It is worth noting that they are an educational charity, not a for-profit organisation.
I think it is great that they have released this - and I think that Broadcom also deserves credit for allowing them to release a schematic giving away the pin-out of the BCM2835 SoC - it makes the Raspberry Pi so much more useful to know what the outputs of the system are.
This opens up the possibility of trying to work out how to use the unused pins on the BCM2835 based on the names. I haven't been able to work out what the HD pin group does (probably not SATA or PATA based on the number of pins, and HDMI has its own pin group). There seems to be a CAM0 not connected to anything - given CAM1 seems to be a camera, it looks like it is for an additional camera. It also has CCP2 outputs for a CCP2 camera. I also haven't been able to work out what SLIM is - but given it has two pins, CLK and DAT, I guess it is some kind of serial data protocol.
Although they look like they are in the middle of the BGA ball array, and probably on a pad that is just floating, so actually connecting anything to the pins that the board isn't designed to allow access to is probably effectively impossible.
Wow... 17 unattached GPIO pins off the CPU. I wonder if they just ran out of room to put them on the board, or if they have lack alternate pin-specific capabilities that the used pins have.
I kind of wish there was a harder to use card edge connector for these - they could likely implement an additional 2 SD card slots or similar I/O.
Man...two unused pins on the external LVDS display header. They could have thrown an I2C pair on that and made it really easy to add an external touchscreen or other kind of display device. Shame.
is there anybody specialized in circuit design? i mean, how hard is it to build a cheap Pi equivalent board? If this is so popular, why no commercial version?
Make no mistake, clones are on the way. For example this device: http://rhombus-tech.net/allwinner_a10/ is supposedly both cheaper and more powerful than the Pi.
The Pi has kicked off some interest in hardware hacking for myself too. My question is how hard would it be to embed an 802.11 radio into the same board for a discreet sensor project
I haven't looked at the schematics yet, but generally a SoC based design like this needs at least a six layer board, which combined with the speeds it runs at is pretty advanced electronic design...
Coupled with that, it can be really hard to get low volumes of some of the parts they use, like the SoC, PMIC (power management IC) and so on (although you can get free samples if you ask sometimes - but they generally won't give them to you unless there's a chance you'll put through a big order with them in the future). And soldering the BGA parts onto the board would be a pain unless you pay for it to be assembled for you...
Unless you know a lot about that sort of thing, it would probably be more successful to design something that interfaces with a Pi.
My university had a wonderful pair of courses in the first year of its CS degree. The first had us build a Z80-based computer on breadboards, with a couple of hulking ancient HP logic analyzers to debug with. For the second, they gave us an assembled Z80 single board computer and had us do something interesting with it -- games displayed on an oscilloscope, control a small robot, etc. For both courses compilers were strictly forbidden.
It was a great way of bridging the abstraction gap between the physical hardware and the code we typed at the keyboard. (We were also required to take courses in basic digital and analogue design and compiler construction to give an overview the rest of the stack, too.)
Nowadays, I think they've switched to using ARM boards for the second course in the build-a-computer series. From a hardware reliability and cost perspective it makes sense, but it's a little sad that the "I built one like this! Just not quite so neatly..." factor has been lost.
As a software guy, the details of the Pi's PCB layout are about as impenetrable to me as the internals of the SoC itself are. It's Magic Electronic Stuff!, but just a tad more powerful than the Z80 I used to play with.
> how hard is it to build a cheap Pi equivalent board?
Cheaper than the RaspberryPi? extremely hard. While a body more experienced than the RaspberryPi with more manufacturing clout can probably do that, the RaspberryPi foundation is a non-profit. Roku/Apple might be able to build a RaspberryPi for 75% the price, but their own profit would make it more expensive.
There are equivalently-or-better specced stuff on the market (Roku boxes that run Linux that can occasionally be found for $50, Apple TV that can occasionally be found for $80). But nothing at the $25/$35 price point with similar specs - and I don't expect any in the near future.
Cheaper than a "real" Pi? Does $35 even get you an Arduino with an SD card and an ethernet port? _Maybe_ someone'll clone these, but surely these not much "fat" in the price to be undercut the "official" ones by enough to make it worthwhile?
The foundation didn't think the device would be so popular. They are ramping up production, so that should help with getting products out there.
As for equivalents: It's a 6 layer PCB, with BGA and other SM components. That's really hard to do at home. So you're unlikely to see hobbyist rip-offs anytime soon.
At some point there might be another similar device - mass made in China, more power, more ram, but probably much worse documentation. There's certainly a big market for tiny dirt cheap computers.
17 comments
[ 3.0 ms ] story [ 46.6 ms ] threadI think it is great that they have released this - and I think that Broadcom also deserves credit for allowing them to release a schematic giving away the pin-out of the BCM2835 SoC - it makes the Raspberry Pi so much more useful to know what the outputs of the system are.
This opens up the possibility of trying to work out how to use the unused pins on the BCM2835 based on the names. I haven't been able to work out what the HD pin group does (probably not SATA or PATA based on the number of pins, and HDMI has its own pin group). There seems to be a CAM0 not connected to anything - given CAM1 seems to be a camera, it looks like it is for an additional camera. It also has CCP2 outputs for a CCP2 camera. I also haven't been able to work out what SLIM is - but given it has two pins, CLK and DAT, I guess it is some kind of serial data protocol.
http://pandaboard.org/node/223/
I kind of wish there was a harder to use card edge connector for these - they could likely implement an additional 2 SD card slots or similar I/O.
If you're trying to duplicate the RPi exactly, you won't be able to. Nobody else can get the BCM2835 chip except the RPi guys.
Coupled with that, it can be really hard to get low volumes of some of the parts they use, like the SoC, PMIC (power management IC) and so on (although you can get free samples if you ask sometimes - but they generally won't give them to you unless there's a chance you'll put through a big order with them in the future). And soldering the BGA parts onto the board would be a pain unless you pay for it to be assembled for you...
Unless you know a lot about that sort of thing, it would probably be more successful to design something that interfaces with a Pi.
It was a great way of bridging the abstraction gap between the physical hardware and the code we typed at the keyboard. (We were also required to take courses in basic digital and analogue design and compiler construction to give an overview the rest of the stack, too.)
Nowadays, I think they've switched to using ARM boards for the second course in the build-a-computer series. From a hardware reliability and cost perspective it makes sense, but it's a little sad that the "I built one like this! Just not quite so neatly..." factor has been lost.
As a software guy, the details of the Pi's PCB layout are about as impenetrable to me as the internals of the SoC itself are. It's Magic Electronic Stuff!, but just a tad more powerful than the Z80 I used to play with.
Cheaper than the RaspberryPi? extremely hard. While a body more experienced than the RaspberryPi with more manufacturing clout can probably do that, the RaspberryPi foundation is a non-profit. Roku/Apple might be able to build a RaspberryPi for 75% the price, but their own profit would make it more expensive.
There are equivalently-or-better specced stuff on the market (Roku boxes that run Linux that can occasionally be found for $50, Apple TV that can occasionally be found for $80). But nothing at the $25/$35 price point with similar specs - and I don't expect any in the near future.
As for equivalents: It's a 6 layer PCB, with BGA and other SM components. That's really hard to do at home. So you're unlikely to see hobbyist rip-offs anytime soon.
At some point there might be another similar device - mass made in China, more power, more ram, but probably much worse documentation. There's certainly a big market for tiny dirt cheap computers.