Hacker News new | past | comments | ask | show | jobs | submit login
Raspberry Pi Zero: the $5 computer (raspberrypi.org)
1271 points by MarcScott on Nov 26, 2015 | hide | past | web | favorite | 513 comments

This is seriously impressive.

While the Raspberry Pi is not the perfect hacker-friendly computer, it has done a lot of good. Some reasons off the top of my head:

1. Providing a low cost computer has given many people access to computers. Giving more people access to the web, email, an office suit, a programming environment AND giving people the ability of safely tinker without the fear of bricking an expensive device.

2. Introduced many different types of people to the FOSS landscape of powerful tools (e.g. distros such as Debian, tools such as Python).

3. The Raspberry Pi foundation has paid developers to write/optimise FOSS (e.g. paid Collabora to optimise WebkitGTK+ -- I think some Wayland work was also done).

4. Built on top of existing FOSS tools (e.g. building Raspbian on top of Debian), instead of doing everything on their own in a proprietary fashion. This has no doubt also helped to introduce new people into these communities.

This is a really good counter-point to all the "locked down" (hacker unfriendly) devices like smart phones and tablets.

This is probably the result of boards like the $9 C.H.I.P. stealing their thunder a little. (Which should actually be able to run Debian, unlike the Pi - Debian Jessie has built-in support for that generation of Allwinner chips.) The Pi has not actually been all that cheap lately compared to the competition. There's no reason they couldn't have done this years ago; like that Allwinner chip, the BCM2835 is an old, slow processor whose R&D costs were probably paid back long ago that they wouldn't otherwise be able to sell. This is especially true of the BCM2835, which is an ARM11 chip that's not widely supported outside the RPi ecosystem.

Olimex alleged[1] that the '$9' CHIP was being sold at a loss (which I guess is par for the course for a Kickstarter project).

>The Pi has not actually been all that cheap lately compared to the competition.

I'm not a big fan of the RPi Foundation & it's relationship with BroadCOMM, but which competition are you referring to? I tried searching for RPi-like boards a while back but only came up with HardKernel & the bananaPi, and I don't think either is a proper competitor.

1. https://olimex.wordpress.com/2015/06/05/how-to-get-in-the-ne...

The Orange Pi PC is competing fairly directly with the Pi 2, for $15. (Though it does apparently have some thermal throttling issues and isn't yet supported by the upstream Linux kernel without patches.) Annoyingly so in some ways; they haven't bothered to pin out many SoC features that the Pi lacks, like line-in support.

Oh, and that Olimex article smells of FUD. As they themselves point out, the price they were quoted for the SoC in the C.H.I.P. wouldn't be competitive - it's the same as a modern quad core SoC for a outdated single core. It's not a new die either, just another (possibly repackaged) version of the old and cheap Allwinner A13, so there's little in the way of R&D costs to recoup.

This could even act as a custom chromecast-ish device. Still cheaper too!

(Although it lacks WiFi AC that the new chromecast has, and probably way weaker graphics, but a cheap WiFi adapter or ethernet and having most stuff streamed should be workable.)

> (Although it lacks WiFi AC that the new chromecast has, and probably way weaker graphics, but a cheap WiFi adapter or ethernet and having most stuff streamed should be workable.)

This irks me a bit with the new "$5" RPi. Yes, it's $5… For a device without a power cord, without networking, and where you need adapters (and/or solder on a UART) to plug anything into it.

So... you will need to buy a power adapter and an Ethernet cable? That will come to, what, another fiver?

There's no ethernet port. So, also an ethernet adapter. But there's only one USB port (µUSB, too, so you most likely need an adapter to regular USB). So if you want more than just ethernet, you'll need an USB hub. But the RasPi's power supply has always been weak, so better get an actively powered one…

And soon you're paying four times as much for the accessories than for the RasPi.

This is an awesome board for embedded development. Not as powerful or I/O rich as the Beaglebone Black, but add $6 worth of CAN transceiver chips and a 12V->5V switching power supply and you've got an incredibly powerful automotive compute unit. The Pi 1 was a bit underpowered for what I'm doing, and this might hit the sweet spot while dropping $35 off my unit cost for small quantities. None of the issues you mention affect that calculation at all.

Who uses ethernet anymore?

That's exactly what I'm going to use this for, too. Right now I have an original Pi B logging GPS and OBD data, but it's crammed into a small storage spot in the car and is a big waste of space. I'm super excited to be able to cut my project to less than half its original size. I don't need ethernet (other than for debugging), four usb ports, etc. etc. I think this is going to be fantastic for creating homemade embedded devices.

Do you have or did you follow a write up for this? I know I would be interested in reading how you set that up!

I don't right now, unfortunately, but a writeup is next on my list of projects after the carputer is complete. I'll post it here if there's interest. For learning about the OBD protocol, though, these resources have been very helpful:

The ELM 327 data sheet (this is the chip in most OBD devices you'll find on amazon): http://elmelectronics.com/DSheets/ELM327DS.pdf

Wikipedia's article on OBD PIDs: https://en.wikipedia.org/wiki/OBD-II_PIDs

"A complete guide to hacking your vehicle bus on the cheap & easy": https://theksmith.com/software/hack-vehicle-bus-cheap-easy-p...

There's a PyOBD python module out there (my carputer is built in python), but it's a fairly simple process so I just rolled by own using pyserial.

How exactly has the power supply been weak? I've been running three PIs (A+, B and B+) for a few years with no problems with just simple cell phone power adapters. However, I'm not plugging any extra hardware into them...perhaps that's why I haven't run into problems?

Yes, the parent was talking about lacking power through the Pi's USB ports thus often requiring a powered USB hub.

My issue is with the Pi's use of micro USB for power, I'd rather it use a barrel connector, it's not like it doesn't have the space. Yes, I know the choice is based on the presumption that everyone already has a micro USB power adapter from a cell phone. The use of micro USB for power on this Pi Zero makes a lot more sense given the form factor.

> However, I'm not plugging any extra hardware into them...perhaps that's why I haven't run into problems?

Yes. Phone chargers provide plenty of power for the Pi itself, but once you add several USB devices like wifi/ethernet you'll need beefy ones.

But even that is normally not much of a problem, because (especially on the B+) each device has their own port. If you plug several into a hub, however, you'll quickly exceed the hub port's power rating. Hence the need for an active hub.

USB-OTG cables with power supply exist.

Ethernet USB adapters tend to be very flakey. (At least that's my opinion/experience.)

It was my understanding that the ethernet on the Raspberry Pi sat on the USB anyway.

Not just an ethernet cable, I think. From the images it looks like there's no Ethernet port on the RPi Zero.

Would be neat with an ethernet adapter with Power-over-Ethernet plus a simple micro-USB converter/USB-OTG cable. Yes, I'm aware that might raise the price a bit...

Other than hack value, I don't see it as a replacement for a $35 Chromecast. Even then probably better to root a Chromecast due to all the hardware benefits mentioned above.

How rootable are they, and will that break updates?

"1. Providing a low cost computer has given many people access to computers. Giving more people access to the web, email, an office suit, a programming environment AND giving people the ability of safely tinker without the fear of bricking an expensive device."

