xbox gamer : I would think the Teensy 4 or 4.1 should be included. It does cost a lot more than the Pico but is based on M7 rather than M0 in the Pico but has big memory like the Pico. Pico will probably be a good alternative to the Teensy unless you want the performance.
Your comparison helps me to choose which board to do what, the charts putting the competitors side is already telling me a lot plus your verdict is gold.
The quality of the Pico boards is excellent as well as having castellated edges for mounting directly. The PIO can be used for additional UARTs as well as I2S and even VGA output, currently messing about with it to see what I can make it do.
Your verdict is quite comparable to mine. I was playing with my pico a few hours. After that, I put it aside and wait now until Arduino or platformio start supporting it. I just dislike creating another toolchain. Thanks for the video.
I think the Pi Foundation is planning to enter the Chip Designer realm. The Name RP2040 is already choosen in a way that new versions with multiple cores and such are a possible future option. I personally love the documentation for the chip, it is done so well.
I freaking love your videos man. I might even pass my Maturita exam from IoT thanks to you!! Our teachers know little to nothing compared to you, so these videos are literally life saving for students like me.
@@AndreasSpiess I have sent our teacher some of your videos in the past but he just dissmissed me with "we don't have time for that" and then proceeded to waste time with BigClown kids toys rather than teaching us how to create our own projects which teaches so much faster. Anyways thanks a lot for what you're doing, I appreciate it a lot.
If you think about it these boards are priced way too cheaply. That's why they need to limit how many you can buy. Several years later you can only buy one RPi Zero at a time.
I forgot to add: PIO (programmable IO) would allow to add CAN or I2S in theory (not sure if the resources suffice in practice) without any performance loss on the main processor. Think of it like bit banging done right, and how that makes you more flexible.
I'd like to see this RP2040 integrated in a full Raspberry board, so you get the best of two worlds: general operating system and a microcontroller taking care of I/O interfacing
@@zeendaniels5809 so perhaps we have answered our question. Maybe the reason for this Pico us for people who already own a Raspberry Pi microcomputer and wants to dip their toes for the first time into the microcontroller world they hear about but have not felt comfortable yet trying out until now..maybe they are wanting the comfort of coming from the same brand they are used to?
Hey, thanks for this comparison, this is very nice to see. I think you missed the point of PIO though, since you listed I2S as missing on the RP2040. The RP2040 does support I2S, it just uses a PIO core. You can find the necessary driver in the pico-extras repository by raspberrypi.
You are right with I2S. It seems to be supported in software. As I said: This is not too important in my opinion. I only used it once in a project. I am curious what we will see of the PIO and I for sure will look for useful stuff and I am glad when I was wrong ;-)
The fact the STM32F4 was released in 2011 and its comparable to this 2020/2021 chip, i would argue that even a bluepill STM32F1 released in 2007 is comparable lol
@@vencdee The Esp32 crushes all of these at the same price. Why bother with anything other than an ESP32 it has wifi and bluetooth low energy and has touch pins as well. Its like a pico on steroids'. Why mess with these weeker boards when you can get esp32 chips for 2 or 3 dollars a piece with 4 dollars shipping. It may take a while to get it but .....
@@greyhnd001 I'm using ESP32 they are great but has some flaws, especially ADC is not accurate and gets blocked e.g. by wifi(ADC2) or by other peripherals. So I must still use the other boards. And RPi Pico has advantages like lower consumption from all the boards when active...
True. STM32G4 is a really good modern replacement. The even more impressive fact is that they are using the exact same pinout as before, so you can replace the MCU with zero tweaks.
@@greyhnd001 ESP32 is good if you are using very high level stuff. For anything low level, they are not good at all. Also they are way too power hungry, even without WiFi.
I've been waiting for this comparison! As a new maker without a tie to the Arduino IDE, micropython is appealing to me especially because of the good documentation
I did not understand why you have only 310k subscriber only.I did not find any other youtuber who gives more valuable knowledge than you on wireless devices and embedded hardware design. Any way thank you so much sir for sharing your knowledge and experience in embedded system.
Thank you for the comparison video, Andreas. I too have been skeptical of this release and question the reasons behind it. It is a fair entry into the microcontroller market. Only time will tell. I will likely pick a couple up at some point for evaluation, but not today.
I've been waiting for the microcontroller authority to chime in! Thanks for the review. Sounds about what I expected, and I'll be getting some when the stores restock. I think the Pi foundation created the RP2040 specifically so they could open source it. The Broadcom patents have been preventing fully open sourcing the other Pi boards. Modern microcontrollers seem to have really good datasheets, but nobody is entirely open source so far as I know. I vaguely recall hearing they even plan to release the RP2040 schematics. 😱
for me in 99% of cases, peripherals like integrated radio and software support is way more important than just speed. Otherwise you know your software is inefficient and sucks.
7:40 ..What is everyday useage ? 32 lines gets eat up pretty fast if you are interfacing to a lcd and touch pad, machine or robitc control with lamy input sensors. Also these boards need buffer chips all around because they are vulnerable and run at less than 3 volts.
Your projects are not "everyday projects" for me. They seem to be quite complex. And maybe I would use subsystems with different boards for such a complex project.
@@AndreasSpiess , A arduino Mega or Due has a lot more i/o lines and enough cpu power for many projects. What would you use that kind of cpu power ( pico , es32 ) for with few i/o lines ?
@@jason.doller1. Hi Jason, Thank you for your reply. 1. I need real-time communication. So for the moment I am using ESP32 with mqtt (more or less real-time). I have 2 HC-12 transceivers. When I get those working right I might shift over to a Teensy 4.1 2. I am running out of pins on the ESP32 NodeMCU S2. If I can off-load the motor control and sensors to a Xiao M0 with i2c then I will be ahead of the game.
@@phillipneal8194 Something to consider is getting an ESP32 CHIP and hooking it's serial comm lines to the Pico. You can then use the USB interface on the Pico to program the Pico and the ESP32 (will require a little work, not too much, to forward Pico serial comms to ESP), and then either program the Pico and the EPS32 to work together, OR slave the Pico to the ESP32 by telling it to accept certain commands via serial and do stuff, for example get the Pico to do the motor control, and manage it via serial from the ESP32. I just picked up a couple of ESP32s and a couple of ESP-12Fs today for exactly that purpose - on my robot car, the ESPs will handle comms, display and LEDs, while the Pico handles motor control and sensors. In my case I've strapped a small cellphone battery pack to the rig which powers the pico, and the pico powers the ESP chips. I want to see if I can get away with the ESP-12Fs because of price, but when I need the extra GPIOs (and I eventually will) I'll use the ESP32. The Xiao M0 is slightly more expensive than the pico based on website pricing, and ~50% more expensive landed in my country (ZA). The pico solution will be cheaper and give you significantly more IO ports, power management, AND is only ~50% bigger.
Of course the reason for PIO is to support the various serial channels which do not have dedicated peripherals, like CAN and SIO. I'm particularly interested in the possibility of implementing the USB-C Configuration Channel without needing another chip.
13:40 why make their own chip, design started before esp32 was a thing, might be integrated on the rpi5(a/b) ! ordered 5 pico and 5 ssd1306 to make a clock/alarm ... might be vertical in 2/1/2 config, use pi0 wifi to get gmt ...
thanks for the comparison. one point: i really think the PIO lines are useful: as you can program them as I2S this makes up for the lack of I2S. also you can programm them as additional UART, PWM, I2C, SPI, ... i looked a bit into the documentation - it seems programming these lines is not as difficult and i am sure there will be soon alot of examples for even the most exoctic protocols. by feeding data directly via DMA into the output this should also give a good speed. so all in all this looks promissing. especially for the micropython: DMA and PIO can do all the heavy lifting and there even a slow python programm can then generate good performance. i am also wondering where the PI foundation wants to go with this. certainly it would not make sense to release just one processor. so it will be interesting to see where they take this...
