Hacker News new | comments | show | ask | jobs | submit login
Raspberry Pi 3 Model B confirmed, with onboard BT LE and WiFi (fcc.gov)
518 points by merah 696 days ago | hide | past | web | favorite | 213 comments

Raspberry is the most important project that made my family interested in "computer things"

It gathered the whole family together to play old couch-games[0], something that most powerful consoles haven't even been close --specially with the grampas

Really exited about what the future will bring us!

0: http://emulationstation.org/

I must be the only one disappointed with old-school games. I was super excited to set up a retropie, after 20 minutes playing the nostalgia wore off quickly. Anyone else find the same?

I have a collection of old C64 games flying around on some of my external drives. Never got to play them. But one day I remembered Deuteros. An old Amiga game. I made it through the whole game back then without knowing English at all. I was so proud back then, I had to pick it up. I couldn't believe it but I played it through all the way again. I really can't say what it was exactly but I guess it was the flashback. The sounds, the graphics and understanding!

I continued with Lure of the Temptress. I only had a buggy version where you couldn't save. I played this with 3 other people. Beginning every Friday. Ending on Sunday... It was fantastic to finally finish it and being able to save.

But if I look at all the games I had back then, most of them are not worth it, but if you seriously look at it: they never were ;)

I totally remember Deuteros. It was riveting. I remember adding more and more stations and bases, I thought I had the game down, and then after deploying a certain number of stations, I triggered a reaction from the aliens and little by little, they started destroying all my empire. I was never able to beat that game.

But what a memory.

Oh ~~ people mentioning Deuteros! That's certainly something I don't see often when people talk about retro games. It was indeed an excellent management game.

You should try again ;)

I think some of that has to do with the fact that a lot of those older games aren't really as fun as we remembered them to be.

There's some classic gems, to be sure, but they were also all we had.

The bigger underlying problem is that we associate those games with happy periods of our lives. The once where we could play games, do projects and not worry about anything else.

By playing an old school game, your brain extracts these happy feelings from the memory. However, it's not a sustainable source of "fun" per say. As opposed to the newer games where "fun" is being caused by the game directly.

I strongly disagree. I have zero interest in today's bloated AAA games. Simple games like Tetris, Mario + romhacks, or Civ II have given me collectively more fun that I would ever have playing some insipid shooter or uninspired pay2win fantasy game. I wouldn't trade them for any of today's games, which I consider severely flawed in many ways.

Sturgeon's law applies. We only play the 10% (or less) of old games that are good. There are good new games, too.

I don't doubt it. I only say that the percentage of good older games is at least equivalent, if not greater, than the percentage of good newer games.

i disagree with you. i think the bad games end up being forgotten.

Nope, they are still getting a lot of exposure on retro websites and even GOG (despite calling themselves Good Old Games there's also some not-so-good games in their catalog).

When they speak of "old" games, I do not think they are thinking of Mario or Tetris, which are a decade newer and orders of magnitude better than some "classic" games.

A 31 year old game I feel qualifies as "old". What is old, then, pong on an oscilloscope? :p

Sure, Moon Lander. Original Sprint. The dragster one. 31 years ago was at about the end of the Golden Age of video games.

The golden age ends roughly 16 years after the birth of the observer.

Ha ha, I get your point, but the GAoVG is a specific era, roughly 1979-1985. :)

Wikipedia has https://en.wikipedia.org/wiki/Golden_age_of_arcade_video_gam...

Restricting to arcade games is probably justified. I'd say we are in the golden age of video games right now---at least in terms of popularity. And you can still get remakes of all the good old games, or even play the originals.

Well, there was a first golden age, which was followed by nintendo and the street fighter era, and so on. That's fine you want to make up new Golden Ages -- more good attention to video games the better -- but be sure people aren't thinking of the earlymid 80s when you use the term for other purposes.

> There's some classic gems, to be sure, but they were also all we had.

Some stuff has aged, but some of the best games are in the past and have never been "reached" by any newer game. Ultima 7 is an example of such games - there's just nothing that comes close to it even in modern RPGs.

For 2D platforms the best was made in the past again. All the newer Mario games are either too easy or uninspired, or both.

There's a whole bunch of genres that has been completely a abandoned by publishers and devs for years (simulators, adventure games, RTS, even RPGs - sure we get Witcher 3 recently, but there's really nothing much else on the radar, there's too few of them now compared to how many we were getting back in the days).

It's been many years that the main production money goes into making GTA, First Person Shooters, Open World games, Racing games and that's about it. Of course the indie market is very much alive, but you can't compare the level of contents made 20 years ago by large teams versus 1-2 folks making games nowadays.

I'd argue that whilst very different in philosophy, The Witcher 3 is about the closest we've come to Ultima VII since it was released.

You might also find Age Of Decadence interesting - it's a super-indie made over a decade or so by a very, very dedicated team. I've not played it and I understand it's flawed, but it looks to be reaching for the same kind of freedom as Ultima VII.

Taking turns with my son, playing Super Mario World on my RPI2 with Lakka is every bit as much fun as playing that game originally.