I seriously doubt about your first point. Adding the cost of accessories (mouse, keyboard, screen) and a shipping cost, Raspberry Pi costs the same as a decent, used laptop that can be used to install a full Linux distro. Even getting Raspberry Pi Zero with essential kit (cabels + the smallest SD card) to my country rises its costs to 20£, add the cost of accessories, and you can get another working used laptop (and as a bonus Ethernet/WiFi connection which RPi Zero lacks). Don't get me wrong, Raspberry Pi is great for tinkering and home projects, I have much respect for what they are doing, but getting "computer to every home" is not it's strongest point.

It seems awfully silly to argue that it costs the same once you add in the price of a retail monitor. You're just saying that the price of the computer is negligible to the setup. They don't manufacture the peripherals, it's weird to blame them for not magically also dropping the price of those other things that they don't make or sell. Besides, it's pretty easy to get free mouse/keyboard/screen via donations from companies or individuals who upgrade if you're that worried about cost.

To add a fifth point if I may - it was also rocket fuel for getting CS into the (what most here will call 'K12') curriculum.

When I (only ~5 years ago) took GCSE (last compulsory qualification) 'IT' - even though it was the slightly different course for 'more able' pupils - we were learning how to make Flash animations, use MS Office, Windows Movie Maker, and sorry things like that.

Forget "fifth point" - I think it's impact on schools is #1.

Pre-GCSE was even worse. Taking screenshots of each step of copying a folder to show that we knew how to do it.

Are you sure about 1? Where are the maths please? In my little circle, the only people who have raspberry already have awesome computers and co.

Curious about that.

How many people are there in your circle who want a computer but don't have one? Perhaps your crowd aren't the target for point 1.

Oh, yep sure! And the visible facts are?

(really, no needs to downvote me. I would love to see this thruth, just sceptical)

Here are some articles in this subject: https://www.raspberrypi.org/blog/tag/developing-world/

Not saying it proves or disproves anything. I'm also interested in this aspect of cheap-o-computing

Personally, I am very much in need of an office suit.

Definitely an impressive feat. I really, really, really wish it had on-board Wi-Fi though. At least that would make it a useful web server.

The way it's designed now, it's kind of unusable for anything useful without making it a 3X larger package. Add a Wi-Fi dongle and the converter from regular to micro-USB is about the size of the board itself. And since it only has 1 USB port, you'll need a full-blown hub to connect any other stuff like sensors, keyboards, or whatnot.

For one I'm glad it doesn't have wifi since that'd add on size and cost. I hope that the ecosystem takes off and shields get created like wifi, bluetooth, gsm, etc. That would be really useful and give everyone flexibility.

From the differences amongst the Arduino, I see the hdmi as a necessary evil. Otherwise it'd be too cumbersome to debug.

I totally agree. One of the reasons I've started using an Onion Omega is because of this reason. IoT is pointless if you can't connect to the I easily, IMHO.

Yeah, I played with an Onion Omega for a while. The downside to it is the horribly limiting 16MB flash (with 64MB RAM, WTF?). When I first read it, I thought it was 16GB and a typo. (We have 128GB micro-SD cards already, and all they could put was 16MB onto it?)

That's where the Pi excels since it has a micro-SD slot (and a much faster CPU of course). But then, no Wi-Fi ...

Not all flash is equal. If it's 16MB of fast NOR flash for direct code execution then it's not so bad.

> I think some Wayland work was also done

Done then abandoned. Doesn't work with their currently shipping hardware/OS.

This is so cool, it's even cheaper than an Arduino. This could turn into a standard embedded development, OS development and learning board for every enthusiast and students.

The cost makes it a great choice for students with less resources.

This is easily my pocket pc now.

Oh Nice! They are finally delivering on what I feel was the great promise of the Raspi in the beginning. Full linux install in the size and price of a micro (read: Arduino).

Some people are bound to gripe about the "lack" of ports but its not like this one displaced the A or B models. Its just another spin of the concept where you don't have to pay for expensive physical parts you don't need. Its a linux server at a price cheap enough to buy one for each little project you want to do and then leave it there. Makers rejoice.

I've got to hand it to the Raspi folks. They've really done an outstanding job creating their product line and getting it out to the masses. When they started, there was nothing but a sea of vaporware and "next-year" promises in the inexpensive SBC linux world. I rather famously doubted them at first. I am very happy to have been wrong.

Buy magazine, get a free computer. It's a pretty good time to be alive.

There is a great new USB standard for this form factor:


It needs to reach a tipping point so the price drops but we're probably only a year away.

I have no idea why Arduino is so popular--it has NO debugging support unless you use an ICE. (People "debug" with printf statements.) We use Netduinos because with .NET we can single step and inspect variables in live code.

Why is Arduino so popular? Brand name recognition, a decent standard library, and a ton of sample code and tutorials online.

I agree that it's not a great platform, though. I personally prefer STM32 these days; it's dirt cheap (a basic development board is <$5) and powerful (72 MHz / 20 KB RAM / 128 KB flash on one of those boards). I believe there's even a way to integrate it into the Arduino IDE nowadays!

Arduino is popular because it was the first one that got big and built up a community. It's the same reason the RPi is so popular despite not being especially compelling from a pure spec standpoint.

You can get more performance for less money with other boards, but you'll have to figure out a lot more stuff on your own.

IME Arduino is so popular because it's so cheap. I can buy Arduino ProMicros on Alibaba cheaper than I can buy the Mega168 MCU itself. And that's not counting the PCB, additional chips or board.

I've got three Raspberry Pis. I don't know why. I don't know what to do with them. None of the projects that I've seen have been particularly compelling to me, and I'm not creative enough to come up with a good idea. So they sit on my desk, gathering dust, and act as a conversation piece when friends come over. "That? Yeah, I can build X, Y, or Z." and they say, "oooh, cool" and then I never actually follow through. I'm a terrible nerd. :(

Just as a counterpoint, I have found many, many uses for Pis in my life. One is hosting my home automation software, I have built several information radiators with it, one is acting as a network-controlled media player, I even made one that acts as a faux remote control for the AC with IR diodes attached to the GPIO ports.

By contrast, I have several Arduino boards sitting here, and never found a good use for them. For me the appeal of the Pi lies in the combination of low energy usage, low price, networking, and comparatively high computing power. Once you get over the "woah, wasting an entire computer just for [insert menial task]" barrier, it's simply incredible what this single component enables you to cook up on a whim. The Pi allows me to make things that are not yet (or will never be) available commercially.

However, I would not say this thing is necessarily useful for all hackers or nerds. It's only useful if you have ideas for small, self-built devices that require programmable intelligence. Take note that this is also fundamentally different from the stated purpose of the Pi, as stated by its manufacturer.

I have a Raspberry Pi 2 and would like to buy a digital scale, and somehow tweak it so that it can sync my weight with my myfitnesspal account anytime I weigh myself, how do you suggest I get started with this? I don't have much of an electronics background.

Why not use bluetooth and a Wii balance board? eg. https://github.com/damariei/wiiscale

I have been searching for the same thing, ideally I'd like to buy an electronic scale with a digital I/O port. You may have more luck finding these in the US than I had in the EU, but in any case I wasn't able to find anything that hooks up to a computer in a non-proprietary way. As with home automation, all products seem to sync to the cloud, most of them in a one-way fashion.

I lost track of this idea a few years ago, so the situation may have changed. But if it hasn't, I see the following options for you:

1) find a cloud-synced scale that offers an SDK or at least some way of retrieving the data from the cloud.