Maybe you are right. I do not judge from the possibility, I usually judge from the problems.Time will tell which of my problems will be solved by the PIO. BTW: in Python most time critical protocols anyway are written in C to gain speed.
Bought a couple pico's last year and just looked at them this week. Have a washing machine that needed a timer so i programmed it to fill until a float switch closed and added times for wash, spin, refill and rinse. Added appropriate relays for the motor and fill solenoids and there ya go! Even put some LEDs on the panel to tell me what cycle it was in.
I think you're underestimating the usefulness of the PIO peripherals. They are designed to implement (along with DMA) high bandwidth and high-speed interfaces. These can include anything from I2S to DVI. There are interesting demo projects showing these off already, and I'm sure libraries making them useful to regular makers will be coming soon. There are a number of architecture details that aren't obvious from the high-level look at cores, clocks and peripherals that make this potentially interesting for more than it appears at first glance, and I'm excited to see what advanced makers do with it and later what libraries will enable for everyone else later on.
@@redleader7988 Exactly this. I know C++ pretty well, I've made programs on pc, but when it comes to finding a damned way to get C++ onto a micro, I just use Arduino. There's very little that I absolutely *can't* do using Arduino rather than raw C or C++ and some esoteric SDK/toolchain.
Thank you for a concise review. Somehow you covered more ground in less time than 2 other TH-cam reviews I watched on the Pico. I wish I hadn't wasted my time on those. Also, I appreciate your comment about the Pi foundation's "notional" pricing policies, notably on the Pi zero. It is a pet peeve of mine. List prices are typically only available through official distributors but only with a variety of unfavorable terms.
I love the Pi and it is great value for the price. A Linux system for 5$ is probably not possible for the moment. But I agree it sucks if you promise something you cannot deliver...
Do you know how something like the teensy 4.1 figures in all of this? I saw it has an ethernet controller on board and was seriously considering the implications ^^
@@Argosh How so? I was doing ethernet with the LPC1768 back in 2011/2012 with the very first mBed boards, before ARM bought the mBed brand. Honestly, the NXP RT1011 is rather more interesting, at LQFP-80, with a 500Mhz Cortex-M7 instead of the 600Mhz variant on the BGA RT1062 on the Teensy4.x. Problem with BGAs is needing to x-ray them to be sure none of the balls bridged. That costs a little bit extra per board. Or, well, since you want ethernet, the RT1020 in LQFP-100.
8:22 minutes into Andreas' video he said he never used DMA, but sought an explanation from the audience. . I have a lot of theoretical/practical experience with other microprocessors/mini/mainframe systems, but little with ARM/ESP32 microcontrollers so what I can share is generic knowledge and nothing specific so I might not directly answer your question, but I'll probably point you in the right direction. . Computer systems often need to move large blocks of data (disk drives, ethernet/WiFi controllers, High-Speed DAQ, Oscilloscope A-D converters). Typically you could use software routines (microcode, machine code, C++) to implement data block moves. . The problem with the "software solution" to data block transfer is that "software code" simply takes too long. Consider that the transfer of data in a computer system could happen at the rate of one byte/word/longword per system clock cycle, while on the other hand consider the number of clock cycles required for several lines of code to move a block of data, then it becomes apparent that moving the data at the rate of the system clock using DMA hardware logic is much faster. . In addition, using the "software solution" to data block transfer uses up CPU resources. . In essence, DMA is nothing more than a simple digital hardware implementation (an address counter that keeps track of the address and some simple I/O control circuitry). DMA is primarily used for high-speed RAM-to-I/O data transfers (disk drives, ethernet, DAQ, etc. Anything high-speed) and sometimes for RAM-to-RAM if the hardware supports this and a block of data needs to be moved in RAM. . Basically, DMA is a technique of moving blocks of data using a hardware solution rather than a software solution. Software is only used to initialise the transfer, but then the "heavy lifting" of moving the block of data is left to actual hardware logic. . Qns: So where would you typically use DMA on an ESP32 or ARM CPU? . Ans: Ethernet/WiFi, microSD cards, USB Block Devices, High-Speed DAQ, Camera. Basically anywhere where blocks of data need to be moved at high speed. In practice, the CPU will never keep up with the required data transfer speed of a camera, disk drive or ethernet controller, where DMA is the only solution . I hope my "generic knowledge" points you in the right direction. Unfortunately, I have very little (none!!) knowledge/experience with ESP32/ARM so I can't go into the specifics. . th-cam.com/video/cVHCllbN3bQ/w-d-xo.html
Thank you for your explanation. I knew what DMA is, but not what the difference of the two implementations are. I just read that the ARM DMA seems to be very good. Whatever this means. This is why I did not want to compare it with the ESP32 DMA. I hope for a specialist who used both and could tell us: The difference is not big ( as your explanation somehow suggests) or: This is the huge difference...
Even better. I stick to ESP8266 (on Wemos D1 mini board), still much cheaper and fit probably 99% of my home IoT needs. Not sure where I would need Pico board.
@@ziomalZparafii I like your design philosophy. This is why I also never went bigger than Raspberry Pi Zero. All my servers in the house are provided and for almost all the time the Pi Zero is using an average of 1% of its CPU. I do like the ESP32 for all the toys in the box it gives to play with but if I was designing for low power current efficiency I could also see going back to an ESP8266, battery and small solar panel(s) for a final design until I reach a point where I find myself hitting my head against the limitations I try to stop myself from defaulting to the most powerful tool in the box.
Contrary it is relatively easy. Tensilica Xtensa cores (used in ESP32) look great on paper but when you start scratching surface you will see that they miss many things, maybe that is reason why most of microcontroller manufacturers pick ARM cores instead of Tensilica even licence for Tensilica is in order of magnitude cheaper
4 ปีที่แล้ว +2
Can't wait for the ESP32-C3 video. :) I guess there's some kind of embargo around these boards, because Bitluni also has one at hand, but still no video.
@ It's not that bad. Just invoke `idf.py set-target esp32c3` [1] before invoking cmake. I didn't have any trouble building the examples; but bigger projects like micropython still haven't landed support patches yet. But I was able to build older projects after converting them to cmake, following the documentation, like github/bozimmerman/Zimodem and a few others, none of which fiddle with the hardware under FreeRTOS directly. [1]: docs.espressif.com/projects/esp-idf/en/latest/esp32/api-guides/build-system.html#selecting-idf-target
This is the review/comparison I was waiting for! The two things that stuck out for me were 1) no radios. A lot of my project ideas need WiFi (or maybe bluetooth to a full Raspberry Pi), so I'd need to get some other board for this, which defeats the purpose of a cheap, integrated chip. If you're going to wire it to a WiFi controller, why not just get an ESP32? 2) using ARM rather than Xtensa could help enable broader compiler support; specifically, it might enable running Rust. The port of Rust to the ESP32 is stalled on LLVM compiler support for the CPU, and has been for over a year, so I think Espressif just lost interest. Maybe the RISC-V core of the ESP32-C3 will improve on that. I had a similar impression as you: neat that they are making a competitor, but nothing really groundbreaking about this.
It is nice for something to be made in Wales/Japan instead of China for a change - sustainable manufacturing! The availability from local suppliers is a great positive for me compared to Black Pill / ESP 32 where you have to wait for slow delivery times from PRC. I wonder if it could be the basis of a hobby grade USB osciloscope given its USB host and DMA support.
The first interesting video on this board ! Thanks a lot for this analysis ! Ahah awesome introduction, I had the same feeling about slice bread ! So there’s still one man on this planet who is able to analyze before saying that it is extraordinary 😄 Congrats !
Thanks Andreas. For novices like me, the Pico has the advantage that the documentation about MicroPython and the online activities, explaining how to use Pico in projects, are brilliantly clear and in simple language. This may not be a board for people with the knowhow - but it's ideal for people like me and the price point means that girls and boys can buy them and get really interested in electronics. As someone else says: "Wait! The guy with the Swiss accent for sure will do it for me!". Thanks for a very clear comparison!