Nostalgia is a powerful emotion. People always remember the truly great games but seldom remember the countless bargain bin cartridges they popped out as a kid after muttering to themselves, "Wow, that sucked."

Well, you know what they say, "Nostalgia ain't what it used to be".


I dunno. I just set up DreamSNES for my son the other day and he played Metroid until we basically had to threaten to burn all of his toys just so he would put down the controller. But yeah, I guess speaking of myself there's not a lot of lasting interest there.

Metroid or Super Metroid?

IDK the one where you jump out of a crashing spaceship and explore some lonely tunnels.

That's Super Metroid. That's the Metroid I played first---on an emulator as well. It's one of the greatest games for the SNES.

I tried the NES original Metroid (also emulated) later, and I was surprised of how many of the game elements were already in place on the much weaker system.

It depends on the game and your expectations. Super Mario Kart is basically obsolete unless you have very specific acquired tastes. Super Mario Bros 3 is still one of the greatest games ever created, but if you go play it, it will just be better than most other platformers, not bring back the magic of childhood.

Gaming has evolved tremendously and there are a lot of not so great elements to vintage computing. I go back to games I spent days straight playing and find them impossibly awful.

That said there are a lot of absolute gems. To find them search youtube for "best <system name> games".

Also, lots of classic PC games have fan made patches (or even remakes) that often remove bugs and help balance.

Master of Orion and the original XCom are just two examples. More recently, the gamemaker version of Spelunky has people still fixing it.

... Kings Quest, Falcon 4.0

I have thoroughly enjoyed exploring the old 8-bit days with my kids - but yeah, the games are pretty boring after 10 minutes.

That said! There are tons of great gems in the old archives.. you've got to find them .. we could play Chucky Egg for days on end in our household .. but we definitely had to search to find it ..

One thing you shouldn't discount is the idea of giving your kids access to the 8-bit machines to learn how to program them. This has been immensely valuable to our 8 and 5 year olds' .. they play crap games, but then right away can get access to the code, understand what it does, and learn from it. My 8-year old spends more time reading code (BASIC) than playing games now .. and that is a delight to see.

> but yeah, the games are pretty boring after 10 minutes.

But most of them were even back in the day. When playing C64 games with friends back in the day, we'd spend almost as much time figuring out what games to play and loading different games as we did playing.

Some games we'd play for ages, but many were 10-minute entertainment at the time.

And others we'd play for ages only to suddenly stop when they got out of the sweet spot in terms of challenge (e.g. Commando and Tapper were two of my favorite C64 games for a long time, but in both cases they had fixed points for bonus lives and once you got to a certain skill level you could basically play "forever", just cycling through the same few levels over and over while accumulating lives, and suddenly they were no fun any more)

I've found that they are great fun to play with groups of people, particularly when it is a group of people that remember playing the games when they were new.

I, like you, have been mostly disappointed when it comes to playing them alone.

I totally not feel that way. I've had more fun playing Fire Emblem games lately than I had with any game released in the past 5 years.

It must be due to the fact it is not the original hardware with the same controllers, etc. That's one thing that goes against the nostalgia feeling (other than bad games).

For example, when I see an original arcade machine (of an excellent game), I cannot not play it.

I don't get how ROMs for the old systems continue to be so easily available. I'm really glad they are, but how have all of the ROM hosting sites note been DMCA takedown'd into oblivion?

Some game have "lost" ownership. Say that old game studio that developed GameBoy game has gone bankrupt, no one is going to enforce the copyright

Most game or not reedited so there is no money to gain fom preventing them from being online. It may even help to keep the brand alive in case a reboot in programmed (Eg: tomb raider)

Often it's a simple calculation of how much revenue could be won if you factor in the fact that you need to pay the team to find the ROM's, send the letters and enforce them once in a while. Basically it's nearly never worth it.

Are you French?

"reedited" should be "republished", I think :)

There's no money in it.

Virtual Console? Also wouldn't publishers be worried about retro classics competing with their new AAA offerings?

If they are worried about that, they have bigger problems.

Some of them have (although I'm not aware of any that only host older games that have).

In any case: this is the internet. You get whatever you want, and the effect of DMCA on that is indistinguishable from zero.

Because not everyone is in the US.

Retropie makes emulationstation on the pi really easy!

+1 for Retropie. So much fun.

If you have a Raspberry Pi sitting around that you don't know what to do with, just do this. You won't regret it.

For real though, it's crazy how easy it is. I found my original Pi buried in some old stuff and was just like 'screw it, I'm going to do something with this' and then I was playing Pokemon Yellow that weekend. I've been working on the design for a cabinet for a little bit, but haven't had the chance to build it yet:


If I may add - do yourself the pleasure, and use a standard usb Xbox 360 controller, otherwise you are in for a world of pain. I've tried a NES30 controller over bluetooth and after a week of trying different configurations nearly chucked the whole thing in a bin. Also some of the emulators are too slow to be usable, even on the Rpi 2, NES and SNES work absolutely fine, but MAME, N64 and Dreamcast are more like curiosities, very few games run at anything even close to full speed on them.

Actually, for N64 emulation, I have a 100% full speed fork of Mupen64Plus for the Pi 2 sitting around on my hard disk that I hacked together a while ago. I should really clean it up and post it to GitHub.