2) buy a low-cost digital scale, take it apart and see what you can do with a voltmeter and a soldering iron. At the very least, you should be able to hook into the raw analog weight data which you could feed to a Pi with an A/D board. The nicer solution would be to siphon off the digital value. Even in scales where one chip does everything you could still try to pick up data going to the LCD. However, that might require some epic cabling and soldering, not sure it's worth it.

3) cheesing it: hook up a networked camera and point it to the scale's display. You can do this with any old android phone and a remote webcam app. Have the software on the Pi (or some other computer) pull the image when you step on the scale, run OCR on it, write the value out to wherever you want.

What about option 2.5? Rather than hack the hardware, hack the software; buy a cloud-synced scale and sniff the traffic it sends out with the pi.

Wouldn't that be option 1.5? ;)

Yes, in principle. If they are not obfuscating and/or encrypting their data, which is a big if. And you'd be bound to keeping that packet sniffer online forever, piping matching data to where you need it.


I had a conversation the other day about side projects. People would ask me what I did for side projects, since I have experience in software, electrical engineering, control systems, mechanical assemblies, etc.

My answer was generally that at my day job I got to write software to control electromechanical systems with a budget in the millions. What am I going to do on the side to compete with that?

Sound familiar? I'm guessing that a lot of your "I don't know what to do with it" is because even a cheap laptop, desktop, or tablet is so much easier to work with and has pretty much greater capability than the RPi that the latter just sits and gathers dust.

Oh, and if it helps, in addition to my RPi gathering dust, I also have a BeagleBone and about a dozen flavors of Arduino :-) At least I do some consulting work with Arduino code, so they get used a bit.

It used to be called a "hobby", but now it's a "side project"?

The pi was an odd thing. Out of the box it was 'suited' for GPIO from python. For any better project you'd need to know electricity. Or going below and learn the SoC weird architecture...

But recently I used one as an IRC bouncer, and another as a XBMC host.

Before that I just wanted to see how arch+dwm+emacs would feel on a 700MHz prev-gen. ARM CPU. Acceptable. I tried to underclock it down to my first desktop computer freq (p75), I've never seen linux run on such slow freq.

I also enjoyed the pre-compiled cross compiler so I could write factorial on my x64 machine and then deploy it on my rpi cluster of 2. None of this is amazing and no hard requirement on a pi (I could have done the same in VMs or with MIPS emulators but ... it would be more setup and less toying).

Impulse buy for most.

ps: One more major downer for me, hardware compatibility, too frail. SDCard would fail, usb would fail, first psu would not push enough amps ... It made you realize how the ATX desktops shielded you. And now I was at the point where I couldn't be sure of anything without researching wikis for hours. Again electricity.

Build them out as dummy media boxes and sell them for a markup. I've had people knocking my door down after they saw my Kodi setup but I simply pointed them to the how-to guides and told them to learn for themselves. I'm sure there's a few people out there who'd take one for $50 or so.

Probably. But at $50 each you're not really making a profit.

There is a "Dragon Box" on Amazon that sells for $270ish and appears to be a lower-powered Android Kodi box. I imagine that a second generation Pi could perform just as well.

chinese seem to do just fine at that price point


I set mine up with a camera to do timelapse.


This is cool, did you follow a guide to set it up, or just wing it?

Mine is collecting dust, meanwhile I have a list I wrote while waiting for it to come in the mail of all the things I was going to try on it:

* ZNC for IRC connectivity * connected to backup server for incremental backups * connected to backup server for bittorrent downloads * media server streams bittorrent downloads to macbook * test web server * git server * owncloud or ftp server for sharing files * controls light switch via http->gpio * nes emulator through hdmi to tv with usb nes controller * ddclient for dynamic DNS of c1.jrcii.com on afraid.org * Run printer jobs/server * snort + snorby * web SDR * WPA2 cracker * in-room temp humidity to menubar and blinds * pet feeder/waterer * http://www.youtube.com/watch?v=PXmCiaRc9XU * piphone

Familiar stuff... at the end of the day, I just end up doing the software side on my macbook. Like running a server, or some old console emulator. Media boxes are useful, but it's a bit like buying cakemix and making a cake, it's so trivial at some point you stop caring about the process and just buy a cake.

As for hardware projects... I've never had an idea for one I'd actually want or care for. Things like a lightbulb that flashes when you get a tweet, and that reads it out loud when you wave your hand infront of it... pet feeders... etc I can do without. Would be fun to build but I'm so busy these days I can only motivate myself to make time if I like the end product I'm building, too.

I hear ya. One of my Raspberry Pi's sat around for a year or two. Then I got this idea to connect my sheet-fed scanner to it and scan pages directly to Dropbox. That worked nicely. My 2nd RPi I got 2 Canon Selphy Printers and setup a gmail account where I could send pictures to. The RPi picks up the images and send them to the printers in a round-robbin fashion. ... I guess what I'm saying is the sky is the limit... So how do you follow-through? The idea needs to have some practical purpose. In my case, I use both those setups regularly.

What brand of sheet feed scanner?

I used my first Raspi to build one of these - got me interested in other music projects like DACs etc.


The real fact is the Raspberry Pi are massive in the IoT arena. Just because Joe Average tinkerer doesnt know what to do with them this size reduction is a big deal for industry, Easy money.

I suppose it's good these sold out so fast. I probably would've bought six or seven, used one, and left the rest to collect dust.

I hear ya, I've been "lucky" and got my order in on every model of Pi on the day of release - even the original. I went through the same process of "ooh, shiny thing!" when I saw this announcement, then thought back to my other RasPi's sitting on my desk doing nothing and decided to leave it for someone who might actually get some use out of it. I'll be getting a Zero when they're more generally available though, if only to complete the collection!

That said, they have been used in the past for various things - a simple SSH server, a networked webcam, a serial interface for my Atari ST, and I've even done some tinkering with a breadboard and the GPIO, so I have had some use out of them at least.

I'm new to tinkering and discovered you can build an SNES emulator out of a Pi, so that's one Xmas gift right there! I have 2 more Pis on the way. Not sure what I'm going to do with those yet, but maybe a NAS for my Dad?

why not build a htpc or an arcade box? very straight forward and you'll have more to talk about when your friends come over!

Can you actually build an HTPC? My understanding is they can't really handle video, except from the purpose-built camera module.

That's exactly wrong. They have the same processors as used in phones, which means they have hardware level decoding for video. My B+ can stream massive 1080p blu-ray rips over ethernet with zero problems. They make fantastic HTPCs.

My phone doesn't ever play 12GB blu-ray rips; and the rpi hardware is older than current phones (which use armv7 and armv8). Given the speed of the cpu, it's surprising it can decode a full H264 stream.

But It's good to know. The down-vote seems unnecessary.

It's not so much software running on the CPU doing the work here. It's hardware support for decoding via the GPU that enables it.

I've been running R-Pi based HTPC since the original R-Pi release. I current run a distro called OSMC (debian based, with Kodi on top) on my R-Pi 2 and it plays everything I throw at it with out an issue.


Been running OpenElec on mine for years now. Works great and does 1080p output without transcoding or anything like that.

There are several HTPC implementations (eg. RaspBMC). They're all pretty nice, but personally, I just use omxplayer, which is the command line package that powers RaspBMC under the hood. It's a command-line-only media player, but it plays anything I can throw at it, including live-streaming 1080p YouTube videos.