You are welcome. You might also love the Adafruit products. They also have a good documentation and a much broader range (including sensors). More expensive, thou.
@@AndreasSpiess Thanks Andreas. I 'am' on a tight budget but I find Rasp Pi products are so good from the point of view of getting info and help, from forums, that I think I'll stick with them. I have used Adafruit wearables and gone into the Arduino IDE for that but I love that I'm using a subset of Python with R Pi products.
Thanks Andreas, very timely. I also liked the pin out on the board, either add headers for use in a breadboard or you can surface mount the board. Personally I could not see a strong use case for the board for stuff I do, vs the esp32, I do mostly standalone IOT so the wifi is a must have. However I am just started looking at some robotics and could see the pico working well as a interface from the main robot computer and managing hardware subsystems like the motors and encoders of a wheeled platform. Something like the Pico should be a good way of getting the PI community more involved with microcontrollers, I like that.
Was waiting for this comparison, thank you Andreas. Will stick to ESP32 for now, they got very cheap. Also, when can we expect an update on the superpower project? Very interested in that.
Thank you for the video. Have you never used a PSoC 5LP board such as the CY8CKIT-059? It seems very interesting also because of the very intuitive IDE.
@@cr6925 It's not exactly an interpreted language. Some implementations are, others are not. From what I understand, Micropython embeds a compiler, which compiles the code into bytecode, and then every time the system is used, it's this bytecode that is used. It can still compile code on the fly (like for executing commands given in the REPL), but main usage is through pre-compilation. I believe you can even compile the code on your computer and just upload the byte-code if you prefer. If you're good at C/C++ though, you will definitely have it running faster, and take advantage of the strong typing.
@@briceparent593 All far away from the ethos of a pure embedded microcontroller running dedicated compiled code with no overheads to serve one purpose. Still, I guess it gets the gawd awful Python language into "microcontrollers" as imagined by the Raspberry Pi foundation.... IMHO a very bad place to start from. Still, they're academics eh?
Thanks for a interesting video. It's always good to learn about new products on the marked. Most of my projects include Bluetooth or WiFi, so personally I don't see much use for this board. But it will be interesting to see what people make with it :)
Exactly. They're not investing in specs, instead bidding on marketing and reputation. It's like a strategy game where they aim to shoot down the collapsing empire of Arduino and acquire their massive territory in ease.
Andreas, I initially shared your feelings about the limitations of the Pico, until I listened to an Adafruit video where Limor Fried took the pedagogical approach of first criticizing the lackluster list of built-in interfaces, only a limited amount of ADC, I2C, SPI and a number of GPIOs. But then Limor got to the plot twist: how PIO enables you to softload a lot of these 'missing" interfaces into the PIO: The clocked state machines of the PIO is essentially a very limited subset, but much easier to work with, of the flexibility to 'program' hardware that people usually turn to FPGAs to get(or their less expensive siblings CPLDs) Limor recited a lot of hardware behavior that was already implemented as soft loaded state machines in PIO, that she in the previous part of the video stated was missing from the chip: PWM, I2S audio in/out, SPIF, VGA and DVI generation (look in the Pico/RP2040 hardware guide) UART rx/tx, sdcard driver, additional I2C and SPI channels. These two repos list a lot of what was available when Pico launched github.com/raspberrypi/pico-examples/tree/master/pio github.com/raspberrypi/pico-extras/tree/master/src/rp2_common Manchester encoding, hub75 led panels, st7789 lcd driver, some kind of logic analyser. Seems like you can program additional USB channels on the PIO as well, but I am not sure if that is already implemented in sdk, examples or extras. I guess 3D printer builders, roboticists, and other motor control invested folks will want to use PIO state machines to get exact timings of everything from simple motor drivers to trinamic drivers.
I have no problems with the native IOs, I mentioned that the two lacking ones are exotic. If they can be enabled using PIO that is fine for me, but not too important. We will see what useful we will get with the PIO. The ESP32, BTW has a similar (more primitive) functionality. Nobody even talked about it so far. Maybe replacing additional specialized driver chips is a good use case.
Yes, but then again it is a reasonable expectation that sum commment that eh robot parser can't work out iz what will w0rk. Iff da machine can sea what ewe R up two, then it can ignore the comment.
I agree with your verdict. IMO one strategic reason for creating the new chip design could be to use it as a co-processor on future raspberry pis thus offering better realtime capabilities and low power sleep states. Especially for that application it is probably preferable to use a well documented in-house design.
It for sure will be a good coprocessor. As probably most STM32 chips ;-) Power consumption is probably not too important with the Pi5 if I thing about the Pi4.
On one other RPI worshiping channel i was yelled at, when I commented that i dont really see the where would I use pico and not ESP32. Thank you for very honest comparison. One more comment regarding power consumption: Higer power consumption of esp32 doesnt mean that much, considering it is much faster and does the work quicker, and already integrates wifi&bt. You can power esp32 very long time from lifepo4.
Probably most of us buy this pico "just for try", but as so far, it's depend on what project we would build, choose the most adapted board and after choose the most cheap for the same work. Sometimes I buy boards just for trying and I use these some months later for real project.
about the missing I2C and canbus the PIO is basically capable of letting you create whatever type of interface yout want (it's been used to create VGA output, for example)
Thankyou for this video (i'm saying that with a swedish accent :) ). Your channel is extremely informative, much appreciated, since i'm new to this and trying to learn. 😁👍
As always, my friend in Europe with the Swiss accent keeps it real. Thank you. It will be interesting to see if the Pico goes the way of the Pi Zero. Very hard to get initially and because of that I back ordered it and by the time it got here I'd used my favorite board the ESP32. Heck, I still use an esp8266 board for projects.
@@Cam2Art Everything in our realm requires soldering, that is like saying to a soldier you never will have to march or do parade or go camping with only a knife and some string.
I've rematched now this great video Andreas and found one part I've missed somehow before! The assembler example.. that's insane! Never have seen it in use, thats amazing! Thanks as usual! I'm looking forward to your stm32-s3 review, I could use it for self driving neural robot^^
As always Andreas, you save the rest of us a ton of time by doing all the deep research for us! Thank You! If I look closely at some of the images of the RP, I notice (unless I am mistaken) a 3D WiFi antenna in the corner, but I do not see them in photographs of the production model. Did you notice that ?
I got a Pico free with my HackSpace magazine (stuck to the cover). Have not tried it yet. This is an excellent maker magazine, and the February 2021 issue, all in black, has a number of sample applications for the Pico. I too was left scratching my head, saying "But isn't the ESP32 better and faster, with Wifi and BT?". Thanks, Andreas, for filling out all the details so far. I look forward to trying the free Pico when it supports the Arduino SDK.
I think we look at this chip from our side. The SBC people look from a completely different angle. I also think that an ESP32 would have better fitted their needs becasue it can easily be connected to Linux Raspberries.
They did a good job on the default bootloader stashed in the NOR flash. Enumerates as a mass storage device, waits for .u2f, as soon as the host closes the file handle, the board will reflash itself and reboot. Linux/raspis will freak out about this, but it's fine, the FAT filesystem it exposes to the host is completely synthetic in the RP2040's SRAM. Pretty neat. The documentation is wonderful, and I highly suggest you grab the companion micropython book PDF they have made available after the physical copies sold out so quickly. It really contextualizes what an interpreter can do for you on a fast micro. One thing to note though -- almost all of the other microcontrollers Andreas compares have some kind of internal secure boot mechanism... The ESP32 uses one-time eFUSE links that can be blown, while the STM32 tries it's best to enforce readout restrictions without a mass-erase command clearing the option bytes first. The RP2040? No security at all; no *flash* on die at all.. No flash on *package* at all. Whatever is on the SPI NOR flash is what the RP2040 will (try to) boot, no hashing, checksumming, fancy crypto signatures, or any sort of hardware level protection outside of the simplest MPU (Memory Protection Unit) design from ARM's IP core stash. Buying Bare RP2040 chips? No fancy pico mass storage bootloader. Getting an initial program loader onto your companion NOR flash is up to you to figure out, which is a little more of a hassle than people assume... But you can just swipe the foundation's bootloader, it's open! Crop out the mass storage enumeration and you could harden it for productization, assuming nobody ever cracks the lid and chipclips it. Which might just be how you flash the IPL; or opt for SWD over a TAG-connect... lots of ways to do it. Oh, and by all means, check out what _thread can do for you in micropython. It's extremely cool to just be able to run some code in the background on the second core, and halt it at any time. Try out the pye.py on-device editor as well. Tweak your on-device code on the fly, right over the serial port. Forgot a try/catch? No problem! (Though, ESP32 can get real fun with _thread -- while RP2040 is limited to the main thread and one thread on the other core; ESP32's FreeRTOS underpinnings handle juggling around 12.)