There's no fundamental reason why the N64 can't be emulated at full speed on that hardware. The VideoCore IV blows the RDP away in fill rate, after all. It's just that the available video plugins are old and aren't optimized for mobile GPUs. Additionally, Broadcom's drivers have a tendency to stall in inconvenient places and the emulator needs to work around that as well. But once those are fixed, most of the popular N64 games run beautifully.

Please do! I couldn't get full speed Mario on my i7 Mac which is way more powerful than a pi!

...Which is sad. I played Mario64 in Corn on a 400MHz K6-2 around Y2K. Nemu+1964+Project64 each handled more, but it seems like I had to wait for my Athlon XP before my machine was fast enough to be useful.

Good emulation is hard, especially when you're dealing with code that was built around the bugs, quirks, and timing issues that the original SNES authors had to deal with. Then you have to worry about buggy software that works because of undefined behavior [1, see Speedy Gonzales]. More recent emulators have focused more on correctness than speed, because it means less work trying to patch and hack around broken games.

[1] http://helmet.kafuka.org/state3.html

I definitely know that. I've done work on an NES emulator, and a lot of the software uses the hardware in undocumented ways, like odd timing quirks, undocumented CPU opcodes, changing display registers mid-frame to achieve special effects, etc.

I've looked at the SNES and N64's hardware to consider contributing to an emulator for one of them, and the hardware certainly doesn't get simpler as time goes on ;-)

The comparison I was making isn't really fair, anyhow. Corn was fast, but the last revision only really covered two commercial games (Mario and Zelda). I'm sure that they heavily optimized for the code patterns in those specific games, without regard to accurate emulation of the hardware in the general case. Expand the supported cases and the problem immediately becomes much harder. Comparing Corn to an accuracy-focused emulator is like comparing one of the cut-down, portable-friendly SNES9x builds to Higan.

Please do this. I find the current N64 emulator in RetroPIE lacking regarding speed and support (even overclocked).

I completely agree with this. If you do choose to use a wireless Xbox 360 controller, make sure to get an official xbox controller wireless receiver. My wife purchased a knockoff for ~$7 and I spent a couple hours trying to get the controller to sync. Using an official controller, the drivers with emulation station just worked and I was up and running in five minutes.

I have the SNES30 and the SFC30 working great on my rPI2 - with retropie 3.5, you just have to make sure to create the UDEV rule, then it works like magic.

Goldeneye works quite well on a rpi2

Hmmm, I got a pretty bad frame rate with RetroPIE even with it overclocked.

I recommend installing Retrosmc https://github.com/mcobit/retrosmc over OSMC https://osmc.tv/ and hooking it up permanently to your TV.

I use a PS3 Dualshock controller and it works wonderfully. :)

RPi is IMHO the most important computer product period. It's the gateway to PC 2.0, the modern Altair.

correction - excited

Comparing the photos to my Pi 2, it doesn't seem very different. There's an extra IC near the microsd card, a connector beneath the HDMI port, and a few other small changes.

But nothing that looks like an antenna. Any idea where that might be?

Edit: as lovelearning pointed out, there is a small ceramic-looking piece in the upper-left of the top side, near the GPIO pins and LVDS display connector. That might be it.

Look for a cutaway ground plane as a hint to where the antenna may be.


Wifi antenna and chip set highlighted.

That is it, and the connector beneath the HDMI port is the same (but mounted) JTAG connector as the RPi2.

I don't know what is the chip, though.

Also noticed a 2-pin header between the 40-pin one and the upper USB ones: it looks like the "RUN" from the Rpi Zero.

EDIT: The 2-pin "RUN" header already existed on the RPi2, where the ceramic antenna is now located on the RPi3.

There is a big Diode D5 near the micro-USB power connector.

Also worth noting: the new IC and the suspected ceramic antenna are in about the same place on opposite sides of the pcb. So I think that lends some credence to the idea that is is an antenna.

Man they can make small antennas these days...

First thing that came to mind. All the wifi ready SBC have a very easy to spot chip and some form of visible antenna. I'm almost thinking the shots were just for gross form factor approximation.

According to FCC report they are using a chip antenna.

Second last "internal" photo, top left corner next to the ribbon connector. That thing that looks like a long cap SMD is the chip antenna (just guessing).

All that site returns here is

    FCC Federal Communications Commission
    Security Violation

    Your request looked malicious and has been blocked.
    You can use your browser's Back button to return to the previous page.
    If you have questions, please contact the FCC at 1-888-225-5322
    If you think that you have reached this page due to an error on our part, please contact let us know.

Lucky you, I'm straight up blocked. From multiple IPs even, chrome throws "ERR_CONNECTION_RESET" on 4G.