The command-line interface for omxplayer is a bit verbose, so I wrote a basic wrapper in BASH, so all I need to do is type "play <url>" and the video will begin streaming after a 10-second buffer. I keep a Screen session open on the Pi so I can drop in and command it from anywhere. At any given point, that little Pi can be simultaneously playing 1080p video, while downloading another two videos in the background. Incredibly capable little devices!

Nice idea. 70s era arcade games running on a processor with more MFLOPS than an early Cray at ~1/1000000 the price.

This seems like an important moment for humans and computers. The cable to connect the computer now costs more than a 1GHZ computer.

It feels a lot like a tipping point for something.


The cable to connect the computer now has more markup than a 1 GHz computer. Micro-USB OTG and mini HDMI adapters go for £1 each shipped on eBay, and I think most of that price is shipping. (Likewise, they've omitted the 40-pin GPIO header unless you buy the £10 bundle with the "Essentials kit", compared to £4 for the Pi itself.)

Old, old business model - cut the headline price to the bone, make it back on cables.

I can't really see that — neither the circuit board header nor the cables are proprietary.

Doesn't have to be - they're relying on people not shopping around for the cables and accessories, much like big brick-and-mortar retailers such as Best Buy have done for a long time.

Arduino Uno for $2, NodeMCU for $3, RPi for $5... please let this trend continue. A year or two, and we'll basically have computronium - simple dev boards powerful and cheap enough to just tile your house with them.

I've recently figured out that there's no point in designing your own PCB for placing sensors at home, when you can get an Arduino and an ESP8266 for $5; add power (and some ~$0 of voltage regulation) and you have a base station. Or just buy NodeMCU for $3 and skip on wiring Arduino and ESP8266 together.

Even more awesome: Skip the extra Arduino and use the ESP8266 as your Arduino.


I realise how silly this question may seem but here we go:

If I get an ESP8266, I can have something that run code, sit on my home network wirelessly and use it to do things like report on sensor inputs? Am I missing some larger expense that would be required as this seems very cheap compared to other options I've seen.

The cost is $4 for a NodeMCU, $2 for the sensors, $1 for a breadboard to stick them on, and $1 for a microUSB cable to power them from. You can power the sensors from the NodeMCU's Vin, since that's the 5V from the USB.

I have done exactly this, it sends values to an MQTT server on my NAS every second or so, which the NAS then adds to Graphite and plots them in Grafana. It also has some logic to act on these values and send commands back, like turning lights on and off, etc.

All that terminology at a really low cost, awesome.

I feel like this comment could be turned into a popular blog post.

I can write up the sensor setup I have at home, if you want. It has motion, temperature, humidity, light sensors and IR/RF remotes for controlling devices, and a burglar alarm.

Please do write about it. Personally, I'll be happy to steal as many ideas as I can for my upcoming home installation :).

Great, I'll give it a shot tomorrow. I also have a mini-ecosystem of Python scripts to communicate with the sensors over serial and to read values, etc. I'll post that as well.

Great! :). Will be refreshing /new/ tomorrow.

Hmm, it looks like I open-sourced most of it a while ago (way to go, past me!):


I'll write up the details, but this is a good chunk of it.

I JUST started this exact functionality to communicate with a wireless scale, thanks for sharing!

You're welcome, I'm working on some NodeMCU-based designs now, with IR as well.

So, where's the Show HN? ;).

Man, I spent all day yesterday trying to reimplement the sensor for the ESP8266, as I thought it would be better to have it report over MQTT, but the damn thing keeps crashing in seconds. I literally cannot keep it working for more than a few seconds, no matter what I do.

I'm pretty disheartened, I'll try some more and either succeed or go back to an Arduino over serial... Do you have any experience with making it not crash?

Sadly, my ESP8266 experience is only about to start next week, I'm still abroad. But from the experiences of some of my friends, consider replacing the firmware on the ESP - some units were apparently sold with buggy one that made the module crash after short time.

Good luck!

Yeah, I've tried flashing the Arduino firmware (I think it gets flashed every time regardless). I'll keep trying and let you know, thanks!

By the way, if you don't want to keep checking the comments, you can follow me on Twitter at @stavros and experience my frustration first-hand. I'll be posting about it there.

I'm using HN Notify, so I generally get replies straight to my inbox (except some weird bug once that made the service not work for me for like a month or two). But yeah, maybe Twitter is a better place to have real-time updates. Will be following. Thanks :).

I just got it to run, yay! Things are going well, and MQTT functionality has been working well for the past hour. Just checking in :P

Copy that!

Perfect, thanks, I'll have a look at the NodeMCU, looks like exactly what I'm after.

The nodemcu is definitely the sweet spot. It saves you the trouble of power management, is breadboard friendly, and integrates the programmer. It's way easier to work with than the base ESP-01.

Simply put ESP8266 & Co. are awesomeness incorporated.


Brilliant, I'll have a look through those. Thanks for the link.

The only other part needed to get this running is a programming cable. The ESP8266 does not have a USB port but a serial one, so you need an adapter to connect and programm it. e.g.: https://www.sparkfun.com/products/12977

Just be careful not to connect the 5V VCC to the ESP8266, it will probably kill it.

That's fantastic, thanks :)

The SparkFun ESP8266 Thing Dev Board [0] has a USB-to-serial so you can use a micro USB cable. They're obviously more expensive but ESP8266 breakout boards like SparkFun's Things and Adafruit's Huzzah are more breadboard-friendly an generally easier to work with.

[0] https://www.sparkfun.com/products/13711

Yes. At least for limited values of yes.

You need to run it on 3.3v which means 3.3v versions of your sensore or external level shifting. You've got limited IO compared to, say, Arduino.

But of you've got 3.3v sensors - yeah - all you need is an ESP8266 and some way of providing 3.3v. (And probably some pulldown/up resistors and decoupling caps...)

I'm pretty sure the digital I/O on the ESP8266 is actually 5V tolerant according to the datasheet, though this isn't widely advertised and there's some inaccurate information drifting around from back when the specs weren't fully known. (The same isn't true of some official Arduino boards which should not be exposed to 5V on the I/O pins.) It's a very nice little chip.

to makomk: No first-hand experience, but entotheteeth on reddit, says that 5v data on the I/O will kill your device in this thread: https://www.reddit.com/r/esp8266/comments/3bm061/5v_board_wi...

Fantastic, thanks for the info :)

Good one, thanks!

While I agree completely with what you're saying, I think it's almost unfair to be comparing this to an Arduino or NodeMCU - this is orders of magnitude more powerful.

At this price point, for many projects it doesn't matter; you can now use Raspberry Pi as a throwaway SOC for powering little stuff. I wouldn't, for example, put an $35 RPi 2 in charge of a bunch of sensors over my doors, but with $5 RPi it is an option.


And given the power, you can start offloading computation on what would otherwise be "dumb" control/data acquisition units. Hell, with that much compute you can start running serious DSP on it, which opens some additional use cases.

The only case where it matters for me is that this runs Linux, which means that I'll have an extra computer to secure against intruders. An Arduino can't do as much damage to the rest of the network if it gets hacked.

Linux also isn't a real-time OS which is often important when dealing with motors and sensors.

