Raspberry Pi Pico

[Ultrapurple]

Here's a hi-res X-Ray I made of the Pico board. Licence for re-use: CC BY-SA 2.0.

[DrG]

You would think that by now I would understand the allure of a shiny new penny. You would be wrong. Just picked up three of them at a US electronics chain store.

No shipping charge and no wait. Plus, they were $1.99 each (limit 5 then they are $3.99). What could I do?

[ataradov]

What chain store? Not that I need them, I can wait for 3 days for mail to arrive. But if the store does something good, some advertising will not hurt.

[DrG]

What chain store? Not that I need them, I can wait for 3 days for mail to arrive. But if the store does something good, some advertising will not hurt.

Microcenter. They are listed here https://www.raspberrypi.org/products/raspberry-pi-pico/

Edit: This has come up before and I have touted them. I don't know where else to go if I want some component NOW. It may vary by store, but the ones I have been in have a pretty good selection of basic components (all standard brands and retail prices). It is probably a small part of their sales, but I do like to support them.

This is only my second visit since covid. The first was to get a wireless board for a PC that was giving me heartaches and headaches. Also picked up a pack of printer paper ($2.49) and 2, 8 oz. cans of air duster ($4.99 for the two). Ten min. in and out. I am big on convenience and want to see them stay around.

EDIT: when I went to get them, the online inventory said 25+. One day later (today), they are sold out - as I expected, but did not know for sure.

[benSTmax]

I think in the near future we'll see a lot of cool projects made using the PIO. To me, it seems this subsystem is very flexible and sort of unique when compared with other microcontrollers (from other manufacturers)

[NorthGuy]

It is a cool toy, nobody expects anyone to use it in real projects.

They have a picture of these things in a reel on the re-seller sites. If it's not intended for real projects, why pack them into a reel?

[ataradov]

They have a picture of these things in a reel on the re-seller sites. If it's not intended for real projects, why pack them into a reel?

Probably was just the cheapest way to mass produce.

If the MCU itself may be usable depending on a price, the board is just nonsense for mass produced device. In order to reasonably use it, you have to design a base board. And there is nothing special on the Pico that could not just be placed on that main board.

I really want to see the pricing and availability of the MCU itself. If they commit to selling the Pico until 2028, they surely will make bare devices available.

But also, this commitment makes no sense to me personally. It sounds like a huge mistake.

But also, I'm not sure how to interpret the promise of 2028 supply in the document that at the beginning reserved the right to make any changes without further notice.

[josip]

"daul core cortex M0+"
why?
why?
WHY?

I don't have any problem with this. Two high speed M0+ cores with segmented RAM and bus master / slave without wait cycles. Didn't read datasheet completely, don't know if all peripherals are available for both cores.

For me is very important for device to have dedicated IOVDD pin, and not only one supply shared with USB that limit device to minimum 3.0V if USB is used.

[andersm]

I'm still a bit puzzled as to where the Pico board and RP2040 fit in but time will tell.

As the default board for Raspberry Pi users who want a microcontroller. And now everyone who's been whining about Pis being used for blinking LEDs can start whining about Picos being used for blinking LEDs, and not their favourite Chinese half-cent MCU that you can only buy from shady resellers, has no compiler, no documentation, and no support.

[andersm]

GD32 is fully licensed by all IP licensors it uses.

And if you like them, keep using them. It affects me exactly as much as anyone else using the Pico affects you, ie. not at all.

[mikerj]

"daul core cortex M0+"
why?
why?
WHY?

also HOW?

I agree, the M0+core is optimised for low cost and low power, so shoving two of them into a micro instead of one of the more powerful variants (e.g. M33) doesn't make a lot of sense.  I guess it's possible they intended to have one core running something like a TCP stack to remove the complexity/overhead from a users program running in the other one.

The lack of semi-decent analog peripherals is a major omission if they are targeting this to an educational or hobby market.

[artag]

If it really is that powerful, expect clones to the PIO on GitHub with full Verilog code in a few days.

It's the most appealing aspect of the chip to me - makes it very hacky, with clever things you can do by abusing these flexible peripherals.

A PIO state machine isn't a new idea, there are several such micros (though not as cheap). But I don't see it as something you'd implement with an fpga, because if you have an fpga in the system you can just do those programmable operations in fpga code. There's only occasional use for an fpga that is customised so far, but then has the last few bits tailored in software.

[hans]

I'm actually really interested in the PIO state machine, especially if you're emulating data protocols with timers, SPI, DMA, etc. just to make a differential Manchester go pretty over an AC coupled data line. It's all a mess. Having some if-this-then-that capability running inside the peripheral without CPU intervention could be very interesting.

I wouldn't expect this micro to be a bigseller. No onboard flash sounds like a hassle. But it could be an educational tool, as having 2 cores is not "(as) easy to program", and these experimental I/O features may be very useful to see in other microcontrollers.

