This is a good article but it's a few years old, and out of date now. I can't see its date with my current adblock settings. It would be great if someone could find it and put it in the title.
There are stupendously powerful $1 microcontrollers now, including the RP2040 (dual core cortex m0+ with 264KB ram on the chip), ESP32-S2 (Espressif core with 400k of ram and wifi), and ESP32-C3 with wifi and RISC-V core and again something like 400k of ram.
Jay Carlson also has a series about microcontrollers as cheap as THREE CENTS, i.e. the famous Padauk 3 cent MCU:
> It would be great if someone could find it and put it in the title.
There's no publication date -nor anything marking it as updated at any time afterwards- on the article itself, but first comments are from November 2017 and passing references in the article point to "end of 2017" too.
The article about the Padauk is interesting. Especially the description of the FPPA feature. Multiple CPU contexts? This sounds suspiciously like SMT to me which would be a total novelty for a micro controller (afaik).
The ESP32 has two separate cores, not SMT. But look at the Parallax Propeller which goes back quite a long way. Also the Padauk is not exactly SMT afaict. It's more like time slicing. The idea of SMT in eg. an x86 is that one thread can run while another is blocked waiting for memory.
Sounds like a barrel processor [1], used to hide latency of individual instructions to allow code to be written without having to think as much about instruction scheduling / dependencies.
The Xmos chips give the programmer virtual time sliced contexts over one or more hardware core(s). Say you have a 400 Mhz base clock and 4 threads, each one would see the wall clock time advance at 100 Mhz. This is useful for having concurrent realtime operations on each hardware thread w/o having to resort to trickery.
These chips are ideal for DSP, phased arrays and beam forming. Most of their devkits are centered around voice applications.
It has an interesting trade-off: no internal flash. But I've found that I add a SPI-flash chip to the MCU in many designs, so it's a reasonable choice.
NXP has rt1020 (Cortex-M7, no flash), or rt1024, which is same but 4 MB flash in the package- it's connected to the SPI-flash controller of the rt1020 core, so I think it's a multi-die package.
The RP2040 actually has built in 12Mbps USB 1.1, no serial bridge required. But if you do want to use a serial to USB bridge the max baud rate is 921600.
Those are the three micros I use for projects/products, too. A while back I bought a few hundred of the Padauk chips with the intent of doing a small manufacturing run of a side project, but I haven't seen them in stock at the normal places for a while. It looks like the lead time for cheap micros in general is about a year.
What I want more than $1 microcontrollers is a series of controllers that goes from low price/low performance to higher price/higher performance, but all programmable with the same open-source software stack.
I don't want to read datasheets again whenever I need 100MHz more clock speed, or 1mW less power consumption. Because reading datasheets, at some point, feels like filling out tax forms.
The STM32 line is as close as you will get with this.
Its HAL (Hardware Abstraction Layer) is MIT licensed and is very similar across products. Products in the same family are nearly pin compatible.
They are remarkably similar even at the register level - when you program with the LL (low level) library that essentially exposes hardware registers. I find LL easier to understand and work with than HAL.
BTW the license is mostly BSD not MIT with some additional libraries more restrictive.
...which has 2.14 DMIPS/MHz = 1.027 BIPS (more than 1 instruction per nanosecond). Fast enough?
(BTW, when I see "mhz" I assume this means "milli-Hertz". Similar when people say they have a 42b structure -- I see this as a 42-bit structure. If I wanted to communicate "42 byte structure" I'd abbreviate as "42B structure". Capitals used judiciously transmit meaning compactly.)
FWIW, I've seen capitalization issues in datasheets. When they're this ubiquitous, it's better to use it as an opportunity to apply number sense as a sort of FEC for verbal understanding. A 480 millihertz ARM doesn't really make sense (if you were running at those rates, you're in such a niche that it has ISA considerations where ARM isn't a great option anymore). Similarly a 42 bit structure on an ostensibly 8 bit machine doesn't make sense. Yes it'd probably be a better world if everyone was consistent, but blindly assuming that they are doesn't really help anyone.
Note, I didn't assume anything. But checking assumptions 'should' be 2nd nature for any geek.
As for millihertz: sure it could make sense, if the design were fully static and one was watching (with eyeballs) some LEDs on control and address and status lines -- in fact, in many 'homebrew' computers, having a setting to allow seeing the sequencing of instruction execution is often a built-in feature.
MHz. B for Bytes. $B for Billions of Dollars (not "$BN") etc. Rules like this reduce our overall cognitive burden, when followed by all. That, in turn, makes us more efficient.