I think the pi foundation has more microcontrollers in the pipe. Also, what if they’re priming us for a pi5 that will use a coprocessor setup like a Lattepanda? Just thinking out loud here.
@@AndreasSpiess Not necessarily in the immediate market but I think it would mean more control over your product in general which is perhaps the objective.
Great overview and comparison. Ironic how "pricing" (15:30) on all these chips esculated in 2021 as demand exceeded supply. Hopefully we will see a return to more reasonable supply pricing later in 2022. BTW: The Pico Pi seems to be one of the few Pi options available, that has stock, and pricing hasn't gone soaring. Looking forward to 2023/24 when all the new silicons fabs come online and flip the supply/demand curve.
The Pi Pico only has the name in common with the original. So it is no alternative it the "real" oness are sold out. I am sure things will get better in the next month because now, everybody tries to build stocks...
Great comparison! I got the RISK-V BL602 from Pine64 the same week as the Pico. Both are interesting, but with some issues each. Like why did they not add a Reset button on the Pico. Plenty of space to add a micro switch, and would only have added pennies to the BOM. And with the PineCone, the using jumpers to enter programing mode was a big mistake. But I've worked around that too already...
@@AndreasSpiess the point of the PIO is to do any peripheral you want to. But of course only time will tell if the flexibility of the PIO pays of versus a "standard" built-in. Ease of use and good discoverability, documentation means a lot!
When I started to search for these informations by myself, I thought: "Wait! The guy with the Swiss accent for sure will do it for me!" Many thanks!😁
And he delivered;-)
you lazy lol
Haha, I concluded exactly the same thing. It was only a matter of time, I waited.
@@AndreasSpiess But what about the dedicated FMAC blocks ?
Finally somone comparing it with a proper board insted of Arduino nanos.
If you want to show something is good you best compare it with a 15 year old board ;-)
Most of those vids (the ones I have watched at least) also use an example project that even the Nano is massive overkill for...
xbox gamer : I would think the Teensy 4 or 4.1 should be included. It does cost a lot more than the Pico but is based on M7 rather than M0 in the Pico but has big memory like the Pico. Pico will probably be a good alternative to the Teensy unless you want the performance.
@@klave8511 in my country teensy cost nearly 7 times as much as pico dose.
@@xboxgamer9216 In my country blue Arduino Nano costs exactly $40.
Your comparison helps me to choose which board to do what, the charts putting the competitors side is already telling me a lot plus your verdict is gold.
Glad the video was helpful!
8:18 I think the es32-s2 has a CAN support, it's just called TWAI (two wires automotive industry for license reasons)
You are right. My mistake.
I bought a pair of the S2s specificly for that. Was doubting my choice when i saw the last video with TWAI/CAN not beeing supported.
@Ashok Rajpal Something with trademarks instead of patents, indeed.
@@AndreasSpiess do a video about can bus
@Ashok Rajpal did you compare specs yet? Thanks in advance
Another exemplary video with no commercials. Thank you, thank you!
Glad you liked it!
The quality of the Pico boards is excellent as well as having castellated edges for mounting directly. The PIO can be used for additional UARTs as well as I2S and even VGA output, currently messing about with it to see what I can make it do.
Agreed. I also will look-out for useful PIO projects.
Your verdict is quite comparable to mine. I was playing with my pico a few hours. After that, I put it aside and wait now until Arduino or platformio start supporting it. I just dislike creating another toolchain. Thanks for the video.
You are welcome! I also hope they will hurry up ;-)
As usual, your videos are very informative and no sales & marketing BS ! Thanks again
Glad you like them!
I was looking forward to this comparison :-) Thanks Andreas.
My pleasure!
Looking forward to the Pico W 😊. As always, great video!
:-)
My thought exactly.
I think the Pi Foundation is planning to enter the Chip Designer realm. The Name RP2040 is already choosen in a way that new versions with multiple cores and such are a possible future option. I personally love the documentation for the chip, it is done so well.
Agreed.
I freaking love your videos man. I might even pass my Maturita exam from IoT thanks to you!! Our teachers know little to nothing compared to you, so these videos are literally life saving for students like me.
Glas you can use the content! Maybe you suggest my channel to your teachers ;-)
@@AndreasSpiess I have sent our teacher some of your videos in the past but he just dissmissed me with "we don't have time for that" and then proceeded to waste time with BigClown kids toys rather than teaching us how to create our own projects which teaches so much faster.
Anyways thanks a lot for what you're doing, I appreciate it a lot.
Oh boy I was just thinking about this. Perfect time Andreas.Thank you
You are welcome!
Thanks for this neutral view. Your conclusion fits with my immediate gut reaction.
Glad that I was not completely wrong...
He's Swiss. They're always neutral ;)
@@zetaconvex1987 Except when it comes to banking regulations maybe
Thank you for the video. I no longer fell the need to try to get one of these right away. I will wait until they are available thru normal channels.
Good decision.
If you think about it these boards are priced way too cheaply. That's why they need to limit how many you can buy. Several years later you can only buy one RPi Zero at a time.
That small hand you use to point with is stupidly hilarious
It has a lot of fans ;-)
☝🏼 Proud member of the small hand fan band 🎖️
It's jarring when used with a real human hand in the background
@@AndreasSpiess ya😂
This is BY FAR the best and most comprehensive comparison I've watched on TH-cam yet! Thanks for the content as always!
Glad you think so!
I like your systematic style of presentation - easy to deduce and decide. was very helpful. thanks
Thank you. I am just a trained engineer ;-)
@@AndreasSpiess - exactly what a maker wants - and hopes to become eventually!!!
I ordered the board with the thought that you would do a comparison video soon, you have won the competition the board will arrive next week :)
:-)) I was lucky. I ordered the first day and only got one after I called them...
I forgot to add: PIO (programmable IO) would allow to add CAN or I2S in theory (not sure if the resources suffice in practice) without any performance loss on the main processor.
Think of it like bit banging done right, and how that makes you more flexible.
The State maschins are independent off the main processor. but they run on the same Clock.
0:12 Thank you for the useful video, but as it is breakfast time, I still prefer the sliced bread 🍞 ;-)
Perfect. Enjoy your meal!
@@AndreasSpiess I think this is a mistake- freshly cooked bread is much tastier.
I'd like to see this RP2040 integrated in a full Raspberry board, so you get the best of two worlds: general operating system and a microcontroller taking care of I/O interfacing
I can imagine it will appear on a Pi5
Thanks Andreas - your reviews and projects have helped me through lockdown...
I hope it will end soon. I cannot get my stuff from Germany because of closed borders :-(
the voice of reason after all the hype! My conclusion also - I don't have a direct use case for the RP2040
Especially when you already have arduinos, ESP boards or black pills laying around...
@@zeendaniels5809 so perhaps we have answered our question. Maybe the reason for this Pico us for people who already own a Raspberry Pi microcomputer and wants to dip their toes for the first time into the microcontroller world they hear about but have not felt comfortable yet trying out until now..maybe they are wanting the comfort of coming from the same brand they are used to?
@@pd8559 Indeed, I think that is exactly the case.
Seems like the Pico offers better I/O controller performance at lower power consumption than the others.