However, you can get very far with the built-in real-time capabilities. If your software stack can deal with occasional jitter in sensor updates (and normally you have to take that into account), you can compile a stock kernel with PREEMPT_RT and can use C to build software that has to do IO every 1ms. In POSIX. That is huge.

What would be a real-time OS? I'd be interested in hooking up a bunch of independent boards that could interact with each other, and have as quick as a response to sensors as possible... Do you have any advice as to where to start with this type of project?

Sensor latency will be measured in milliseconds in Linux. A real time OS is more necessary when you have feedback loops, inverted pendelum problems, bitbanging protocols, etc. Not being a real time OS just means that when you say "ping me in X time" it might ping you in X +- Y. Y tends to be too big for critical applications, but generally small enough that it won't matter.

Don't plug an ethernet adapter or wifi dongle into your Raspberry Pi Zero if that's your problem.

How would it communicate?

It could run RiscOS instead, with a very different attack surface?

As a programmer who used to like taking things apart and playing with solder in school seeing these kinds of things fill me with enthusiasm and joy - and then I remember I need a wired power supply or a battery and suddenly enthusiasm is gone :( I wish we had wireless power transfer - for these kinds of things losses wouldn't even matter.

Yeah, powering stuff sucks. With ESP8266 you've pretty much eliminated the need for a data line for home automation, but you still need a power line.

My current solution (to be deployed in a week or two, after I get home): $2 Chinese fast charger with 4 USB ports (2x 2A, 2x 1A). I had a chance to test them personally before buying; they're decent enough for the task and cheap enough to buy a bunch. You can then use either an Arduino ($2-3 with USB cable) and ESP8266, or NodeMCU + $3 USB cable.

Got a link to the fast charger?

It's this model:


The price on Aliexpress is stunningly high though; I bought it for something like $2-$3 in person. I guess this is one of those cases (I've noticed a few), where Chinese just leave the same price numeric value but change the unit from CNY to USD.

You can even flash Espurino (JavaScript runtime) on ESP8266/NodeMCU boards. It's a crazy world!

> Arduino Uno for $2

Are you sure about this? It doesn't sound... legal.

I meant Chinese clones. Arduino is Open Hardware, it's perfectly legal (that's kind of a point of Open Hardware movement). It's equivalent in form and function to the genuine Arduino made in Italy.

Oh okay okay, I see now. Yes, it's perfectly legal (as long as they don't use the Arduino (or Genuino) name).

Do you have a link? I was just about to get started on a project and getting a $2 Arduino would be great for tinkering

Just a little bit more: http://www.aliexpress.com/item/Best-prices-UNO-R3-MEGA328P-f...

But if you are not planning on using shields you can get arduino mini for ~$1.3 or nano for ~$1.8

Go to AliExpress.com and search for "pro micro" or "pro mini".

You can get a Pro Mini clone for less than $1.50.

A full-size Arduino R3 clone (with a cheaper USB chip, the CH340G) is less than $3 with free shipping.

I'm speculating, rather sooner than later, the foundation will let people (read Chinese) clone Pi Zero at will ala Arduino way.

They can't even have a profit margin with that price point. So why not open-source it anyway to let go of the burden of manufacturing? That would be a great boost for the Raspberry platform.

raspberry avoid this by using chips only supplied to them.

Where are those chips actually made? ;).

What does it matter? broadcom wont sell you them. and you cant legally import them yourself.

Just because Broadcom won't sell them doesn't mean the factory won't make some more. This is China, they have a pretty liberal (and one could argue: more reasonable) approach to issues of intellectual property :).

Or at least it used to be so; sadly, in the last few years the US was pressuring Chinese government to crack down on it. Anyway, you can read more about this phenomenon here: [0].

Oh, and one more thing. It may not apply to RPi in terms of cloning it because it's still quite a niche product, but in general, they actually recycle electronics here. Like, seriously, when you throw away your smartphone, it'll likely end up in China for desoldering. So will your laptop. The amount of desoldered components and especially ICs available for sale is staggering. I have no data for this, but I'm pretty sure people here are making small-scale production runs out of recycled ICs. So Broadcom may not sell you their chip, but the lady on the first floor of SEG building in Shenzhen just might.

BTW., I love how they recycle components so much, and how it leads to interesting situations sometimes. For instance, I recently got my phone fixed - $15 and 15 minutes of waiting was all it took to get the front half of Galaxy S4 replaced with a brand new part (screen + touch surface + glass + box) and to get a new camera (electronics, lens and all). The equivalent repair at home would cost me around $240 and would take a few days. The secret of such cheap and fast repair? They took the broken parts from me. They'll regain 90% of value of that repair by simply replacing broken subcomponents in their own time, and then pushing the fixed modules back onto the market.

[0] - http://www.bunniestudios.com/blog/?cat=20

"So Broadcom may not sell you their chip, but the lady on the first floor of SEG building in Shenzhen just might."

And she will also happily affix a holographic "genuine" sticker to the part and package it in "original" "tamper proof" packaging. Just like brand new!

I've seen the stickers today. You can get quite a lot of them for less than a dollar :).

or you get the broadcom+ chip. which kindly breaches your entire network and sends the content of everything it finds to a Chinese ip address...

don't have any data on that either. but I'm guessing it is more likely than getting what you ordered.

Love watching the progress of Raspberry Pi. The attraction of the platform to me (compared to something more capable, with more ram ghz etc) is that the ecosystem is now so mature, you can easily google and get a cut and paste guide to exactly what you want to do. This means you can spend time tinkering with aspects that interest you rather than maintaining an operating system or troubleshooting hardware...

I'm curious to see how far optimisations could go. Analagous to the old consoles where developers could squeeze incredible performance out (compared to the equivalent processors elsewhere) since it was so uniform. e.g. the later games on a nintendo or neo geo were incredible compared to what was capable on a typical 8/16 bit computer of the time.

A $5 version is just going to accelerate this ecosystem... looking forward to it.

There is a lot of capability on the graphics side, which takes up most of the die. Can run Quake3 etc. Not very impressive compared to modern PC graphics, but very handy if you want to put something on a small screen.

Same with Arduino. You can use dozens of off-the-shelf shields and libraries to make it do exactly what you need without having to understand every little detail (however, if you want you can change every little bit or resistor!). Instead you can focus on solving the problem at hand, mostly writing high level glue code.

I would love to see someone sell these as preconfigured as minimal bandwidth tor exit nodes with wifi so they can be spread far and wide. Just connect to power, make sure it is connected to a wifi network and then leave it alone. Just make sure it advertises itself as such so that law enforcement knows that it is likely not owned by the owner of the Internet connection and therefore doesn't make sense to do anything other than find the device and disable it if that's what they want. It should have a similar disclaimer as the standard tor exit not notice, except it should say something like "this tor exit node is operating on a disposable computer and was placed clandestinely on this network without the consent of the owner" or something to that effect.

I know there are ethical implications here, but that doesn't mean that something like this shouldn't or won't exist eventually.

These are not well suited for Tor. No built in networking and a serious lack of CPU power needed for processing encryption.

How much CPU power do you need to process encryption in TOR?

No wifi, no Ethernet. You need to use a USB adapter.

Most wifi networks are NATed, so this would only work if you could also reconfigure the router to forward the necessary ports to the device.

You could use it more easily for hosting hidden services though, as those only require an outbound connection to be made to the Tor network. Hidden services on a hidden device.

upnp is a thing at least some of the time.