Considering the FLASH/SRAM situation I would look at it this way: you can get a STM32L431 with 64kB RAM. With a few lines of linker & assembly code you can segment the SRAM into 16kB of data and 48kB of code, and save 1/3 power (energy). Depending on what you're doing, 48kB of code is still plenty for low power IoT stuff.
With 264kB of SRAM you could provision 64kB code for each core and have a shared 136kB for data. Or do 128K/64K code + 64K data. Optionally you can still run large libraries from FLASH. etc. That still sounds like plenty of memory for such a small and relatively slow core. It's not like this MCU has on-chip peripherals for Ethernet, camera, USB2.0HS, high resolution LCD controllers etc. that can easily chew up all the memory for assets.

I will probably order one too whenever I get around to it.

"daul core cortex M0+"
why?
why?
WHY?

also HOW?

I agree, the M0+core is optimised for low cost and low power, so shoving two of them into a micro instead of one of the more powerful variants (e.g. M33) doesn't make a lot of sense.  I guess it's possible they intended to have one core running something like a TCP stack to remove the complexity/overhead from a users program running in the other one.

The lack of semi-decent analog peripherals is a major omission if they are targeting this to an educational or hobby market.

I think we take great ADC's/DAC's/comparators/opamps for granted from large MCU vendors. I was looking to use the ESP32 ADC lately and I was shocked what I saw. I was almost contemplating just putting an external I2C or SPI ADC just so I don't have to crunch my brain over calibration data, non-FS ranges and large sampling capacitor values.
Still this part is amazingly popular. Of course for it's WiFi/BLE capability.. but still.

I think it goes to show that designing digital systems can be facilitated/prepared very well with modern & free (as in free beer) FPGA tools. Yet analog is still much of a craft that requires a lot of knowledge to get right.

[iMo]

..For the curious, the chip cost 3-4 million GBP to design. This is about what you expect for a design like this. You could probably make it a bit cheaper, but at much higher risks.

Not today and with this chip.. 1/6 of that incl. set of masks..  EDIT: 1/4 -> 1/6
And it needs to be arduino compatible as that made other chips great

[guenthert]

The lack of semi-decent analog peripherals is a major omission if they are targeting this to an educational or hobby market.

    Is it?  The Parallax P8X32A has, other then built-in RAM, counters and a boot ROM, no peripherals at all (not even built-in E(E)PROM or UART) and Parallax expressively caters to the educational and hobby market.  Sure, they are a small player, but are around for more than 30 years now.  Guess, it depends on whether you want to teach how to fade a LED or how PWM works.

     The new (still fairly expensive) P2X8C4M64 has a number built-in 'peripherials', ADCs, DACs, UARTs and filters.  So maybe they did see a need.

[DrG]

I think predicting the success, in the sense of having an impact, of a new product like this is fraught with difficulties. I am not going to bother with an Arduino comparison, but want to make two points from my perspective (not an EE) and my personal experience.

I got back into small controllers on a whim when I purchased an Intel Galileo I. I had only heard about 'Arduino' and was drawn to 'Intel' more than anything else. This board was, in my opinion, a disaster from a performance and "arduino compatability" point of view. They came out with the II which was not much better, and then it faded away.

I later bought an 'Intel' Quark D2000 board (hey, maybe Intel would get it right this time ). I really liked that board because the source code was available. I could go look at the firmware code and learn how to do what I wanted to do and I thought Eclipse was a fine platform (and still do). This sounds odd but I appreciated that it did not have the plethora of "libraries" that allowed users to do something, which they didn't understand at all, far too easily. There is a balance there, although I am not certain I can operationally define it very well. Sadly, the board was slower than a Republican accepting election results and it too (along with the Quark) faded away without having much impact.

Seems to me that the hardware is one part of the equation and the support is the other. Both are needed for impact. The ""EE brain" values the former more than the latter, which makes sense because their onboard KSAs (knowledge, skills and abilities) are well suited to make use of the hardware with limited support. The "non-EE brain" values the latter more than the former, which makes sense because their on board KSAs may not be of much use using the hardware with little support. 

RPI has a brand that is steeped in support (documentation as well as user community). Is that enough to overcome the limitations of this $4 board? I have no freaking idea But, reading this thread suggests to me that, for a $4 piece of hardware, it "aint" too bad....so far.

[Rasz]

PIO allow it to bitbang DVI 720@30 https://github.com/Wren6991/picodvi should be able to bitbang a ton of slower stuff with ease. 1Gbit (RGMII) ethernet, IDE PATA, SCSI, all kinds of parallel camera and display interfaces.
Too bad there are no fast serializers/deserializers on board, and hardware USB is only 1.1

I purchased an Intel Galileo I. I had only heard about 'Arduino' and was drawn to 'Intel' more than anything else. This board was ...
I later bought an 'Intel' Quark D2000 board ...

Did you own a Zune by any chance? https://www.inc.com/jessica-stillman/customers-of-doom-who-you-dont-want-to-buy-your-product.html https://journals.sagepub.com/doi/abs/10.1177/0022243719867935
Just browsing their joke datasheets was a clue enough it wasnt a serious attempt at anything (few GPIO with KHz speeds etc). Intel wont enter a market with less than $1B of profit potential, they passed on iPhones, they sold StrongArm. What did you expect?