"BTW, when I see "mhz" I assume this means "milli-Hertz"."
> As for millihertz: sure it could make sense, if the design were fully static and one was watching (with eyeballs) some LEDs on control and address and status lines -- in fact, in many 'homebrew' computers, having a setting to allow seeing the sequencing of instruction execution is often a built-in feature.
I didn't say the the frequency didn't make sense, only that it doesn't make sense as an ARM, as that niche ends up having ISA considerations where something like ARM doesn't make practical sense . Particularly given that I said "up to 480mhz".
When I see an abbreviation, I assume it means something. Do you think that when one sees a word, assuming it has the usual meaning is 'assuming' something? Only in the most extreme sense could that be true. If I assume you know French, yes, that is an 'assumption' because -- I have no knowledge if you know French or not, so for me to say "You seem to know French" would be an assumption.
When I say "MHz" I 'assume' it is an abbreviation. Because, it does not parse as a 'word'. Therefore, the next question is: WHICH abbreviation is it? Does it have a Capital M at the beginning? If it does, then that usually means "Mega" in Scientific/Engineering circles. Does it have an 'm' in the beginning, such as ml? then 'm' means 'milli-' like it usually does with Scientific/Engineering units. When I see 'mhz' --- well, first little 'm' means milli, h -- oh, I see, it's Hz but not properly capitalized, ok, so it's milli-hertz. oh, what, no, I have to know that ARM (supposedly) cannot operate at the milli-hertz range, so now -- with all that context --- I can now make an educated guess what 'mhz' means.
Really?
Are you suggesting I should not ever assume an abbreviation means what it usually means? And I must adopt your way, which is fully context-dependent (and not even accurate in that case)?
I do feel as if you may be a 'word lawyer' and so, I will stop trying to communicate, since this is going to be painful for me. Good day.
I've wanted something similar when shopping at Adafruit. https://www.pcpartpicker.com has some pretty good search tools that I wish more vendors would copy.
Be careful spending time writing software/firmware for chips that may disappear tomorrow. Cost of the chip itself isn't everything, in fact it's one of the least important things if you're looking at long term cost. This is especially relevant in the current chip shortage climate which has made this point quite salient.
Whether it's a hobby project or a high volume production project, you don't want hundreds of hours of your time spent developing for it to go down the drain. In case of a hobby project, it probably means the one or two units you made are all that you're ever going to make. In case of a product you built in high volumes, it means a large gap in sales and lost revenue while you re-do the firmware for a new chip.
This is a real concern. A lot of these offerings are not useful now because they are not ARM (or, soon, RISC-V). Even with instruction set match, moving between vendors is a huge issue.
The Texas Instruments MSP430 is another good one to look at. At one point, dev kits were available for under $10. It's a good choice for low-power applications. Comes with a free IDE.
I have been struggling to get TI's (free) Code Composer Studio to recognize my MSP430FR2311 boards now on-and-off for months. They 'upgrade' the code, and then I can't program my devices. In fact, so far, I've spent more time trying to get the dev environment to work than developing my code. Not Fun.
Very cool when it does work, tho. I'd prefer to pay a little more for parts and get some English-based support.
I'll work through the Code Disposer Studio issues...again...
It helps if the USB cable one is using actually provides more than just Power and Ground. (red-faced blushing....)
I take back all the mud about CCS; except of course now it wants to put a new bootloader on one of the boards, but seems unable to do that. Good thing the other board is fine.
Wow, didn't know about that. Interesting! thanks. The MSP430 branch is meant for low power; and or getting you and interface translation without needed an FPGA. They are cool -- sort of like pdp-11 in philosophy.
Oh yes, the PIO is very cool and somewhat unique. The list of examples of things that can be implemented with such a tiny "computer" (9 instructions assembly language, program size maximum of 32 instructions) is quite impressive:
In fact, the rp2040 examples and SDK docs are very, very good compared to most traditional manufacturer's documentation and by itself makes it a very good value proposition for hobbyists.
It's up to hobbyists what's worth it to them, but the power consumption of the ESP 8266/32 is not _really_ comparable to the '430. "Deep sleep" mode on the ESP seems to be about 20uA, an MSP430FR5969 can go to a 20nA LPM 4.5 mode (factor of 1000 lower). Harder to compare in a HN comment, but active mode power consumption is also in the MSP's favor for many applications particularly when using the built in hardware accelerators like LEA.