There are also economical implications. Bandwidth isn't free. Now if you also could charge for your service with micropayments, we have a winner.

I believe the implication is that these would be placed in locations that have free wifi available, without necessarily having permission to do so.

McDonalds would be perfect

The 80s are widely considered as the golden age of hacking, but what the hell? I just bought ten microcontrollers that can run Lua/Python/C for $20, there are full-blown computers for $5, and all the supporting ecosystem (sensors, components, etc) is cheaper than most toys.

I am very optimistic about the future, given that people (and children) these days have trivially cheap access to powerful programmable and easily connectable computers, and hopefully they'll start to demand more and more that all their other devices are equally hackable. If most people have a microcontroller at home that they made themselves that controls the coffee maker, they will want to be able to connect other stuff around the house up, and that can only be done with open protocols.

The next few years are going to be very interesting on the maker scene.

Plus standardisation (linux, usb, arm, etc) which I understand was one thing lacking in the 80s.

The standard was pretty simple in the 80's, > 3 V = 1, anything below that counts as a 0.

Lacking because we were building it.

Amazing. If only I'd get my act together and do all the amazing stuff I planned when I bought my first pi - then I could justify buying this

Two years after buying mine, I put the pi to a good use only last week! Now I am using it as ad-blocker thanks to http://pi-hole.net/, works like a charm.

I ported the web interface for pi-hole to Go single executable (https://github.com/girishso/pi-hole-web)

Now looking for something interesting to do on my second pi!

Edit: Added pi-hole-web link.

That's great, thanks!!

I have a pi as a media server* connected to a projector via HDMI, that reads media files from a NAS on my home network; very efficient and lightweight.

I will definitely try pi-hole; maybe I can have both running on the same pi.

* via Raspbmc, which I just found out has apparently ceased to exist...

> * via Raspbmc, which I just found out has apparently ceased to exist...

Try OSMC [1], it's a massive improvement over Raspbmc and is being developed actively. Works like a charm. If I'm correct, the main Raspbmc dev went over to OSMC a while back.

[1] https://osmc.tv

Yes, thanks, but for now raspbmc works really well, I'm not sure I want to go through the whole update now unless there's a big compelling reason.

I might upgrade to a 2B though, it's probably a lot faster (the current setup tends to freeze a little on rapidly changing frames).

You have to overclock to about 900 to get the resposiveness up.

Both of mine are perfectly stable at that speed.

OSMC is actually the successor of Raspbmc (And Crystalbuntu)


My frequently asked question: if I want to watch and record DVB-T television, what media centre solution should I be using?

I have been running the Pi since 2012 from the early days of Raspbmc to OSMC and sometime in 2013 starting watching and recording television using it.

I have ATSC (North American Over-The-Air TV) which is similar to DVB-T. I use an HDHomerun Network TV tuner. I watch using my Pi running OSMC and an Add-on for DVBLink, and when I select record, I use DVBLink software which runs on my Synology NAS. DVBLink however is paid software that can run on the Pi, as well as many other platforms (Windows, Linux, Asustor, Netgear, QNAP, Synology, Western Digital).

I tried using TVHeadend but found the interface and support of DVBLink to be better.

My other Pi currently runs Volumio (https://volumio.org) for music streaming.

How does it do performance wise? I can only imagine that DNS + Blocklist checking is already a big task for a PI. I always wanted to have something like this, and am on the search for something like a PI but a bit better since then.

It works pretty well, in fact page loading times have significantly reduced for me, may be because it caches the DNS records locally. I think PI is good enough for this purpose.

How much RAM does it take? I mean if it runs with less than 250 MB of free RAM I can deploy this right away, without using a dedicated RPi (I have 2 RPis on my network, the one is doing some heavy lifting, the other one has many cycles to spare).

Not much, currently it's using 1.6% RAM of 512MB. Even CPU usage seems negligible. After all, it's just acting as a DNS server, only resolving the DNS queries.. not to mention blocking the ad domains!

I just love the idea.I may finally put my Pi for some good use. I am hoping enough people have looked at it to trust this with my network.

What is the difference between this and privoxy with an ad blocker?

looks like a great idea but doesn't it significantly slow down your network? I think I would rather install an ad-blocker on a tomato/dd-wrt/open-wrt router...

Same here. Had them laying around for two years before I dusted them off to port Processing to it (shameless plug: https://github.com/processing/processing/wiki/Raspberry-Pi).

Still haven't done anything truly original, but I learned a lot about the hardware and the kernel writing the scaffolding and plumbing - and hoping that this will make the curve less steep for the ones that do.

Lol...I'm the same. I have four of the sodding things (Pi#1-B, revised Pi#1-B and two of the model 2 B's) complete with PSU's...all sitting in a box. So many plans, so little time.

I wonder if the problem for most of us isn't the money, but just the time. A cheap computer probably doesn't give us any more time/ease than an expensive computer (which we probably already have 1+ of)

Find a local makerspace and pop in every once in a while. A bit of time spent with others doing things rubs off.

I live in rural Scotland, there is no such thing, and sadly I don't have the time available to start one.

Amusing that most people I speak to (including myself) fall into this category.

Great deal nonetheless, especially when considering the bit where he mentioned being much faster that version 1, which is considerably slow if you ask me.

They're cheap enough to be a "cool, I have to play with this" impulse buy...

At last count I've got about 7-8 different devices like this sitting around that I've aspired to play around with at one time or other, all of them have seen way too little use.

I know what you mean.

I have a Tessel here (Nodejs base micro controller) and never used it...

Didn't know about Tessel, the plug and play modules look very tempting!

Yes, everything works out of the box with npm install. It's really nice.

But I have no idea what I should do with it, haha

This is great, but for people who want to "get started in computing", don't they also need a display, a mouse and a keyboard?

LCD screens are like, $100, a mouse is $10-$20 new and low-end keyboards are $10-$30. At that combined price, why does it matter if the computer is $25 or $5?

And if one is going to outlay the 150 bucks in peripherals, they might as well spend a bit more than $5 on the computer to get a significantly better computing experience.

Is there anything else going on here? Do they have a different approach that I'm not getting? It feels like, yeah, Moore's law is great and computers are cheap, but once RPi got to around or <$50 (which it did with the first version anyway), the computer was already cheaper than everything you needed to plug in to it.

With shipping (and the fact that Pis are currently hard to get) Pis can go up to $40 or $50.

In countries like India, that's expensive. Not prohibitively, but you think twice before using a Pi for a project. I'm currently studying at a tech institute and have been part of the group trying to get folks to do more tech. Generally we focus on Arduinos for freshmen because they're 3x/4x as cheap as a Pi and easy to get your hands on. I'd love to have more folks doing projects with Pis, but it's just too expensive.

Peripherals? They're lying around. You can find an old monitor or TV anywhere, or just share one with an existing machine. A lot of projects with Pis don't even need peripherals.

There is a lot of peripherals lying around. Even years ago when I didn't have two pennies to rub together I could still get unwanted keyboards/mice and even smaller unwanted screens.

On the flip side, I don't need a screen for all my computers. Some are network devices now (SSH only) and some share the same screen because I use them for different things at different times.

Right? But there are also plenty of computers with 512 MB of RAM lying around that you can get for free. And they're 'programmable' too.

Physical space, noise, and running costs are also important considerations for many people.

Yes, running this as a useful network device (storage, irc bot, etc..) off a small solar panel is pretty neat side project.