And here's "curl" from my server

  * Hostname was NOT found in DNS cache
  *   Trying
  * Connected to apps.fcc.gov ( port 443 (#0)
  * successfully set certificate verify locations:
  *   CAfile: none
    CApath: /etc/ssl/certs
  * SSLv3, TLS handshake, Client hello (1):
  * SSLv3, TLS handshake, Server hello (2):
  * SSLv3, TLS handshake, CERT (11):
  * SSLv3, TLS handshake, Server key exchange (12):
  * SSLv3, TLS handshake, Server finished (14):
  * SSLv3, TLS handshake, Client key exchange (16):
  * SSLv3, TLS change cipher, Client hello (1):
  * SSLv3, TLS handshake, Finished (20):
  * SSLv3, TLS change cipher, Client hello (1):
  * SSLv3, TLS handshake, Finished (20):
  * SSL connection using DHE-RSA-AES256-GCM-SHA384
  * Server certificate:
  *        subject: C=US; ST=District of Columbia; L=Washington;; O=Federal Communications Commission; businessCategory=Government Entity; serialNumber=06 19 1934; CN=apps.fcc.gov
  *        start date: 2015-09-13 15:38:08 GMT
  *        expire date: 2016-09-14 00:27:01 GMT
  *        subjectAltName: apps.fcc.gov matched
  *        issuer: C=US; O=Entrust, Inc.; OU=See www.entrust.net/legal-terms; OU=(c) 2014 Entrust, Inc. - for authorized use only; CN=Entrust Certification Authority - L1M
  *        SSL certificate verify ok.
  > GET /oetcf/eas/reports/ViewExhibitReport.cfm?mode=Exhibits&calledFromFrame=N&application_id=Ti%2FYleaJNSl%2BTR5mL5C0WQ%3D%3D&fcc_id=2ABCB-RPI32 HTTP/1.1
  > User-Agent: curl/7.35.0
  > Host: apps.fcc.gov
  > Accept: */*
  * SSL read: error:00000000:lib(0):func(0):reason(0), errno 104
  * Closing connection 0
  curl: (56) SSL read: error:00000000:lib(0):func(0):reason(0), errno 104

Is it being slashdotted and responding like it is a DoS attack maybe?

It worked for me.

If someone were to call the FCC regarding this and write about their experience, I would definitely read that article ;) The phone number is right there!

Copy/paste the URL in a new tab, it blocks connections with the referer [sic] header set to HN.

Given Broadcom's connections to the project, it's almost a surprise they didn't have some kind of Broadcom option for WiFi/BT earlier. Must have stung at Broadcom to be powering the most popular hobbyist ARM Linux board, but almost every one using WiFi powered by Realtek,Ralink,etc.

Broadcom's "connection" to the project often seems circumstantial, like the Pi project is a blip on nobody's radar. Several of their engineers used to work at Broadcom and they've been under NDA lock and key, bending over backwards to help people progress on things that they can't officially talk about by leaving the least subtle clues ever so that they can move the process forward.

The GPU/camera integration in particularly is really painful - people have gone to insane ends to reverse engineer how it's working so that they can access features that aren't documented.

As a heavy Pi+camera user (I use 70 at a time every time we ship a unit) I appreciate the efforts of all these folks so much.

>bending over backwards to help people progress on things that they can't officially talk about by leaving the least subtle clues ever so that they can move the process forward

That sounds pretty interesting, where can I read more about it?

If you hang out in the camera board and read a certain kind of post - generally people trying to get the camera to do stuff it's not easily able to do - you see guys like 6by9 and jamesh doing their best to help without getting in trouble.




I think even the founder Eben Upton works at Broadcom too.

There is an official Raspberry Pi foundation wifi dongle that uses a Broadcom wifi chip with full support in the kernel. For a time it was the only dongle that worked with Windows 10 IoT on the Pi. Can see it here: https://www.adafruit.com/products/2638

Ah interesting, thanks!

It'll be interesting, given the state of Broadcom wifi drivers on linux.

Looks like it's using a broadcom WiFi & Bluetooth chip: http://i.imgur.com/exuZy58.jpg

I'm a little disappointed that (from a brief glance at the documents) it doesn't appear to support the 5 Ghz Wi-Fi band. I wonder if it's a power issue, or just limited space on the PCB. The 2.4 Ghz band gets kind of crowded in dense urban environments.

Probably a cost saving measure. The entire board feels like it was put together to tick off a bunch of features. 64 bit quad core with bluetooth and wifi? Check. Meanwhile the clock is still a measly 1.2GHz, the wifi is only 2.4GHz, the ethernet is still 100Mb and the chipset is still broadcom. They are struggling to stay relevant. (Remember, it was only 1.5 years ago that their official word was "No RPi2 until 2017".)

The Odroid C2 goes on sale in a few days for $40. It's got a 64 bit quad core CPU (just* like the Pine and RPi3) but clocked at 2GHz. And has real gigabit. And has twice as much ram. And can drive a 4k screen comfortably. (Video-out was the one place where the RPi has traditionally excelled.)

* Okay, not entirely just like. It's an Amlogic chip instead of Broadcom or Allwinner. Amlogic is one of the few without blatant GPL abuse.

Same here. I moved the AppleTV and the Roku to a wired connection and had to set up a second wireless network to be able to browse from the kitchen. Any time I fire up Wifi Analyzer on my tablet I see three dozen different networks.

Making computers easy to use was, definitely, a huge mistake.

Assuming that last sentence is a joke...

It's known that many a true word is spoken in jest.

I have this problem too. 2.4Ghz in my house, no matter how good of a router we use, gets up to 10% packet loss on average.

That's a bummer. Other than the Nest, the Pi is the only 2.4ghz device left in the house.

Also, good on the Pi Foundation for keeping this listing in the open rather than requesting 6 month confidentiality for the pictures/test report (which almost all companies do).

I don't think they did: the confidentiality period is now 45 days only by default, see http://acbcert.com/documents/misc-docs/Memo-Short-Term-Vs-St...

Certification took place at Track (now Element UK) on 2015-11-23, given the external picture dates, or 95 days ago, so it looks like it has been extended once.

They didn't request short-term confidentiality for anything: the exhibit summary list at https://apps.fcc.gov/oetcf/eas/reports/ViewExhibitReport.cfm... has a column for short-term confidential which is all No.

Also, the confidentiality letter would specifically request short-term if they wanted it.

You are right, but today it is very difficult to get an extension to the 45 days period, unless you can provide a release date, as it has been abused so much in the past.

Why does Raspberry Pi not need to conform to the new "standards" of the FCC that require locking down firmware in routers? If it has onboard wifi, and you can turn it into a router... what's the difference between someone installing OpenWRT on a raspi vs. installing OpenWRT on a TPLink?

edit: This is a legitimate question, not an opinion.

Okay, first a bit of background on the wifi ruling. The WiFi issue is because the 5ghz WiFi band has some rather sensitive pre-existing operations, like weather radar, that are intermingled in some of the channels given for WiFi usage. There are a lot of times and places where such items aren't in use and can be used as additional wifi channels, but when there is a need, those channels need to be vacated and not considered usable for at least a half hour after receiving a signal. However, there were a lot of companies that weren't upholding their end of the bargain, so the FCC put in more stringent testing requirements, including ensuring that the radio cannot transmit where and when it's not supposed to. Because of this, most companies just took the easy way out and locked down the whole router, instead of putting the wifi chipset's control software in ROM or some other way of locking down the firmware.

Now, the Pi 3 does not talk on the 5ghz band at all, so even in Soft-AP mode, it's not capable of even coming close to violating the rules. Other desktop chipsets that can run in the 5ghz band have programming to not allow them to be APs in the 5ghz band, again steering clear of those rules. The 5ghz ruling is complicated because there's a lot of issues that tend to be glossed over in articles about it, which just muddy the waters.

The FCC was mostly concerned about DFS in that case, which only applies to the 5GHz spectrum, and is only applied by the AP in a wireless network.

Note that if you have a 5GHz wifi dongle, you'll see that the Linux regulatory framework will disallow running it in AP mode for this reason - almost all lack DFS.

My guess is that it's simply because it's specifically not sold as a router. Just like I could turn my laptop into a router, but it's my decision, the device is not sold as a router.

As far as i'm aware those are all still proposed regulations and not in effect yet.

I think TP-Link is already shipping with a firmware in lockdown


But that's just because TP-Link wanted to, not because of any FCC rule (at least, that's what an earlier discussion here on HN about that router said)

Wouldn't that apply to any motherboard that has a wifi antenna on it?

Perhaps their wifi chip can not operate outside FCC limits? The flashing question comes up for devices which can be made to operate outside the frequencies and power levels allowed by the FCC.

Do you have any more info on this new "standard" from the FCC? If true, that…sucks.

I believe he or she is thinking of the FCC proposal regarding software-defined radios.

It was widely (and incorrectly) reported as requiring all routers to have locked firmware, but that's not really the case, and if the proposal were to go forward there are a variety of ways that the SDR could be locked down while allowing firmware updates in general.

It's not clear that the FCC is even going to go forward with it, though; industry pushback was quite heavy.

edit: it looks like they went forward with the guidance, after amending it to be clear they specifically weren't going after open-source firmware

TPLink has already started locking down its firmware in response to FCC:


Eh, I'm skeptical that a chat customer-service flunky really has deep insight into high level regulatory-compliance decisions -- especially surrounding regulations that have only been floated as a proposal and which don't, yet, even exist (edit: the guidance seems to have moved forward, but only after amendments making it clear that the FCC specifically wasn't banning open source firmware)

TPLink has always been a shit company when it came to firmware, and routers with locked firmware predate by decades this FCC proposal. I'm not convinced there's a cause-and-effect relationship here.

Edit: Anyway, from the horse's mouth back in Nov 2015 after the comment period closed on the proposal:

"One immediate outcome of this ongoing dialogue is a step we’ve taken to clarify our guidance on rules the Commission adopted last year in the U-NII proceeding. Our original lab guidance document released pursuant to that Order asked manufacturers to explain “how [its] device is protected from ‘flashing’ and the installation of third-party firmware such as DD-WRT”. This particular question prompted a fair bit of confusion – were we mandating wholesale blocking of Open Source firmware modifications?

We were not, but we agree that the guidance we provide to manufacturers must be crystal-clear to avoid confusion. So, today we released a revision to that guidance to clarify that our instructions were narrowly-focused on modifications that would take a device out of compliance. The revised guidance now more accurately reflects our intent in both the U-NII rules as well as our current rulemaking, and we hope it serves as a guidepost for the rules as we move from proposal to adoption."