@@lawrencedoliveiro9104 lower power thanks to no wifi or bluetooth on the pico
Hey, thanks for this comparison, this is very nice to see. I think you missed the point of PIO though, since you listed I2S as missing on the RP2040. The RP2040 does support I2S, it just uses a PIO core. You can find the necessary driver in the pico-extras repository by raspberrypi.
You are right with I2S. It seems to be supported in software. As I said: This is not too important in my opinion. I only used it once in a project. I am curious what we will see of the PIO and I for sure will look for useful stuff and I am glad when I was wrong ;-)
What is PIO?
@@requiem4adreamc Programmable I/O
@@mikiex Thank you :)
Very fair and informative analysis of the Pico and competitors.
Thank you!
The fact the STM32F4 was released in 2011 and its comparable to this 2020/2021 chip, i would argue that even a bluepill STM32F1 released in 2007 is comparable lol
These are great pieces of HW, so underestimated ! I'm new to microdevices but this I knew for sure !
@@vencdee The Esp32 crushes all of these at the same price. Why bother with anything other than an ESP32 it has wifi and bluetooth low energy and has touch pins as well. Its like a pico on steroids'. Why mess with these weeker boards when you can get esp32 chips for 2 or 3 dollars a piece with 4 dollars shipping.
It may take a while to get it but .....
@@greyhnd001 I'm using ESP32 they are great but has some flaws, especially ADC is not accurate and gets blocked e.g. by wifi(ADC2) or by other peripherals. So I must still use the other boards. And RPi Pico has advantages like lower consumption from all the boards when active...
True. STM32G4 is a really good modern replacement. The even more impressive fact is that they are using the exact same pinout as before, so you can replace the MCU with zero tweaks.
@@greyhnd001 ESP32 is good if you are using very high level stuff. For anything low level, they are not good at all. Also they are way too power hungry, even without WiFi.
Very informative video. Proper tech-journalism way above the standard. Thank you!
Glad you liked it!
I've been waiting for this comparison! As a new maker without a tie to the Arduino IDE, micropython is appealing to me especially because of the good documentation
A good choice!
I did not understand why you have only 310k subscriber only.I did not find any other youtuber who gives more valuable knowledge than you on wireless devices and embedded hardware design. Any way thank you so much sir for sharing your knowledge and experience in embedded system.
You are welcome! The subscribers are growing day by day ;-) But my topics are not mainstream.
Been waiting for this video. Thank you!!
You are welcome!
Thank you for the comparison video, Andreas. I too have been skeptical of this release and question the reasons behind it. It is a fair entry into the microcontroller market. Only time will tell. I will likely pick a couple up at some point for evaluation, but not today.
I think we agree on the situation
Grandpa Andreas! Can't thank you enough :") I really feel like you are the grandpa of the maker community. Always bringing it together :D
You do know right that he isn't that old? I mean look at his hand ! 0:40
You're the best!
@@Hasan... Hahaha He is our spiritual grandpa
@@AndreasSpiess You made my day :') Thank you!
I just wanted to make a similar video.... couldn't get it yet due to backorder in India......All your videos are great and straight to the point.
Thank you!
I've been waiting for the microcontroller authority to chime in! Thanks for the review. Sounds about what I expected, and I'll be getting some when the stores restock. I think the Pi foundation created the RP2040 specifically so they could open source it. The Broadcom patents have been preventing fully open sourcing the other Pi boards. Modern microcontrollers seem to have really good datasheets, but nobody is entirely open source so far as I know. I vaguely recall hearing they even plan to release the RP2040 schematics. 😱
I hope they can do that. Arm IP is not open source.
This video makes me feel like baking some Swiss fresh bread and buy few Pi Picos. Excellent chanel !
Thank you!
Will you do a video on the Teensy 4.0? Seems like that is the one to go for if a lot of speed is needed.
for me in 99% of cases, peripherals like integrated radio and software support is way more important than just speed. Otherwise you know your software is inefficient and sucks.
@@flipschwipp6572 My use case would be huge matrices. Kalman filtering, particle filters. etc.
Maybe you watch my STM32 comparison video?
I was searching for this comparison but no one else ever done it. Thank you for the video
Glad it was helpful!
I thought of this and the video was landed....thank you 🙏
You are welcome!
7:40 ..What is everyday useage ? 32 lines gets eat up pretty fast if you are interfacing to a lcd and touch pad, machine or robitc control with lamy input sensors. Also these boards need buffer chips all around because they are vulnerable and run at less than 3 volts.
Your projects are not "everyday projects" for me. They seem to be quite complex. And maybe I would use subsystems with different boards for such a complex project.
@@AndreasSpiess , A arduino Mega or Due has a lot more i/o lines and enough cpu power for many projects. What would you use that kind of cpu power ( pico , es32 ) for with few i/o lines ?
I agree with your analysis. I will stick with ESP32 for now. Thank you for all your work.
You are welcome!
Will you stick with the ESP32 if your project needs to be power efficient? Or do you mean as a general board to play with?
@@jason.doller1.
Hi Jason,
Thank you for your reply.
1. I need real-time communication. So for the moment I am using ESP32 with mqtt (more or less real-time). I have 2 HC-12 transceivers. When I get those working right I might shift over to a Teensy 4.1
2. I am running out of pins on the ESP32 NodeMCU S2. If I can off-load the motor control and sensors to a Xiao M0 with i2c then I will be ahead of the game.
@@phillipneal8194 Something to consider is getting an ESP32 CHIP and hooking it's serial comm lines to the Pico. You can then use the USB interface on the Pico to program the Pico and the ESP32 (will require a little work, not too much, to forward Pico serial comms to ESP), and then either program the Pico and the EPS32 to work together, OR slave the Pico to the ESP32 by telling it to accept certain commands via serial and do stuff, for example get the Pico to do the motor control, and manage it via serial from the ESP32.
I just picked up a couple of ESP32s and a couple of ESP-12Fs today for exactly that purpose - on my robot car, the ESPs will handle comms, display and LEDs, while the Pico handles motor control and sensors. In my case I've strapped a small cellphone battery pack to the rig which powers the pico, and the pico powers the ESP chips. I want to see if I can get away with the ESP-12Fs because of price, but when I need the extra GPIOs (and I eventually will) I'll use the ESP32.
The Xiao M0 is slightly more expensive than the pico based on website pricing, and ~50% more expensive landed in my country (ZA). The pico solution will be cheaper and give you significantly more IO ports, power management, AND is only ~50% bigger.
Hope you enjoy your summer holiday 2022 with your wife and think about the Pico W and projects that might run with that in the autumn of this year.
I will wait till they support Arduino, I think...
Tommorow i have my Final Exams for my Aprentenceship as Electrican for Automationtechnology but a Video from you must be watched :D
Good luck tomorrow!
Good luck with your exams
Automatiker?
@@gummibear654 ?
Of course the reason for PIO is to support the various serial channels which do not have dedicated peripherals, like CAN and SIO. I'm particularly interested in the possibility of implementing the USB-C Configuration Channel without needing another chip.
True. Let's wait till we see the first useful projects. Probably not VGA for me ;-)
10:00 neither ESP32 nor -S2 support SWD, unfortunately just JTAG
Yes, wanted to add the same!
You are right. I mixed it up!
13:40 why make their own chip, design started before esp32 was a thing, might be integrated on the rpi5(a/b) !
ordered 5 pico and 5 ssd1306 to make a clock/alarm ... might be vertical in 2/1/2 config, use pi0 wifi to get gmt ...
Nice project! I assume you also ordered 5 ESPs for WiFi ;-)
thanks for the comparison. one point: i really think the PIO lines are useful: as you can program them as I2S this makes up for the lack of I2S. also you can programm them as additional UART, PWM, I2C, SPI, ... i looked a bit into the documentation - it seems programming these lines is not as difficult and i am sure there will be soon alot of examples for even the most exoctic protocols. by feeding data directly via DMA into the output this should also give a good speed. so all in all this looks promissing. especially for the micropython: DMA and PIO can do all the heavy lifting and there even a slow python programm can then generate good performance. i am also wondering where the PI foundation wants to go with this. certainly it would not make sense to release just one processor. so it will be interesting to see where they take this...