While this is not enough to run a latest MS or IntelliJ IDE and Chrome with 100 tabs open, 512 MB is adequate to run vim/emacs, some Ruby / Python / Go / Node.js / Rust / you-name-it code, and a moderate browser session. Quite enough to gain some highly marketable skills using hardware that costs a couple of cups of coffee to nothing.

Looks like your link broke walmarts website. Don't expect that from larger companies...

Working for me. Could have been a hiccup due to Thanksgiving/Black Friday shopping traffic.

As useless as walmart's website is, I doubt that HN could break it by simply visiting. As the other poster said, if it's broke, that's probably due to black Fri/holiday traffic.

Being on the front page of HN is worth about 20,000 hits a day in my experience. My $5/month VPS running every single page load through the incredibly slow BeautifulSoup parser handles it no problem. A link in the comments is probably an order of magnitude less traffic than that.

A lot of people already own TVs that would work as a display for this. My local Microcenter (which sells this new RPi too) has mice for $1 and keyboards for $4. You probably need to spend a few more dollars on cables and storage but I think you could easily walk out with a functional system for less than $20 total, maybe less than $15.

At thrift stores they can barely give away mice and keyboards. Just checked craigslist, keyboards and mice were $5 each and a monitor was $20. So a few hours work at minimum wage and you have a computer.

> This is great, but for people who want to "get started in computing", don't they also need a display, a mouse and a keyboard?

Who are you quoting there? It's not the article.

Is Raspberry Pi being marketed at beginners, or for conventional desktop use? That wasn't my impression.

That may not be the goal of this particular model, but it was the point of making the Raspberry Pi in the first place, yes. It was meant to be a modern equivalent to the BBC Micro; a way to give every kid affordable access to a computer with an aim towards teaching them computing (as opposed to using an office suite). They'd need a keyboard of some sort, but they could use a television as a monitor (as most of us once did). Languages like Scratch were part of the project. That seems to have gotten lost among the possibilities that a small, cheap Linux machine offers to the grown-ups among us.

Good point although remember that the (full size) Raspberry Pi has composite video output that can be used with the cheapest of TVs.

IMO the Raspberry Pi Zero is more of a reaction to the C.H.I.P $9 computer on Kickstarter and not a cheap computer for people who cant afford a full size Pi.

There are pin headers on the zero for composite out - should be cheap+easy to bodge a jack, but it would pretty much weigh as much as the zero itself...

True, but imagine an university that could have a few peripherals and each kid could have their own RPi and share the monitors.

Maybe it also makes it easier to do more daring things with it? it it breaks it's only $5.

Commodore 64s, ZX Spectrums, BBC Micros and the like never came with monitors, you just used your TV. This board has HDMI out on it, or composite if you solder in a connector, so you can use it with your existing TV or monitor.

A mouse is optional too, you can do everything you need to from the keyboard.

So really, it's only a keyboard that you need.

I'd be more interested in a cheap LCD+keyboard plastic thing with batteries. I hope raspberry is considering this option, or that somebody is trying it.

To be honest the cheapest android phone costs 60 bucks, so I wonder if there should not be some attempt to have a development suite that can be usable with a touch screen.

I'm guessing it's a rework on the compute module : small enough connectors to avoid having to remove them to start with and cheap enough to put it anywhere an arduino isn't quite powerful enough to work.

This sort of stuff is great for setting up some kind of automation. Places where you would want a small computer, without all the peripherals.

The use cases for such a device go as far as one's imagination.

Ah well. I recently (a week ago) started a new project with the Arduino and then they release this beast. It's basically cheaper than an Arduino (except that there's no storage in the Pi) AND faster than the Arduino.

Sure, it won't have the same IO capabilities as there's much more layers but it will be pretty close and I don't really need great performance anyways.

I guess, I'll have to switch to the Pi, it's not even a fair race at this point.

(luckily for me, I was mostly writing the SaaS that would work with the Arduino and only spent about a full day's work doing C/C++ coding so far)

Well, I would like to note that people on a budget have been buying Arduino Nano with USB for $2 from eBay for a long time.

Or a clone of SparkFun's ProMicro, which is absolutely awesome. It's basically Arduino Leonardo (Mega32U4) in the 'micro' form factor and able to work either on 3.3V or 5V. You can get it at around $3

Next awesome step would be RPi with integrated wifi/bluetooth, and it would become ultimate IoT/embedded development platform.

Such step would increase price, I know, but AFAIK, most people are almost always buying either wifi or bluetooth dongles anyway.

All in all, RPi ables to deliver exciting, rather (for me) unexpected and most importantly great and user-friendly products.

You can find zigbee, wifi or bt dongles for $8 or less these days. I suspect board development costs are the limiting factor for the Foundation here, it's easier to slap on USB and tell users to plug in what they need.

hm, I probably should check now, but were I live (non US), the last time I've checked zigbee prices started from 19$ :(

But still the best thing will be that:

- you will not need any dongles

- can embed your project immediately

- better compatibility with hardware and software (hopefully) :)

The headers/shield solutions could easily provide that too, so it doesn't have to be integrated into the SoC if you want to add something not already there. These type of electronics have essentially become digital Lego.

Have you heard of this? http://vocore.io/

I have read of many alternatives, though RPi has one feature, nothing can beat it today - community support.

99% problems you'll see on your RPi, you can find a solution to them in 5 minutes with Google.

Maybe it will sound like nitpicking and also RPi has versatility and extensibility for rapid development. Want add camera? No problem. Connect to screen? No problem. Few USB devices? No problem. GPIO control? No problem.

Probably it's just a rant with possibly very long discussion, but if you do not know what crazy or fun thing you going to do tomorrow, RPi probably got your covered, whereas others - not so much. So if RPi (let's say the 'normal' one, not the Zero) had builtin WiFi and BT, personally I would be covered in any way I need and would not need any dongles at all.

Wifi is a game changer IMO. I've just found out about esp8266, and its a mini revolution on the Arduino world.

Wow, 2x CPU, better Wi-Fi and they even managed to add BLE into this? I have just one question: how? :O.

Too bad they didn't include camera module connector (CSI). Low-latency hardware-encoded h264 stream from camera is one of the most cool features in RPi platform. The Zero would be great e.g. for FPV fying folks, if it supported camera module. (USB cameras aren't answer here, as they typically don't have low-latency hw-encoding.)

Yeah, I had hoped for that as well. But connectors cost money. Could the camera module be attached via the header?

> Could the camera module be attached via the header?

No. The Raspberry Pi GPIO header doesn't include all pins, and in particular it doesn't include pins that were used for other peripherals, such as the camera and display interfaces.

Interestingly you can get out DPI display signal from GPIO pins. Adafruit is using this technique with their Kippah display, giving you fully accelerated display signal (unlike those crappy low-fps-low-res framebuffer driven GPIO displays).

Correct. My understanding is that trick was made possible by design -- one of their engineers specifically chose pins to bring out to the GPIO header to facilitate it.

I've seen some variants on the "Kippah" to directly drive VGA.

I asked this on RPi forums and got clarification from their engineer. It's not possible to connect CSI camera to Zero via GPIO. Required CPU pads are not connected.

its not the connector, I have a suspicion Zero is 2 layer pcb

That's a shame. I was wondering if I could use a Pi0 with an old USB camera as a cheap IPCam.