Of course the MSP430 doesn't have a WiFi modem or a giant hobbyist community like the ESP, so I'm really not trying to make an argument for which is "better" just trying to add some contrast to the low power features you asked about.
I think the platform is very much still alive and kicking. It is still the only uP platform I am aware of which is available with embedded FRAM which is low power, non volatile, extremely more write durable than flash and rad hard. Again, maybe not top selling features for a hobbyist project but absolutely compelling for certain industrial use cases.
Thanks - you got me excited about MSP430 again. I am wondering if I should build an AC motor controller for fun, since I happen to have an AC motor on my bench.
No problem! You are not the first person I've heard express the impression that the MSP430 is slipping into irrelevance, and while I agree the places it shines may be getting squeezed by low power ARM parts I think they can be pretty compelling for applications where they are a good fit. They are simple, yet modern enough to support full featured c compilers, have a history of focusing on low power and the FRAM is a pretty solid differentiator if your use case can take advantage of it. Jay's article here can't really show any of that off because the one dollar price point is below the sweet spot where most of the nice features kick in
Semi-related: Since few months I am digging into tinkering and electronics. It’s really great to learn about modern era solutions which enable cool projects with low entry barrier.
However, I happen to learn about new solutions (new sensors, actors, boards, ICs) more randomly than systematically. I tried to google for databases, blogs or news outlets, but couldn’t find anything suited for my consumerish needs. Going through hundreds of pages for similar products on online shops also didn’t yield any good output. Do you guys know any such source? How do you keep yourself updated on interesting new products on the market?
Usually if you have accounts at the major distributors(Arrow, Digi-Key, Newark, etc.), you'll get occasional updates passed on from some of the manufacturers whose parts they sell. But those are typically not the hobbyist/consumer items you may be looking for.
Perhaps sign up for a SparkFun account? I get a lot of OSHW and new module information from them.
(total noob) I like looking at the components adadruit builds in-house. For most of them they have open-sourced the schematics, and you can then look up the building blocks on sites like DigiKey.
Absolutely, but I find them more valuable because of that. I don't know enough to pick the ideal component, but I know if I get what they use it'll be good enough.
It's amazing that humanity can produce this sort of hardware under $1. This has implications that are more profound than previous industrial revolutions.
In centuries from now, our time will be studied as a remarkably revolutionary one.
I would argue that this kind of price is only doable because people and resources are being exploited for it. Not specifically only for these microcontrollers, but most our consumer products in general.
In a worst case scenario, the resources for this $1 microcontroller were mined by an african child, barely earning anything. These resources are then shipped to China(probably), where people along the way don't earn much either. Except the white collar workers managing it all, of course.
These controllers are then built in a factory, either by robots, or low-earning workers. Then shipped to a warehouse.
For me, the fact that such things can be bought so cheaply is not really an achievement if we look at how broken the supply chains are if we consider the humans who are affected by them. And also the environment which gets destroyed quite a bit.
I am of course talking from a moralist and empathical perspective. What I am trying to say is, that this is only an achievement because we accept or ignore the drastic consequences.
>I am of course talking from a moralist and empathical perspective.
Speaking from a business and engineering perspective, I'd be surprised if (even in an ideal world with no environmental externalities or exploited third-world labourers) the cost of the raw resources, shipping and e-waste processing were a significant part of that $1.
We're literally talking about something 1/16 the size of a postage stamp that gets stamped into a similarly tiny plastic shell.
From my admittedly "peak mount stupid"[1] position as an engineer that uses these things, I'd suspect that the biggest contributors to that $1 figure come from the expert labour required to both R&D the chips as well as run the machines that manufacture them.
I'm sure there's truth to that. But I also think a large proportion of the price (and in general the plummeting price, and apogee-ing performance of many high tech components, such as batteries) is due not to exploitation, but to increases in efficiency. The bequeathed legacy of the thousand accreted technologies and improvements that have layered to produce contemporary supply chains.
Of course...everything we do today is based on exploitation: 1000s of years of exploitation of humans, frequently the worker and poorer classes' labor, 100s of years, and billions dead in war...all our achievements today literally stand atop this ginormous pile of blood and death...but, that ever-present historical reality aside...I think a large proportion of the skyrocketing perf and ocean-depth-plumbing prices of components is due to the compounding and intersecting effects of all our technological efficiencies.
The first step in solving these problems is getting the accounting in place, so we know exactly who is paying what. It is beyond me why we are not even doing that.
Anyway, we could solve these problems if we doubled the price. And then it would still be an amazing achievement.