Maybe you are right. I do not judge from the possibility, I usually judge from the problems.Time will tell which of my problems will be solved by the PIO.
BTW: in Python most time critical protocols anyway are written in C to gain speed.
Bought a couple pico's last year and just looked at them this week. Have a washing machine that needed a timer so i programmed it to fill until a float switch closed and added times for wash, spin, refill and rinse. Added appropriate relays for the motor and fill solenoids and there ya go! Even put some LEDs on the panel to tell me what cycle it was in.
That is what I call a useful project!
Great review, I fully agree, nothing really amazing, but it's good to have as an option
I like the low power numbers for the chip. I don't really have a use for it though.
Thank you!
Great video and really informative. Came across the video from a google news feed. Glad I did. Can't wait to look at some of your previous videos
Thank you. There are around 350...
I think you're underestimating the usefulness of the PIO peripherals. They are designed to implement (along with DMA) high bandwidth and high-speed interfaces. These can include anything from I2S to DVI. There are interesting demo projects showing these off already, and I'm sure libraries making them useful to regular makers will be coming soon. There are a number of architecture details that aren't obvious from the high-level look at cores, clocks and peripherals that make this potentially interesting for more than it appears at first glance, and I'm excited to see what advanced makers do with it and later what libraries will enable for everyone else later on.
I hope I underestimate it and we will get lots of useful libraries. Otherwise they will probably not be used.
Andreas is the Swiss army knife of electronics and his brain is much sharper. Thank you for taking the time to help us all not wasting our money.
You are welcome!
14:20 as someone that learned C++ as their first programming language, I had to LOL at that
After 2 hours of frustration making VS work with the Pico, I quit. MicroPython and Thonny are way too easy and life is too short.
I believe he was commenting on the SDK, not his C++ skills.
@@redleader7988 Exactly this. I know C++ pretty well, I've made programs on pc, but when it comes to finding a damned way to get C++ onto a micro, I just use Arduino. There's very little that I absolutely *can't* do using Arduino rather than raw C or C++ and some esoteric SDK/toolchain.
So far Andreas i am watching a very good video. Thank you! i will watch it all eventually, i think pico has a lot of potential.
Unfortunately, it is not well accepted by the Arduino community and you do not find a lot of projects :-(
Thank you for a concise review. Somehow you covered more ground in less time than 2 other TH-cam reviews I watched on the Pico. I wish I hadn't wasted my time on those. Also, I appreciate your comment about the Pi foundation's "notional" pricing policies, notably on the Pi zero. It is a pet peeve of mine. List prices are typically only available through official distributors but only with a variety of unfavorable terms.
I love the Pi and it is great value for the price. A Linux system for 5$ is probably not possible for the moment. But I agree it sucks if you promise something you cannot deliver...
Thanks for your detailed and fair review, Your channel is a Must for all Hackers, Makers, and Developers!
I hope so ;-) Thank you.
Do you know how something like the teensy 4.1 figures in all of this? I saw it has an ethernet controller on board and was seriously considering the implications ^^
I made Teensy benchmarks in my STM32 video... Different league.
@@AndreasSpiess I was more thinking in terms of usability. With ethernet on the teensy it's potential has just exploded.
@@Argosh How so? I was doing ethernet with the LPC1768 back in 2011/2012 with the very first mBed boards, before ARM bought the mBed brand.
Honestly, the NXP RT1011 is rather more interesting, at LQFP-80, with a 500Mhz Cortex-M7 instead of the 600Mhz variant on the BGA RT1062 on the Teensy4.x. Problem with BGAs is needing to x-ray them to be sure none of the balls bridged. That costs a little bit extra per board. Or, well, since you want ethernet, the RT1020 in LQFP-100.
8:22 minutes into Andreas' video he said he never used DMA, but sought an explanation from the audience.
.
I have a lot of theoretical/practical experience with other microprocessors/mini/mainframe systems, but little with ARM/ESP32 microcontrollers so what I can share is generic knowledge and nothing specific so I might not directly answer your question, but I'll probably point you in the right direction.
.
Computer systems often need to move large blocks of data (disk drives, ethernet/WiFi controllers, High-Speed DAQ, Oscilloscope A-D converters). Typically you could use software routines (microcode, machine code, C++) to implement data block moves.
.
The problem with the "software solution" to data block transfer is that "software code" simply takes too long. Consider that the transfer of data in a computer system could happen at the rate of one byte/word/longword per system clock cycle, while on the other hand consider the number of clock cycles required for several lines of code to move a block of data, then it becomes apparent that moving the data at the rate of the system clock using DMA hardware logic is much faster.
.
In addition, using the "software solution" to data block transfer uses up CPU resources.
.
In essence, DMA is nothing more than a simple digital hardware implementation (an address counter that keeps track of the address and some simple I/O control circuitry). DMA is primarily used for high-speed RAM-to-I/O data transfers (disk drives, ethernet, DAQ, etc. Anything high-speed) and sometimes for RAM-to-RAM if the hardware supports this and a block of data needs to be moved in RAM.
.
Basically, DMA is a technique of moving blocks of data using a hardware solution rather than a software solution. Software is only used to initialise the transfer, but then the "heavy lifting" of moving the block of data is left to actual hardware logic.
.
Qns: So where would you typically use DMA on an ESP32 or ARM CPU?
.
Ans: Ethernet/WiFi, microSD cards, USB Block Devices, High-Speed DAQ, Camera. Basically anywhere where blocks of data need to be moved at high speed.
In practice, the CPU will never keep up with the required data transfer speed of a camera, disk drive or ethernet controller, where DMA is the only solution
.
I hope my "generic knowledge" points you in the right direction. Unfortunately, I have very little (none!!) knowledge/experience with ESP32/ARM so I can't go into the specifics.
.
th-cam.com/video/cVHCllbN3bQ/w-d-xo.html
Thank you for your explanation. I knew what DMA is, but not what the difference of the two implementations are. I just read that the ARM DMA seems to be very good. Whatever this means. This is why I did not want to compare it with the ESP32 DMA. I hope for a specialist who used both and could tell us: The difference is not big ( as your explanation somehow suggests) or: This is the huge difference...
Thank you for the competition:
I'll stick with ESP32 or even Nano...
As probably many of us...
Even better. I stick to ESP8266 (on Wemos D1 mini board), still much cheaper and fit probably 99% of my home IoT needs. Not sure where I would need Pico board.
@@ziomalZparafii I like your design philosophy. This is why I also never went bigger than Raspberry Pi Zero. All my servers in the house are provided and for almost all the time the Pi Zero is using an average of 1% of its CPU. I do like the ESP32 for all the toys in the box it gives to play with but if I was designing for low power current efficiency I could also see going back to an ESP8266, battery and small solar panel(s) for a final design until I reach a point where I find myself hitting my head against the limitations I try to stop myself from defaulting to the most powerful tool in the box.
8:03 Just been looking at the RP2040 datasheet, and on page 324 it mentions doing both I2C and “three-pin I2S”.
It can be done using the PIO (in software), you are right.
@@AndreasSpiess That’s the whole point of the PIO, that it can be programmed - indeed, _has_ to be programmed - to support all these protocols.
I don't see any advantage beside the lower power consumption.
I think it will be hard for any MCU chip to compete with the ESP32.
Agreed.
Contrary it is relatively easy. Tensilica Xtensa cores (used in ESP32) look great on paper but when you start scratching surface you will see that they miss many things, maybe that is reason why most of microcontroller manufacturers pick ARM cores instead of Tensilica even licence for Tensilica is in order of magnitude cheaper
Can't wait for the ESP32-C3 video. :)
I guess there's some kind of embargo around these boards, because Bitluni also has one at hand, but still no video.