You can. Just get a USB webcam. The GP just said you can't use the CSI webcam.

I just... I don't... I'm a bit lost for words.

The sheer amount of "stuff" you get for $5 (albeit USD) is staggering.

I am designing some simple electronics gadgets for Burning Man, and the electronics (low-end MSP430 based) for that is costing me a significant portion of $5 yet its significantly less powerful.

I know, I know, volume is a key issue, but honestly, that doesn't make it any less impressive to me.

The Pi foundation being good friends with Broadcom also helps.

Broadcom must be delighted to have managed to find a use for their obsolete part from 2007 or so. Some accountant, somewhere, is happy :-)

Probably a neat little board too, but unlikely to be $5. It's at £11.88 [0] ex VAT ex shipping at farnell UK (and can't be ordered anyway) -- that's more like nearly $20 these days.

[0]: http://uk.farnell.com/raspberry-pi/raspberrypi-zero/sbc-zero...

It's £4 in the Pi Hut (disabling the extras): http://thepihut.com/products/raspberry-pi-zero

Nice one, they are not too extortionate on the 'essential kit' either.

That pi zero might become an alternative (for me) to the Olinuxino Micro [0] and the NanoPi [1] -- these are ARM9 based so even more dated than the Pi Zero; however I can boot these two in 0.7 seconds to an application, so they've been the first draft of quite a few professional projects for me...

[0]: https://www.olimex.com/Products/OLinuXino/iMX233/iMX233-OLin... [1]: http://nanopi.org/NanoPi_Feature.html

On the cynical game I'd add that selling a pi zero at 5 then many phats to get audio, network etc, might provide more global revenues and maybe more profit :)

€9.61 EUR from PiHut in the UK, shipped to Ireland.

£5.99 if you can still find the MagPi magazine with one on the cover. I picked up two on my walk in to work this morning.

The Specs:

- A Broadcom BCM2835 application processor

- 1GHz ARM11 core (40% faster than Raspberry Pi 1)


- A micro-SD card slot

- A mini-HDMI socket for 1080p60 video output

- Micro-USB sockets for data and power

- An unpopulated 40-pin GPIO header

- Identical pinout to Model A+/B+/2B

- An unpopulated composite video header

- Our smallest ever form factor, at 65mm x 30mm x 5mm

Source: http://raspberry.piaustralia.com.au/products/raspberry-pi-ze...

Don't forget, also, that the ARM11 in the BCM2835 is actually just a secondary processor; the main CPU is a blisteringly fast dual-core Videocore 4, which lots of interesting DSPish and vector processing instructions. (You can call out to it from the ARM.)

See: http://maazl.de/project/vc4asm/doc/index.html

And if you're into shader programming, the GPU has a 24 GFLOP shader unit.


Only suitable for very specialised purposes, but astonishingly fast.

You can't actually get the USD$5/AUD$7 version from piaustralia/littlebird though. It's a kit for AUD$25 (plus AUD$7 shipping) and no other option.

I ordered one from PiHut in the UK because that was the cheapest I could find (AUD$16 including shipping).

I'd like an updated Compute Module. Given how small the Pi Zero is, it should be doable to fit it inside the layout and with the connectors of a MiniPCIe board (though, of course not with the pinout).

Also, the compute module has support for USB Slave mode but there is no documentation for this - I'd like to see some expansion in this area as well.

Or choosing of a CPU with a MII interface to allow real GBit or, heck, fiber/powerline/wifi adapters...

Same here. What I'm fantasizing about is a kit with a set of compute modules in a "tower stack" with a power supply, with horizontally oriented boards stacked on the height, with all compute modules connected to ports in the tower.

And you could choose where the pin-outs leads - either straight out to connect to some electronics on the outside, or straight down to a router in the bottom.

Then you could dedicate some pins on all boards to link to the router, making for a small compact HPC setup (easy to ventilate too) and connect stuff like per-module LED rows for module status info and whatever else you might like.

Bonus point for a "split" tower with modules on both sides, folding down like some toolboxes do: https://www.howdentools.com/data/howden/ul/1406847600/140933...

Google for PC/104 its exactly what you're looking for although its aimed at industrial control users (and unfortunately priced accordingly)

Of course if you could guarantee a production run of 5 million then a contemporary PC/104 board could probably cost as little as a ras-pi.

Note that there's PC/104 the concept and form factor, and another concept with the same name was a 104 pin header that was vaguely a PC/AT bus. After all, S-100 can do almost anything so an extra 4 pins shouldn't hurt anything LOL. Anyway my point is if you find references to the 80s era 286-AT level of performance bus, don't worry for a decade there's been stackable PCI equivalents and all that.

Something to think about WRT "limited" PC/104 bus speed is its not like spec-ing gamer graphics cards... if the purpose of the bus is to connect a 4 kilosamples/sec 12 bit A/D converter, even something as slow as RS-232 would not be a limiting factor, for example.

Neat. Not exactly what I imagined but probably a lot more practical OTOH. I'd just like a standardized high speed interconnection port with power supply and a shared router of some form.

Probably would be more plausible when high speed optical links becomes cheap and ubiquitous enough so that a single thin cable can be drawn from each module to an PoE equipped router, so there's not a large mess of wires. Then the on-chip pinouts could be used independently of the networking.

"Optical" and "PoE" do not go together?

One cable can hold both fiber and copper wires. I think some thunderbolt cables do that.

What's this stack for? I'd be very surprised if it beat PC hardware + GPU for MIPS or FLOPS per $ or watt.

Agreed for a supercomputer design it would be awful, but for a stereotypical 99.9999% CPU idle applications, the big win would be total cost of ownership / capex.

On a long term view from many decades of observation, the digital world is bifurcating into one world thats 100% utilized all the time in data centers or real time control or video gaming, and another world that's 99.999% idle but when its active the users want it infinitely fast. And the two worlds don't talk or cross pollinate and are gradually using separate hardware.

Having spent some time in electronics labs, given experimental results with smoke emitting diodes, and given the GPIO ports, there is minimization of the total cost of hardware experiments. If you assume 5% of boards will be destroyed per PID/motion control lab, then minimization of board cost is key to minimization of cost of running the lab. Implementing a PC that costs 20 times as much but provides 100 times the performance would only result in increasing the cost of the lab by 20 times.

There is also a total wattage ecological argument. I have a pi running a 3-d printer and it works great and only draws a couple watts, but the web application is extremely fat and pointlessly featureful "because the cycles are there so may as well use them" and its not like installing a 100 watt PC would make better prints, it would just make an even more elaborate UI to use all available CPU cycles thus burn more coal. A 300 watt tor node doesn't really do anything a 3 watt tor node does, other than burn 100 times the coal.

Most of this is the case for using one Pi, but not a stack of Pies. Unless you have some sort of micro-AWS which turns off the 99% unused ones. Ten boards drawing 100mW at idle are the same as one board drawing 1W at idle.

Except that when one board fails out of 10, there are 9 others still alive whereas in the other case you are toast.

I would expect the complexities of distributing a workload over 10 boards would result in downtime far exceeding that of a single more powerful machine left doing its thing, for all but a few highly battle tested applications.

A fair accounting includes the fragility of software, not just hardware.

Mainly experimentation.

Reference: https://www.parallella.org/

Registration is open for Startup School 2019. Classes start July 22nd.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact