Possible Hardware fixes

Ohh you right I missed that one.

Cost and the fixation on the credit card size form factor are very lame design guidelines, It is not bad to have an approximate target size, but they didn't have any mechanical restrictions or requirements, so component and connector layout could have been much better.

-J

Why do you think it will save 20% CPU, you are still using USB and it needs to be polled. The LAN9512 acts as a USB hub so even if you go external is the same thing.

-J

you wouldn't be going through the lan chip at all, because you're

not using the RPi's USB or ethernet ports, you're using the SD interface.

I'm not sure that the keyboard/mouse would work, since the SD interface

is expecting a file system.

My area of head-scratching:  how to turn the SD card into a network boot.

Are you prepared to keep an sd card installed with some minimal code on it ?  If so then perhaps u-boot. There's been some efforts to port it to the Pi and as there appears to be somewhat functional usb and network drivers it's possible to have u-boot load from the sd card then pull kernel etc from something on the network.

There are some adapters that convert SD into USB, such as this one:

Won't the problem with those be that you'll need to emulate an SD memory card ?   You should certainly be able to hook up an SDIO network device, but what would the on-SoC boot code make of that ?

Ohh, I see you mean using that SD<->USB converter not going directly to the USB pins (sort of model A) of the SoC.

But how efficient is that thing ? is it available ?, can't find any similar to that one

Also it would have been nicer to have more of the GPIO pins available somewhere, even on a non populated 2mm header.

-J

>Won't the problem with those be that you'll need to emulate an SD memory card ?   You should certainly be able to hook up an SDIO network device, but what would the on-SoC boot code make of that ?

the adapter allows a USB stick to emulate a SD memory card.  But I don't know

how much further you can go beyond that, by plugging a USB to ethernet adapter

into the USB port on the SD to USB adapter.

selsinork wrote:

 

I'm thinking seriously if after all there is enough interest and momentum to start a true "open source hardware" development.for a Pi like board, with parts that are documented.

I could certainly be interested, but I see two stumbling blocks. First is that anything with a GPU is unlikely to have good enough docs, second is that anything without a GPU seems unlikely to have the sort of mass market appeal to get the cost down to something reasonable.

 

The rhombus-tech guys seem to be a reasonable way along this road already, with some interesting info like the cpus they've evaluated: http://rhombus-tech.net/evaluated_cpus/

However they seem to be a long way from an actual released product and I'm not at all convinced of the EOMA-68 form factor. Although I do like the idea of a set of standard interfaces.

This is extremely hard to do without corporate sponsorship.  You have to build a lot of boards before they become cost-effective.  Sure, at $89 BeagleBone sounds like a lot of money, but I expect it's quite reasonable given the quantities sold.  If more people get behind BeagleBone, the costs would come down to something closer to RasPi.

Then you have the BeagleBone "capes".  They're pricey, but again it's because of the quanties manufactured.  You really cannot make anything with high-density SMT without specialized equipment and know-how, so it's almost always cheaper to contract it out.  However, the fixed costs of doing so are quite high so you need to pretty large quantity to get something reasonable.  Custom bare PC boards are pretty cheap now, but populating them with BGAs and QFNs isn't.

So my recommendations would be:

1.  Support the most open existing boards, e.g., BeagleBone.

2.  Take existing tablets and make the software open, like Make Play Live's Vivaldi and ZaReason's ZaTab.

3.  Take a look at Xilinx Zynq.  I don't know if that has a complete technical reference yet, but there's no reason it shouldn't completely document the ARM A9s and its hard peripherals.  And it's technically possible to design a completely-open GPU using the FPGA fabric.  Yes, a big job but so is reverse-engineering closed GPUs.  And yes, the FPGA image is (I assume) closed as well, but you can design at the Verilog or EDIF level.

It's really too bad that the Zed Board is so expensive.  I could afford one, but it's unlikely to get a mass following at the current price so my work wouldn't find wide distribution.

Sure, at $89 BeagleBone sounds like a lot of money, but I expect it's quite reasonable given the quantities sold.  If more people get behind BeagleBone, the costs would come down to something closer to RasPi.

Looking at BeagleBone on uk.farnell.com the other day, out of stock. Approx 300 available for delivery sometime between now and 2nd week in August.

Compare to the 12000+ available for delivery on 23rd July that the Pi was showing a few days ago and the difference in scale is clear.

You have to give the RPF credit for managing to get the interest and backing to be able to ramp up to where they are in such a short time. They seem to be the only ones to have managed it so far. As everyone keeps pointing out there's plenty of other boards out there, but none seem to be gaining the necessary momentum.

How much of the success of the R-Pi is down to aiming low to begin with and praying that the interest will be there ?

>is it available ?, can't find any similar to that one

I've looked, and strangely, I can't find anything like it,

although quite a few people seem to want such a thing.

The closest I've found is SD to WiFi adapters, such as

eye-fi, but I'm not sure if special drivers would be needed.

http://www.eye.fi/how-it-works/basics

See also the Toshiba FlashAir wireless LAN SD card

http://www.engadget.com/2012/01/11/toshiba-flashair-hands-on/

said to be the first card implementing the SD association's

Wireless LAN Standard.