No embargo. Maybe you will understand why after my video.
@@AndreasSpiess then my next guess would be "near zero software support" :( Anyway, still very interested :)
@ It's not that bad. Just invoke `idf.py set-target esp32c3` [1] before invoking cmake. I didn't have any trouble building the examples; but bigger projects like micropython still haven't landed support patches yet. But I was able to build older projects after converting them to cmake, following the documentation, like github/bozimmerman/Zimodem and a few others, none of which fiddle with the hardware under FreeRTOS directly.
[1]: docs.espressif.com/projects/esp-idf/en/latest/esp32/api-guides/build-system.html#selecting-idf-target
Can't wait for your review about risc-v and esp32-c3
2 weeks :-(
This is the review/comparison I was waiting for! The two things that stuck out for me were
1) no radios. A lot of my project ideas need WiFi (or maybe bluetooth to a full Raspberry Pi), so I'd need to get some other board for this, which defeats the purpose of a cheap, integrated chip. If you're going to wire it to a WiFi controller, why not just get an ESP32?
2) using ARM rather than Xtensa could help enable broader compiler support; specifically, it might enable running Rust. The port of Rust to the ESP32 is stalled on LLVM compiler support for the CPU, and has been for over a year, so I think Espressif just lost interest. Maybe the RISC-V core of the ESP32-C3 will improve on that.
I had a similar impression as you: neat that they are making a competitor, but nothing really groundbreaking about this.
My next video will be about Risc-V ;-) But you might be disappointed.
It is nice for something to be made in Wales/Japan instead of China for a change - sustainable manufacturing! The availability from local suppliers is a great positive for me compared to Black Pill / ESP 32 where you have to wait for slow delivery times from PRC. I wonder if it could be the basis of a hobby grade USB osciloscope given its USB host and DMA support.
Agreed! Got mine and it's made in Japan
I agree, too. And now since Wales and Switzerland already have trade agreements we get the stuff hopefully cheaper ;-)
The first interesting video on this board !
Thanks a lot for this analysis !
Ahah awesome introduction, I had the same feeling about slice bread !
So there’s still one man on this planet who is able to analyze before saying that it is extraordinary 😄
Congrats !
Thank you!
Thank you. Now I don't feel bad that I don't have the Pi Pico when I have ESP32!
(Will be ordering some when supply is much more available)
Thanks Andreas. For novices like me, the Pico has the advantage that the documentation about MicroPython and the online activities, explaining how to use Pico in projects, are brilliantly clear and in simple language. This may not be a board for people with the knowhow - but it's ideal for people like me and the price point means that girls and boys can buy them and get really interested in electronics. As someone else says: "Wait! The guy with the Swiss accent for sure will do it for me!". Thanks for a very clear comparison!
You are welcome. You might also love the Adafruit products. They also have a good documentation and a much broader range (including sensors). More expensive, thou.
@@AndreasSpiess Thanks Andreas. I 'am' on a tight budget but I find Rasp Pi products are so good from the point of view of getting info and help, from forums, that I think I'll stick with them. I have used Adafruit wearables and gone into the Arduino IDE for that but I love that I'm using a subset of Python with R Pi products.
When I heard about the new Pi Pico, I immediately thought "When will the guy with a swiss accent will make a video about it ??" ^^
ASAP, of course...
Thanks Andreas, very timely. I also liked the pin out on the board, either add headers for use in a breadboard or you can surface mount the board. Personally I could not see a strong use case for the board for stuff I do, vs the esp32, I do mostly standalone IOT so the wifi is a must have. However I am just started looking at some robotics and could see the pico working well as a interface from the main robot computer and managing hardware subsystems like the motors and encoders of a wheeled platform. Something like the Pico should be a good way of getting the PI community more involved with microcontrollers, I like that.
Some people do not like the pin descriptions on the bottom side. Bot it is a good general purpose board with a fair price.
Was waiting for this comparison, thank you Andreas. Will stick to ESP32 for now, they got very cheap. Also, when can we expect an update on the superpower project? Very interested in that.
As soon as we have the boards and tested them...
Thanks for the comparison, very well-structured and presented, as always.
Glad you liked it!
Thank you for the video. Have you never used a PSoC 5LP board such as the CY8CKIT-059? It seems very interesting also because of the very intuitive IDE.
So now we have 300K people in the front row.
Congratulations! Love from India.
True. Thank you!
After all the hype it is great to see a fact-based review. Much appreciated. Micro-python is a non-starter for me. I will stick with C/C++
Yep! An interpreted language in a microcontroller is just daft.
I think Python will take its space in MCUs...
@@cr6925 It's not exactly an interpreted language. Some implementations are, others are not. From what I understand, Micropython embeds a compiler, which compiles the code into bytecode, and then every time the system is used, it's this bytecode that is used. It can still compile code on the fly (like for executing commands given in the REPL), but main usage is through pre-compilation. I believe you can even compile the code on your computer and just upload the byte-code if you prefer.
If you're good at C/C++ though, you will definitely have it running faster, and take advantage of the strong typing.
@@briceparent593 All far away from the ethos of a pure embedded microcontroller running dedicated compiled code with no overheads to serve one purpose. Still, I guess it gets the gawd awful Python language into "microcontrollers" as imagined by the Raspberry Pi foundation.... IMHO a very bad place to start from. Still, they're academics eh?
@@AndreasSpiess Hopefully no. It's not a place to start from when designing embedded systems.
Nice comparison!!! was finding details about all these boards to compare myself and i found this video!
Glad you enjoyed!
Thanks for a interesting video.
It's always good to learn about new products on the marked.
Most of my projects include Bluetooth or WiFi, so personally I don't see much use for this board.
But it will be interesting to see what people make with it :)
I agree. Arduino will add an ESP32 Wi-Fi module to it. Maybe this is a good concept for stable WiFi?
I got my Rpi Pico boards for $4 bucks each. Shipping was a couple of bucks extra. -- Thanks, BTW for making the video. As always, I enjoy watching!
Glad you liked it. Mine was a little over 5$ (plus 8$ shipping)...
Well done! Got MicroPython running on my Pico yesterday. Totally painless and moar boards is always better ;)
Agreed!
I think another great microcontroller board to compare to is the Imx-rt based teensy boards. Lots of performance.
You are right. I compared the fastest Teensy in an earlier video.
The Pi Pico looks like a great way to move kids from the Scratch programing level across to python and electronics.
The Pi foundation has schools as their primary market.
Exactly. They're not investing in specs, instead bidding on marketing and reputation. It's like a strategy game where they aim to shoot down the collapsing empire of Arduino and acquire their massive territory in ease.
Andreas,
I initially shared your feelings about the limitations of the Pico, until I listened to an Adafruit video where Limor Fried took the pedagogical approach of first criticizing the lackluster list of built-in interfaces, only a limited amount of ADC, I2C, SPI and a number of GPIOs. But then Limor got to the plot twist: how PIO enables you to softload a lot of these 'missing" interfaces into the PIO:
The clocked state machines of the PIO is essentially a very limited subset, but much easier to work with, of the flexibility to 'program' hardware that people usually turn to FPGAs to get(or their less expensive siblings CPLDs)
Limor recited a lot of hardware behavior that was already implemented as soft loaded state machines in PIO, that she in the previous part of the video stated was missing from the chip: PWM, I2S audio in/out, SPIF, VGA and DVI generation (look in the Pico/RP2040 hardware guide) UART rx/tx, sdcard driver, additional I2C and SPI channels.
These two repos list a lot of what was available when Pico launched
github.com/raspberrypi/pico-examples/tree/master/pio
github.com/raspberrypi/pico-extras/tree/master/src/rp2_common
Manchester encoding, hub75 led panels, st7789 lcd driver, some kind of logic analyser. Seems like you can program additional USB channels on the PIO as well, but I am not sure if that is already implemented in sdk, examples or extras.
I guess 3D printer builders, roboticists, and other motor control invested folks will want to use PIO state machines to get exact timings of everything from simple motor drivers to trinamic drivers.
