 128+ I/O and at least a SDRAM interface...
SPI is quite buggy, there are some ICs that won't work with the hardware implementation and you have to bit bang with DMA. Not the end of the world, but you get a performance hit. Sometime you end up looking for a similar component with a parallel interface as otherwise perfomance is not acceptable.
Well, firstly they're hobbyists chips, so they're aimed at different markets such as (supposedly) education. So it's hard to moan about a MUCH smaller amount of chips destined for that market.
Secondly, I used a lot of STM chips, but show me one that's $6, has Wifi and has education/newbie resources.
I'm assuming by your confidence that you know the average power consumption of this new chip, and how it compares to the STM line? Or are you evaluating purely on the number of I/Os/pin specs?
Why on god's earth would these chips have the PIO abilities that you talk about?
Mmm, best go storm the AMD production facilities and get them to stop churning out CPUs, so the fab capacity can be given over to Intel...
Not sure just how much clout the the RPi Foundation has with fabs, but I suspect that if the likes of ST, TI etc. were *that* desperate for whatever small slice of production capacity is now being occupied by production of the RP2040, then the RP2040 simply wouldn't be in production right now. The fact that it is in production, along with a load of other chips which you could just as easily argue (*) don't need to be in production due to them just being wheel-reinventers of no discernable use to anyone, suggests that the supply problems with some of the more high-profile chips such as STM32 is a little more complex than a simple case of "is there enough fab capacity of any type anywhere in the world".
(*) at least if you subscribe to the previous posters apparent belief that once there's something already on the market filling a particular niche in said market, no-one else needs to or should attempt to bother coming up with an alternative.
Apples and oranges - if the sort of stuff you're working on really does need 128+ I/Os PLUS a seperate SDRAM peripheral, then clearly you're not working on the sort of stuff where lower pin count micros would be of use. Many of us do however work on stuff where the capabilities of the Pico would be a nice match (especially in this W form - how many other small processor modules are there that feature onboard wifi like this, hmm?), and some of us will even be working on stuff where it'd be considered rather over-provisioned in the IO department.
So if you can't see any real positives in having this on the market, then that's because it's not aimed at you, but it doesn't mean the positives aren't there for others to see and have their interest piqued by...
...some of us will even be working on stuff where it'd be considered rather over-provisioned in the IO department
I'm willing to bet that MOST uses won't need anything like the various I/O it supports.
From memory, that's two I2C channels, two SPI channels, two independent state machines with programmable clock, up to the processor clock speed (125MHz by default, but can run faster), 3 D/A converters (16 bit IIRC), plus a fourth one hooked up to an onboard temperature sensor, and a whole a bunch of bare pins for binary bit-fiddling if you need them. I reckon most people will find themselves hooking it up to one or two I2C/SPI sensors, and maybe an LED or two, and using it to monitor stuff, do some basic analysis, and connecting to whatever is listening for the results of what is being monitored either via the onboard USB or the wireless. Bear in mind that these things are probably cheaper than the sensors you'd be connecting them to, so the attitude in most use cases should be: if you need to do more stuff, use more devices.
If you are after more parallel pins, then why not just use an I2C or SPI device that provides them?
Give me RP2xxx with 128+ I/O and at least a SDRAM interface...
This not the $6 plug-in-and-go development board you are looking for. The RP2040 chip is probably not the $1 chip you are desiring either.
It's got USB, host and device, Wi-Fi, oodles of other interfaces spread over 26 GPIO pins, including three ADC, on a 40-pin carrier. That's plenty for most hobbyists and makers - the people this board is targetted at.
One of the 'sensible' things that has been done with the pico (or at least the PR2040 chip) is a spiffy macropad kit that a US maker (Adafruit) sells for around 50 US rubles; there's also an encoder knob and a cute little display on it, along with programmable RGB lights for all the keys. The kit comes with everything except a USB-C cable- the keys are Kailth Linear Red, and it's programmable using Adafruit's flavor of MicroPython (CircuitPython) and the macropad software is super easy to customize and change on the fly. (Easier than QMK and VIA, at least!)
I'm using it at the moment as the ten key for my new keyboard setup, but it pulls double-duty as a macropad as well for some of the company apps and things I use on the regular.
Such things as I2C serial-to-parallel buses exist. Okay, so you're going to be reducing the throughput by serialising it, so that 100 KHz bus is going to manage a 32-bit parallel bus at around 3 KHz. If you need something faster, I'd suggest that this is not the hardware you're looking for.
Not exactly -- if you look at the board, there's a triangle leading to a rectangle right next to the wireless chip, where the debug pins previously were.
That's a clever third-party licensed antenna that works by having a tuned resonant cavity (there's fibreglass in there, but from the point of view of RF signals, it's effectively void). This first appeared on the Pi Zero W. It's called Niche and it's patented by ProAnt.
While signal lines running close to it may cause problems, the bigger issue here is simply that there's no space on the board at that end any more.
I get it, but it's a shame. The 2.4 band is basically useless in my area unless you are literally underground. 3 non-conflicting channels really limits utility when you have neighbors. Wish it wasn't the industry (bog) standard for embedded systems, as there is WAY more channel space on the 5g bands, and more still up in 6.
If wishes were fishes, I'd have the manufacturers drop 2.4 for wireless data entirely, and issue future single band chips at 5 instead. Instead people's home networks are knocked offline because their printer insists on pooping out it's own SSID out of the box, as does their fancy toaster and security cameras(shudder, streaming video on 2.4 on g-era wireless). And no one turns it off because then they can't connect to their shiny gizmos.
I agree. What's worse, there are still larger devices that ship with 2.4-only WiFi. It's not true in every product type, but a lot of things at the cheaper end leave out that comparatively simple feature. There are a lot of places where the performance on 2.4 is not good at all.
I have a RISC V sbc. It’s good to try out, but it’s not ready in terms of using in projects unless you are prepared to spend the extra time getting all the software working reliably, because it’s all a bit flakey still. For any new project I need to get on with I will still choose established devices with solid support and community.
RISC V will come but as you say yourself, it’s still distant.
Just tried reading the spec but it seems quite dependent on what is used. I'd imagine its low enough to run from a smallish PV setup so if the next Fuzix version can manage the wireless I think I may finally get a wireless feed to my chicken shed to test the Pi4 keeping an AI eye on my chicken shed!