If TPLink really is locking out firmware "because of the FCC", they've simply misinterpreted things.

I couldn't spot the typical wavy PCB antenna pattern anywhere. Is that small cube in the upper left corner of external top view a ceramic antenna? Looks to me like the 2 LEDs have been shifted further down towards lower left and replaced with something new.

Update: Just noticed the radio testing report says single PCB mounted chip ceramic. I still can't spot it with certainty.

Yes, this is the antenna.

I cant believe it took them that long, Broadcom is famous for their BT/Wifi chips, buildin wifi was a no brainer, and cost shouldnt be an issue when you have great connections inside the company.

From their blog posts and forum discussions, I have the impression BC was not exactly enthusiastic about the Pi to start with, but got convinced only after Pi v1 become a hit. Perhaps there is still some reluctance and resistance inside BC? After all, they did take ~3 years to upgrade from ARMv6 to ARMv7 while all the other SBCs were already using ARMv7 cores much earlier.

Are you sure that's the antenna? It looks more like it'd be a balun. It even has what looks like a feed line going to the keepout that's right next to it. I feel like the one in the picture is missing the actual antenna.

balun (white ceramic part with brown rectangle marking) appears to be on the other side between the chip and via to the antenna

http://www.digikey.com/catalog/en/partgroup/2-45ghz-balun/52... or similar

Oh, it's on top (which they totally said). I'm not a smart man.

Is a ceramic wifi antenna's range comparable to that of a regular PCB antenna?

Usually slightly better, as it can be a 3D structure.

More accurately, better if considering antennas of equal footprint area, however chip antennas are usually designed to be very small (like this one) as the extra cost is only justified when space is at a premium.

In my experience in practice a good-sized PCB antenna beats tiny ceramic antennas. I would expect the radio performance of this Pi to be pretty poor, though better than using one of those tiny USB adapters (which also typically use even smaller ceramic antennas with really awful ground reference when used in a Pi as they are designed for use with laptops).

Also, another reason for their smaller size is that ceramics can have very high dielectric constant which reduces the size of the antenna.

It looks like the ceramic antenna is the light brown SMT between the DSi connector and the upper left hole in the top view pictures.

I love how the FCC still has a Sun web server

Slightly surprised to see that in combination with a *.cfm URL. I guess ColdFusion had a SPARC port?

Sun web server is a Java web server, and ColdFusion is a Java application.

ColdFusion runs on the JVM, I think.

I noticed the favicon too. :-)

That doesn't understand mime types for PDF downloads.

Neat :) It'd be cool if they use non-usb ethernet too, fingers crossed!

Are the BTLE and WiFi the only changes?

Still same CPU and RAM as the 2?

Since this is the first showings of it, we don't know yet. I haven't read the data yet but it'll likely cover only the newly added RF stuff, and the part 15 compliance for the FCC. We'll have to wait to hear from the pi foundation on the rest of the specs.

From elsewhere, I've read that it's the same 1G of RAM, but the CPU is now a quad-core 1.2Ghz arm64.

Yea that got confirmed last night on the release of it. I can't wait for the software stack to catch up for the arm64 stuff. I hope the pointer size change doesn't negate too much of the speed gains it can bring.

Video and images of the Raspberry Pi available here: https://www.youtube.com/watch?v=7yXIC1UVKx0

The Raspberry Pi website says they're unlikely to have a new version in the next few years. Do they have FCC now? Maybe they just want to add on the radio modules.

From the FAQ: > there are no immediate plans to release any more new models. A further new model may be released in 2-3 years, but this is not a firm schedule.


Saying you've got an awesome product around the corner is a good way of causing the Osborne Effect [1]. Also, plans can and do change. It could very well be an issue where they weren't planning on it, but the people at Broadcom came up with an offer that was too good to pass up. Plans can and do change.

[1] https://en.wikipedia.org/wiki/Osborne_effect

just like they had no plans for rpi 2 until the day it was released :)

They said that, then they Released the Pi Zero.

That statement was in Feb 2015 after the pi2... the PiZero was Release Nov 2015....

One step closer to accessibility for that don't have all sorts of usb dongles laying around.

All we need is USB 3 and it would be perfect.

I also want USB slave/device/OTG. I've been running a BeagleBoneBlack which does have it.

As I understand it, the port that USB is implemented on for the Pi machines is actually meant as an OTG port, made to act as a host by the drivers. On models with more than one USB port, there's a hub+ethernet chip. On the single-USB boards, they can actually be configured to act as OTG devices.

The whole situation means that OTG is unavailable on most models, models that have it only get one USB port (so OTG/host is an either-or proposition), and for the multi-USB+ethernet models, performance over those interfaces is worse than you'd expect (if I remember, BBB gets several times the transfer over USB and network, all at lower CPU usage, right?)

Pi Zero has OTG, but only one USB port total.

or one PCIe line. Something, anything to get fast I/O.

I am curious of your application for USB 3? Care to elaborate?