Given our love for destruction, and ongoing events, it will be talked about the same way as we talk about Babylon, Egypt, Sumerian, ancient Roman and Greece, Aztecs and Mayas.
I've used a lot of micros as a hobbyist, and many on this list. For my time and money, I like the PSOC, even though the author pretty much dismisses them.
PSOC always seems to be what I need. I've used it to read CCDs (timing is very strict, and works much better on the SOC than in software). I've made capacitive soil moisture sensors using the capsense. The psoc5 has good options for USB devices.
But I don't make things to mass produce, so $1 has no appeal to me. $10 dev board is fine for me, if it does what I need it to do. There's a lot to love about the PSOC, but if you ignore all of the things they do that these other microcontrollers don't do, it might not look great by comparison. Also they have a development environment that is free and easy enough to learn that it doesn't require living in datasheets.
Read this when it came out, it helped me to understand that with modern silicon geometries or process nodes, computers are "free". The largest part of the cost here is the packaging and testing of individual parts.
None have a PC-architecture (x86+other HW standards), so no, they aren't. It would be possible to make a microcontroller that can, but nobody does. Intel did make a few experiments with x86 microcontrollers, but even those were afaik not PC-like enough, because you generally don't want a lot of those things in a microcontroller.
Generally a microcontroller has no MMU. A microprocessor may mean the same thing. The industry term for something with an MMU is an application processor (Cortex-A). There are other terms that can be used, but it will need to be specified.
There is some fuzzy area between MMU-less, small processors, and processors with no MMU but large enough to run Linux in a MMU-less way.
It's not very hard at all. The issue is exclusively licensing fees.
Cheapest depends. There are Allwinner parts that are $2? But with the move to RISC-V, we will see costs decrease further, probably so that a Linux-capable part is just as expensive or cheaper than an MCU.
Whenever possible I reach for a part that can run Linux, because most of any embedded task nowadays is connectivity. A lot of parts include a microcontroller on the same die for realtime tasks.
There are some recent x86 SoC's which basically implement i586 and have been made to run both MS-DOG and Windows 9x. Commonly used by retro- and vintage computing hobbyists. None that would cost $1, though.
Things called "microcontrollers" usually have these properties:
* They are actually CPUs, but with very weak feature sets related to desktop & scientific computing
* They may or may not have hardware to do any kind of advanced computation, starting with plain old multiplication and division of integers (addition and subtraction of integers is enough for what they do). Yes you read this correctly. The vast majority of 8-bit microcontrollers cannot even do division or even multiplication in hardware.
* They run with clocks from kilohertz to low megahertz
* They might be 8-bit, 16-bit or 32-bit, with rarity going up as the bitness goes up. There's a ton of 8-bit microntrollers, and 32-bit are much rarer. 64-bit ones are very very rare, since if you need that processing width, you might as well go for a single-board-computer like a Raspberry Pi.
* They are used to DIRECTLY drive other hardware, not in an abstract way. They literally regulate voltages on their pins to make stuff happen with connected hardware.
* They might have RAM space ranging from a couple of BYTES (yes, BYTES, see e.g. this one having 32 bytes of RAM: https://www.microchip.com/en-us/product/ATTINY9#) to a couple of hundreds of kilobytes. It's very rare to have a microcontroller with a megabyte or more of RAM, because, as before, in that case you might as well just use a full-featured desktop-ish CPU.
* They are usually lacking a special bit of hardware called a MMU (Memory Management Unit) which is used by "real" operating systems like Linux and Windows to manage process isolation and many other useful things related to memory. If you need a MMU, again, just use a SBC.
So where do we use microcontrollers? Well, what do you think happens in your garage or car keyfob when you click on the buttons? What tells your microwave when to finish and with which power level to nuke your food? What drives your electric toothbrush? Your car's gas valves? Microcontrollers, that's what.
They are currently (or have been a few years ago) so cheap we can readily add a microcontroller to every toilet paper leaf, and it would only bring the price of a roll up by a dollar or so. What's stopping us to do just that is the sorry state of current batteries and energy harvesting in general - those technologies simply are not as developed (yet).
So, no, you wouldn't describe a chip which can run Windows 98 as a "microcontroller".
I stopped tinkering with electronics despite being a full-time embedded software engineer. I have designed small boards, but I only know enough EE to get myself in trouble. I prefer working with EEs that know their stuff, where our skills can compliment each others’.
I no longer work on “hobby” projects; such things become an albatross around my neck. If you deploy a hobby project in your life, you must maintain it. Forever. For that time cost, I want to work on a product that could go into production and pay for its own maintenance.