[Bud]

And now everyone who's been whining about Pis being used for blinking LEDs can start whining about Picos being used for blinking LEDs

It seems you've not been around for long enough to remember that blinking LED was the EXACT way the PI was marketed.

[andersm]

And now everyone who's been whining about Pis being used for blinking LEDs can start whining about Picos being used for blinking LEDs

It seems you've not been around for long enough to remember that blinking LED was the EXACT way the PI was marketed.

I've been around for long enough to grow tired of the constant whinging about how awful and terrible and wasteful that is. Well congratulations, now you have something new to complain about.

[DrG]

PIO allow it to bitbang DVI 720@30 https://github.com/Wren6991/picodvi should be able to bitbang a ton of slower stuff with ease. 1Gbit (RGMII) ethernet, IDE PATA, SCSI, all kinds of parallel camera and display interfaces.
Too bad there are no fast serializers/deserializers on board, and hardware USB is only 1.1

I purchased an Intel Galileo I. I had only heard about 'Arduino' and was drawn to 'Intel' more than anything else. This board was ...
I later bought an 'Intel' Quark D2000 board ...

Did you own a Zune by any chance?/--/ What did you expect?

I did, in fact, own a Zune for about 12 hours and returned it for a full refund and the clerk understood and showed me a pile of returns. Subsequently, I bought a cheap Sandisk and I still use it today.

I did not know what to expect from the Galileo. Since you obviously did, you should have no problem at all predicting the future and making large sums of money - maybe you could email a few bucks to "make me whole".

[hans]

PIO allow it to bitbang DVI 720@30 https://github.com/Wren6991/picodvi should be able to bitbang a ton of slower stuff with ease. 1Gbit (RGMII) ethernet, IDE PATA, SCSI, all kinds of parallel camera and display interfaces.
Too bad there are no fast serializers/deserializers on board, and hardware USB is only 1.1

That's a really cool demo of the hardware while only a tiny bit outside the normal spec. But I think this is where it gets tricky. A few posts back I commented that no one is going to use this micro to use it with high bandwidth peripherals such as USB2.0HS or Ethernet, because they are not on-chip, and therefore the perhaps a small SRAM after segmentation is not a problem.
Okay, obviously people are going to implement them using PIO and explore the limits. But similar to how you can bitbang VGA from a STM32F4, you probably can make something to work and create a cool demo for the aforementioned peripherals. But to then actually have the CPU and memory subsystem remain useful enough to do something else..

Like, I was amazed when Microchip launched the PIC24FJ GC006 series. A 10MSPS 12-bit ADC in a hand-solderable MCU! Yet the MCU only has 8KB SRAM, runs at 16MIPS andPMP as high bandwidth peripheral (which can't really interface with much, maybe a FPGA). So good luck doing anything with that data, especially for prolonged periods of time. The ADC can fill the entire MCU SRAM in under 400us. Talking about a bursty workload..

In that regard I'm still interested in the concept of PIO on the other side of the spectrum: namely in ultra-low power operation without CPU intervention. MSP430 can intelligently clock gate peripherals in bursts to perform short workloads. EFM32 has a 'peripheral' reflex/event system that works even when no HF clocks are running. I think it would be awesome to see a MCU that could combine several of these concepts (e.g. PIO that can also run using LF clocks, perhaps with gated oscillators/clocks that can chain/generate events or IRQs) for ultra low power operation. Sometimes only a few gates or flip-flops of logic is all you need...

[jonroger]

Not clear to me why they didn't just support bare metal programming on the Zero and Zero W.    Performance is so much better.

[ataradov]

They wanted to grow their business and enter a new market. Nothing wrong with that. And this gives them ability to release better peripheral sets in the future, since current vendors are dropping the ball on this hard.

MCU market is not that easy to enter, but they have huge customer base out of the box, so that helps.

If you judge the performance of this chip on its own, it sucks. But if you look at what possibilities it can open in the future, it is great.

[luiHS]

 
I have no interest in this product.
I am not going to stop using the NXP Kinetis, i.MX or the STM32 from ST, to replace it with an unknown product, Cortex M0 (I use a minimum of Cortex M4). And worst of all, without a serious IDE like the ones based on Eclipse (MCUxpresso, STM32CubeIde).

What IDE does Raspberry offer for their microcontroller? They talk about Arduino, which seems like a terrible option.

A good IDE and a good sample source code package is the least I ask of a microcontroller to decide to use it. That's what NXP with MCUXpresso offers me, that's why it's my main product, and I sometimes turn to STM32s for special things like MIPI and hardware JPEG decoding.

I'm not going to buy these Raspberry microcontrollers, no matter how cheap they are, I already have Teensy to play with, and for professional development I make my own PCB using NXP or ST microcontrollers.

[ataradov]

Again, what makes you think they will not build up this infrastructure. Nobody just develops things for decades and then releases it all at once.

The exact same thing was said about esp8266 when it came out. Now that thing is in commercial products shipping millions.