On the Pi 2, the ethernet port was really a USB-to-ethernet port plugged into a hidden USB port. So network speeds were limited to USB 2 speeds. I'm guessing the Pi 3 also works like this, instead of having pure onboard ethernet? But I don't know yet.

> So network speeds were limited to USB 2 speeds

Given that USB 2.0 is 480Mb/s, and Ethernet is 100Mb/s, there is no bottleneck when using the Pi's Ethernet port. I really wish this myth that "Pi Ethernet is slower than dialup because USB" would just die.

Of course, like everyone else I'd love to see a new version of the Pi with Gigabit Ethernet directly connected to the system bus, and USB 3.0 while we're at it. But that would greatly increase the cost of the device and also require a newer SoC that supports such things.

>Pi Ethernet is slower than dialup because USB

I can't find any info to say which bus the MMC reader is connected to, but I know if you use a USB drive as well as the built in ethernet port, speed will suffer.

None of them. The SD card connects directly to the CPU.

I just want a 64bit one. I put in an order for a Pine 64.

pee 3 will be 64bit :)

What's a Pee 3? :)

We can still dream can't we? I hope to see this in the future.

If it's using the same bcm2836 chip, then I think it'll have the same limitations. The markings on the chip in the pictures aren't quite legible enough to me, playing around in a photo editor, but it's marked the same way that the Pi2's chip is, from what I could see.

I could use it as a very cheap file server for my home / small office.

Also, USB 3 can drive video at pretty fast rates too (for multimonitor setups).

Fast I/O, for example ? Check the Pi 2 cluster vs Xeon quad core

I'll be picking one up. Onboard WiFi is a big deal.

I'm still waiting for a Zero. A Zero with onboard Wifi would be brilliant: really small, needs little power, and you don't need to hook it up to anything but the power supply.

You're going to be waiting a long time.

Not necessarily. I picked one up about a week ago, I guess? Impulse purchase.

Keep an eye on etailers that won't back-order, and just nab one when they come into stock. There's a stampede, but it's then luck of the draw.

I wouldn't mind onboard power supply though

With a Zero with an on board power supply, Wifi, and Bluetooth the possibilities are endless.

Here's someone working on a ESP8266 module for the Zero, which will include a simple power supply:


Looks like they are planning to offer a few for sale eventually.

Get a Chip!

Seems like you'd get a Pi Zero before you'd get a chip. They don't start shipping until June, and I see Zeros in stock for a few seconds (some on Adafruit yesterday, although they sold out in seconds).

I have a CHIP, from the Kickstarter campaign. It really is a better deal than the Pi Zero, unless you have Pi-specific needs. It's much more powerful than the Zero in my testing (though not quite as powerful as the Pi 2) and the built in Wi-Fi/BT is awesome. The fact you can attach a LiPo battery and have it managed by an onboard power regulator is worth the whole $9 as far as I'm concerned.

The only downside is no HDMI or even VGA until the respective adapters come out...in June. I currently only have one screen in the house with a composite input and it's not always available for tinkering.

Another downside is the size of the community. I think this is why the pi is going to be around for a long time despite not having the best tech specs around. There are all kinds of tutorials for hundreds of projects and the prevalence and availability of rpi forums further lower the barrier for entry.

Couldn't you already just plug a fairly cheap USB dongle in your rasperry pi ? I'm not sure I see what the big deal is, aside from probably making the price cheaper and freeing up a usb slot

If you've ever tried to get wifi working (esp on earlier RPis) you'll understand it is not that simple. Especially for newbies. Having said that its a good baptism of fire.

Raspberry Pi with built in wireless comms means it uses a standard piece of hardware so the software stack can be configured to work out of the box. That can't happen with the existing arrangement. Also the power supply problems with original Pis are frustrating to say the least. These are most evident when plugging in USB wifi devices. Again having a standard integrated piece of hardware will (hopefully) avoid this.

This doesn't really reflect my experience at all with the RPi 2. The edimax or official wifi dongles "just work" with NetworkManager.

With the early RPi it was a bit different because of USB power problems and kernel bugs, but those are mostly solved.

Experiences vary. I agree with tonylemesmer, getting them to work has not been trivial for me.

Oh you can, its just little kids tinkering around (and the parents who are going to buy the thing) having it "batteries included" is good thing.

Aren't those two pretty big upsides?

You're fairly limited in terms of antenna size/placement when you use a tiny dongle. If it's built into the Pi, they could have the antennae wrap around the edge (similar to how the iPhone does it) to allow for a much higher quality connection.

At least, that's what I hope they're doing with it.


But you can get almost any wireless ability and keep modern with whatever standard changes by simply using a usb stick.

They are super tiny now and barely stick out of the usb port.

Why hardwire wireless and outdate a board a few years later?

I think in 5 years we got g -> n -> 5ghz -> ac and three bluetooth variants.

Now there is indoor lte and other stuff coming.

So, how much time did it take you to get the WiFi stack working on your RasPi? Oh? You've never actually tried to get WiFi working on a Pi? Thought so. Getting WiFi to work on a Pi has historically been an exercise in very, very careful shopping for a dongle that actually works. Or getting lucky. Things may have changed recently, but many Pi users have stubbed their toe on the finding-a-Wifi-dongle-that-works rock. Have WiFi built-in that "just works" is going to be hugely beneficial.