Unfortunately, I cannot stand being around people anymore. I had my entire life savings stolen by an ex, with the government’s help. When I tried to open back up and make new friends after that shit show was over, I was drugged and raped by the first guy that I invited into my house in over two years; he was never even arrested for what he did. And after over 25 years, I stopped playing music, because those “friends” all actively abandoned me when I decided to start carrying a sidearm daily. Collectively, these events destroyed my belief in the just rule of law and the goodness of humanity.
Overall, the above factors make me read articles like this with a mixture of excitement and possibility combined with depression and hopelessness. I do not have the resources for a side project, financially or emotionally. I cannot design and launch a side project, even though I am fully capable and that is what I most want to do.
I'm sorry to hear about your loss of faith in humanity. Especially hearing how you don't play music with others anymore. I'm a musician and not being able to make music with my friends would be a huge loss.
I hope you find peace and healing. People often suck, but they can also be beautiful if given the chance. And you deserve that chance.
Sounds like you've had a series of incredibly traumatic experiences. I'm sorry to hear that, and I hope that you're able to find your way into friendships and relationships that allow you to trust again.
> I no longer work on “hobby” projects; such things become an albatross around my neck. If you deploy a hobby project in your life, you must maintain it. Forever.
Alternatively: If you do something yourself, you get the option to be able to maintain it forever. If you buy something, the vendor will probably stop caring the moment the sale completes; any maintenance you want to do yourself becomes a reverse engineering effort as your work into a system that you didn't design.
> If you deploy a hobby project in your life, you must maintain it. Forever.
Perhaps your definition of hobby is different to mine, but hobby projects are exactly the kind you can abandon in a heartbeat. No long term customers, no guarantees, no obligations. In fact, we hear about countless HW and SW projects being abandoned left and right, right here on HN on regular basis.
> For that time cost, I want to work on a product that could go into production and pay for its own maintenance.
Hobby and high volume production aren't necessarily mutually exclusive.
I wanted to build a custom board to use USB-C power in my projects. Except the cost of a dozen boards assembled was quite high. So I made a thousand or so to subsidize the cost of development and production for myself. The chip I used went out stock (global chip shortage related) but until it did, my "hobby" project paid for itself and actually made me a small profit as well.
You don't mention where you live, but that last one suggests to me that you might consider moving to a different part of the country: I know several people, including myself, who carry sidearms. Where I live, everyone can do it, some people do do it, but no one gets upset by it or cares that much.
The author, Jay Carlson, is a really rad guy. I've got a few stereoscopic camera boards of his floating around that I happened to run into and a few emails later I had them up and running!
I just came across this site on Friday when looking for a possible replacement for some Digispark (ATTiny85 board) clones at work to connect to some 5v relay modules. I'm not too invested in the Arduino ecosystem at this point and am still looking into other options, haven't finished reading through the whole list yet.
I just used some attiny85 as a software based USB interface with v-usb for interfacing with an atmega328p. I especially love the DIP size, its amazing that i could solder the entire thing learning about capacitors and pull up resistors in the process. The use of multiple processors also allowed for some security by design, having functionalities capsuled with strictly defined directions of communication.
I think the ecosystem is really quite impressive with a really low entry hurdle from a teaching perspective. It was possible to get the power consumption of the atmega328p low enough for battery operation even with just Arduino, giving an easy introduction into prescalers and flags. From there you can step deeper into the functionality of avrdude (shoutout to USBASP) and standard AVR makefile projects with for example the micronucleus bootloader.
I should've specified that I don't dislike the ATTiny85 or Arduino, far from it. I just needed to find a replacement for a Digispark that I bricked (flashed the wrong bootloader) and wanted to see what other options were out there for a low-spec, low-cost MCU/dev board. I've been using Arduino for years and it's hard to beat the docs and community.
The mods here are really, really good. I do occasionally see spam & junk comments on HN, mostly early in the (US) morning, but they clean it up quickly. It's one of the reasons I like HN so much, it's very well curated.
There are stupendously powerful $1 microcontrollers now, including the RP2040 (dual core cortex m0+ with 264KB ram on the chip), ESP32-S2 (Espressif core with 400k of ram and wifi), and ESP32-C3 with wifi and RISC-V core and again something like 400k of ram.
Jay Carlson also has a series about microcontrollers as cheap as THREE CENTS, i.e. the famous Padauk 3 cent MCU:
https://jaycarlson.net/2019/09/06/whats-up-with-these-3-cent...