I have no problems with the native IOs, I mentioned that the two lacking ones are exotic. If they can be enabled using PIO that is fine for me, but not too important.
We will see what useful we will get with the PIO. The ESP32, BTW has a similar (more primitive) functionality. Nobody even talked about it so far.
Maybe replacing additional specialized driver chips is a good use case.
Free comment to increase viewing related metrics.
Yes, but then again it is a reasonable expectation that sum commment that eh robot parser can't work out iz what will w0rk. Iff da machine can sea what ewe R up two, then it can ignore the comment.
I agree with your verdict. IMO one strategic reason for creating the new chip design could be to use it as a co-processor on future raspberry pis thus offering better realtime capabilities and low power sleep states. Especially for that application it is probably preferable to use a well documented in-house design.
It for sure will be a good coprocessor. As probably most STM32 chips ;-) Power consumption is probably not too important with the Pi5 if I thing about the Pi4.
7:56 What about ADC sampling frequency? Because you know you can trade bits for sampling rate with oversampling.
You are right. I compared them in the other video I mentioned. I thought there is no need to repeat it here
On one other RPI worshiping channel i was yelled at, when I commented that i dont really see the where would I use pico and not ESP32. Thank you for very honest comparison. One more comment regarding power consumption: Higer power consumption of esp32 doesnt mean that much, considering it is much faster and does the work quicker, and already integrates wifi&bt. You can power esp32 very long time from lifepo4.
You find a lot of stuff on Low power and deep sleep on this channel...
Nice overview ! Well done !
:-)
Hello Andreas, Gratulation for 300k Subscribers, many more will follow. Keep going and nice greetings Edi
Thank you very much. You also just got a new subscriber!
Probably most of us buy this pico "just for try", but as so far, it's depend on what project we would build, choose the most adapted board and after choose the most cheap for the same work. Sometimes I buy boards just for trying and I use these some months later for real project.
Here usually the availability of libraries play a big role for decision making. and WiFi, of course.
about the missing I2C and canbus
the PIO is basically capable of letting you create whatever type of interface yout want (it's been used to create VGA output, for example)
You are right. Back then, no libraries for these protocols were available
Thankyou for this video (i'm saying that with a swedish accent :) ). Your channel is extremely informative, much appreciated, since i'm new to this and trying to learn. 😁👍
Thank you for your feedback!
As always, my friend in Europe with the Swiss accent keeps it real. Thank you.
It will be interesting to see if the Pico goes the way of the Pi Zero. Very hard to get initially and because of that I back ordered it and by the time it got here I'd used my favorite board the ESP32. Heck, I still use an esp8266 board for projects.
The chip is only one week old. So we will see where this moves...
Adafruit has plenty, got mine yesterday, $5 with pins which require sodering.
@@Cam2Art Everything in our realm requires soldering, that is like saying to a soldier you never will have to march or do parade or go camping with only a knife and some string.
I've rematched now this great video Andreas and found one part I've missed somehow before! The assembler example.. that's insane! Never have seen it in use, thats amazing! Thanks as usual!
I'm looking forward to your stm32-s3 review, I could use it for self driving neural robot^^
You will be disappointed:-(
As always Andreas, you save the rest of us a ton of time by doing all the deep research for us! Thank You! If I look closely at some of the images of the RP, I notice (unless I am mistaken) a 3D WiFi antenna in the corner, but I do not see them in photographs of the production model. Did you notice that ?
Why would they add an antenna?
I got a Pico free with my HackSpace magazine (stuck to the cover). Have not tried it yet. This is an excellent maker magazine, and the February 2021 issue, all in black, has a number of sample applications for the Pico. I too was left scratching my head, saying "But isn't the ESP32 better and faster, with Wifi and BT?". Thanks, Andreas, for filling out all the details so far. I look forward to trying the free Pico when it supports the Arduino SDK.
I think we look at this chip from our side. The SBC people look from a completely different angle. I also think that an ESP32 would have better fitted their needs becasue it can easily be connected to Linux Raspberries.
thanks! and the micropython workflow demonstrated in linux gets me curious to learn it
Hope you enjoy it!
They did a good job on the default bootloader stashed in the NOR flash. Enumerates as a mass storage device, waits for .u2f, as soon as the host closes the file handle, the board will reflash itself and reboot. Linux/raspis will freak out about this, but it's fine, the FAT filesystem it exposes to the host is completely synthetic in the RP2040's SRAM. Pretty neat. The documentation is wonderful, and I highly suggest you grab the companion micropython book PDF they have made available after the physical copies sold out so quickly. It really contextualizes what an interpreter can do for you on a fast micro. One thing to note though -- almost all of the other microcontrollers Andreas compares have some kind of internal secure boot mechanism... The ESP32 uses one-time eFUSE links that can be blown, while the STM32 tries it's best to enforce readout restrictions without a mass-erase command clearing the option bytes first. The RP2040? No security at all; no *flash* on die at all.. No flash on *package* at all. Whatever is on the SPI NOR flash is what the RP2040 will (try to) boot, no hashing, checksumming, fancy crypto signatures, or any sort of hardware level protection outside of the simplest MPU (Memory Protection Unit) design from ARM's IP core stash.
Buying Bare RP2040 chips? No fancy pico mass storage bootloader. Getting an initial program loader onto your companion NOR flash is up to you to figure out, which is a little more of a hassle than people assume... But you can just swipe the foundation's bootloader, it's open! Crop out the mass storage enumeration and you could harden it for productization, assuming nobody ever cracks the lid and chipclips it. Which might just be how you flash the IPL; or opt for SWD over a TAG-connect... lots of ways to do it.
Oh, and by all means, check out what _thread can do for you in micropython. It's extremely cool to just be able to run some code in the background on the second core, and halt it at any time.
Try out the pye.py on-device editor as well. Tweak your on-device code on the fly, right over the serial port. Forgot a try/catch? No problem!
(Though, ESP32 can get real fun with _thread -- while RP2040 is limited to the main thread and one thread on the other core; ESP32's FreeRTOS underpinnings handle juggling around 12.)
I think the pi foundation has more microcontrollers in the pipe. Also, what if they’re priming us for a pi5 that will use a coprocessor setup like a Lattepanda? Just thinking out loud here.
Well possible. But does all of this need their silicon?
@@AndreasSpiess Not necessarily in the immediate market but I think it would mean more control over your product in general which is perhaps the objective.
Great overview and comparison. Ironic how "pricing" (15:30) on all these chips esculated in 2021 as demand exceeded supply. Hopefully we will see a return to more reasonable supply pricing later in 2022.
BTW: The Pico Pi seems to be one of the few Pi options available, that has stock, and pricing hasn't gone soaring.
Looking forward to 2023/24 when all the new silicons fabs come online and flip the supply/demand curve.
The Pi Pico only has the name in common with the original. So it is no alternative it the "real" oness are sold out.
I am sure things will get better in the next month because now, everybody tries to build stocks...
@@AndreasSpiess Agree, it's going to be 6-9, or more months before inventory pressures ease.
Great comparison! I got the RISK-V BL602 from Pine64 the same week as the Pico. Both are interesting, but with some issues each. Like why did they not add a Reset button on the Pico. Plenty of space to add a micro switch, and would only have added pennies to the BOM. And with the PineCone, the using jumpers to enter programing mode was a big mistake. But I've worked around that too already...
Nothing seems to be perfect ;-)
As far as I understand the RP2040 actually does support i2s. There is even a audio project that uses a i2s adc/dac.
Edit: i2s can be done via the PIO.
Another viewer posted a link which uses the PIO for that. No native support, thou.
@@AndreasSpiess the point of the PIO is to do any peripheral you want to. But of course only time will tell if the flexibility of the PIO pays of versus a "standard" built-in. Ease of use and good discoverability, documentation means a lot!
@@AndreasSpiess The point of PIO is to make it native support as well as being flexible.
I personally would call that being supported.