It's now easy to find reviews on amazon for many wifi dongles specifically mentioning rpi compatability. And they just work.

Im sure you could still buy something unsupported if you didn't do your homework, just like nearly anything.

> So, how much time did it take you to get the WiFi stack working on your RasPi? Oh? You've never actually tried to get WiFi working on a Pi? Thought so.

Wow, attitude much?

Just to counter your anecdotal data with some other anecdotal data, I have 2 different random USB wifi dongles I got from ebay for a couple of £ (different chipsets) and they were both plug-and-play.

Despite having left Linux as my full time platform before wifi was ubiquitous, it took me 10 minutes or so. That included looking up how the ui was supposed to work. I don't have a dog in this fight, but getting wifi running took me less time than figuring out the volume settings.

You can still add on any wireless-usb you want. And for a $35 computer, upgrading to a newer model probably won't be much of an issue.

I personally like this, as it cuts my initial setup-time/cost.

The FCC's OET website is down at the moment, I'm guessing probably because of HN traffic. You can find the document cached here though:


having recently sourced near identical components, the combo wifi/bt chip was around $1.40 in large volume. I'd wage in 40nm its < 50c for BRCM to produce depending on the break they got. ant is < 30c.

gotta assume this is connected via USB so it won't be the lowest power - taking up the 2836's sdcard slot would be 6 IO out of the GPIO map which would be over kill. so most likely this solution needs the onboard USB hub / ethernet chip and its less likely they will bring connectivity to the cheaper versions.

pretty happy with this - saves 2 USB peripherals!

wifi chip is most likely connected over SDIO

Please have GbE

Raspberry Pi's 4th birthday coming up...


"We are going to be doing some celebrating here at Pi Towers on Monday: February 29 is the 4th anniversary (or 1st, if you’re prissy about leap years) of the first sales of the Raspberry Pi 1."

Might be an auspicious date to launch a new version of hardware?

3 Generations of PCBs and still no RTC onboard...

Probably because of lack of demand. Most people use the RPi in a scenario where they either have internet, or don't care about time.

Wow, this will give my chip a run for its money/performance. My favorite thing about the chip is the onboard connectivity.

Excellent news. I wonder what the radio components will do to overall power consumption of the board. It should be considerably lower than a USB combo wifi / bt device, anyhow, no?

I don't think it's going to happen but it would be really neat if it supported Vulkan. I want to try to build an arcade machine out of cheap, underpowered hardware.

It also has BT Classic: section 7.5 of the test report states "The EUT is a small, single board, computer with WiFi, Bluetooth and Bluetooth LE connectivity."

I must check that out. Reading the comments makes me want a retropie. The link is not working.

Can someone explain to me the room that the test photos are from? Is it sound proof? Why?

That is called an anechoic chamber [1]. Basically, the walls absorb all the outward radiation from the device so the test instruments do not have to sort out reflected waves. It mimics the device being outside with infinite space in all directions.

[1] https://en.wikipedia.org/wiki/Anechoic_chamber

Close - radio waves not sound waves.

Turns out that the wavelengths involved are similar, although one form of radiation is acoustic and the other electromagnetic. So those anechoic chambers often work pretty well for both purposes.

It's a Faraday cage (https://en.wikipedia.org/wiki/Faraday_cage), It keeps external radio signals out so they can check for EMI (https://en.wikipedia.org/wiki/Electromagnetic_interference)

This is partially incorrect. The objective is not only to isolate the device from external radiation but to avoid internal RF reflections within the chamber. That's why the geometry of the wedges, which are covered in a RF-absorbent material. A Faraday cage wouldn't help with this.

Anyone got a copy of the files? The FCC has pulled them down!

Please have a header for an on/off button.

The link seems no workie. FCC server problems.

And here i was just about to order a 2...

The onboard Wifi and Bluetooth really makes the barrier of entry for IoT $5 (assuming the stick with the $5 mark)

It's already $4 with the new ESP8266-EX from WeMos!

Full WiFi stack, 11x GPIO, Analog in, SPI, I2C, 80/160Mhz, 4MB Flash, power/flashing over USB, supported by the Arduino IDE, or just use the NodeMCU Lua firmware with modules for most sensors, displays, MQTT etc. I'm having a blast with them right now.

(Edit: http://www.wemos.cc/wiki/doku.php?id=en:d1_mini)

WTF is WeMos? ESP8266 module itself costs below $2

Ummm, I believe it's a board you can surface mount the ESP8266 on to.

The Raspberry Pi Model B has typically been around $35, not $5.

As the other poster mentioned, they've typically been $35. The barrier of entry for IoT is still pretty low though, you can get an ESP8266 for ~$2/ea. That'll handle wifi and can also be your application processor. For another ~$2-3 you could add on a BLE module too.

ESP32 is coming with both Wi-Fi and BLE: http://espressif.com/products/hardware/esp32/overview


This looks like a full board with ethernet and all the extras. I doubt it will be $5.

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