Raspberry Pi - Model A - Where are you ?

Despite the Model A not actually existing, I suspect that it played a huge role in creating the worldwide popularity of the Model B.  The $25 price tag was positively earth-shattering, the $35 not so much --- good, but not earth-shattering.  In fact, the Model B is  riding the PR wave created by the Model A.  I hope it's grateful for the tow.

For headless or standalone/mobile projects though, the Model A is quite likely to be the preferred board, especially now that wifi dongles can be obtained for far less than the difference between $35 and $25.  In fact the Pi is poorly suited to Ethernet communications anyway, since the Broadcom SoC doesn't provide Ethernet natively and so the external Ethernet controller chip hangs off USB which is not ideal.

It's a bit odd, but now that the dust has settled, it's pretty noticeable that various things about the Pi are pretty screwy.  It's using a less than ideal SoC, the power design is highly questionable, and the CSI/DSI connectors are a very bizarre carbuncle that shouldn't be part of the Pi design at all.

Still, we are where we are.  I sure hope that some of this gets fixed for the next significant redesign though.

Morgaine.

Despite the Model A not actually existing, I suspect that it played a huge role in creating the worldwide popularity of the Model B.  The $25 price tag was positively earth-shattering, the $35 not so much --- good, but not earth-shattering.  In fact, the Model B is  riding the PR wave created by the Model A.  I hope it's grateful for the tow.

For headless or standalone/mobile projects though, the Model A is quite likely to be the preferred board, especially now that wifi dongles can be obtained for far less than the difference between $35 and $25.  In fact the Pi is poorly suited to Ethernet communications anyway, since the Broadcom SoC doesn't provide Ethernet natively and so the external Ethernet controller chip hangs off USB which is not ideal.

It's a bit odd, but now that the dust has settled, it's pretty noticeable that various things about the Pi are pretty screwy.  It's using a less than ideal SoC, the power design is highly questionable, and the CSI/DSI connectors are a very bizarre carbuncle that shouldn't be part of the Pi design at all.

Still, we are where we are.  I sure hope that some of this gets fixed for the next significant redesign though.

Morgaine.

Morgaine Dinova wrote:

 

Despite the Model A not actually existing, I suspect that it played a huge role in creating the worldwide popularity of the Model B.  The $25 price tag was positively earth-shattering, the $35 not so much --- good, but not earth-shattering.  In fact, the Model B is  riding the PR wave created by the Model A.  I hope it's grateful for the tow.

 

For headless or standalone/mobile projects though, the Model A is quite likely to be the preferred board, especially now that wifi dongles can be obtained for far less than the difference between $35 and $25.  In fact the Pi is poorly suited to Ethernet communications anyway, since the Broadcom SoC doesn't provide Ethernet natively and so the external Ethernet controller chip hangs off USB which is not ideal.

 

It's a bit odd, but now that the dust has settled, it's pretty noticeable that various things about the Pi are pretty screwy.  It's using a less than ideal SoC, the power design is highly questionable, and the CSI/DSI connectors are a very bizarre carbuncle that shouldn't be part of the Pi design at all.

Another advantage of the Model A is significantly less power consumption.  I find that when Ethernet is hooked up, my RasPi can get pretty hot.  Now that I've mounted her vertically the air flow across both sides keeps temperature reasonable.  The LAN9512 data sheet says that when 100baseT and high-speed USB are in progress, the chip dissipates 763 mW (typ) or 231 mA (typ) of 3.3V.  This must be produced by 5V -> 3.3V linear regulator RG2, which then dissipates an additional 0.4 W.  So when you remove the LAN9512 and use an external hub and Ethernet dongle, you're removing over 1 W from the RasPi board.  Putting the Model A in a closed plastic case is a lot more feasible than the Model B.

Yes, I know there have been lots of claims about how the BCM2835 has been tested in ovens (I haven't seen an actual temperature range mentioned) and how cell phones don't have fans (I'm not a fan of cell phones either).  However, as far as I know nobody has published actual thermal test results for RasPi.  My own experience is that I've never had a problem with a board that ran too cool, and it's usually software guys who say it doesn't matter if a product runs hot.

Regarding CSI/DSI, I suspect that the RasPi designers have plans for those.  We've already seen cameras prototyped.  I bet working on these add-ons is a lot more fun than the drudgery of getting the power subsystem user-proof.

John Beetem wrote:

 

Regarding CSI/DSI, I suspect that the RasPi designers have plans for those.  We've already seen cameras prototyped.  I bet working on these add-ons is a lot more fun than the drudgery of getting the power subsystem user-proof.

Indeed they are, but it's a case of the tail wagging the dog.  They're designing camera and display boards for Pi because the BCM2835 provides that functionality,, not because camera input and LCD display are educational goals of the project.

It really questions the ulterior motives --- a board design that omitted the CSI/DSI connectors would have been usefully cheaper and easier to lay out, as well as providing more hardware interface options with educational merit.

Just imagine the possibilities ... drop the CSI/DSI connectors, add proper power management.

I think the RPI as a robot-controller with a "vision system" and "display" would be very educational. Especially of both these addons could be bought for about $10 or $15. I think that's what they have in mind.

It's not like there's a shortage of USB webcams to achieve roughly the same thing though, and at prices that are plummeting and specs that are rising with each passing year.  I doubt the Foundation is set up to compete with Logitch and unbranded Chinese manufacturers in that niche.

In any case, even if image processing were held to be important for education (which I haven't heard the Foundation suggest yet although perhaps they might try that defence some day), the BOM and layout cost of the connectors and board area shouldn't be born by everyone.  I very much doubt that a significant proportion of Pi users are going to be doing image processing.  Leaving such devices to be attached by USB would have been more cost-strategic for the Pi community as a whole.

I just think it's so sad that to have proper power management was regarded as negotiable, whereas the CSI/DSI connectors were not.  Something is badly wrong there.

Morgaine.

Morgaine Dinova wrote:

 

In any case, even if image processing were held to be important for education (which I haven't heard the Foundation suggest yet although perhaps they might try that defence some day), the BOM and layout cost of the connectors and board area shouldn't be born by everyone.  I very much doubt that a significant proportion of Pi users are going to be doing image processing.  Leaving such devices to be attached by USB would have been more cost-strategic for the Pi community as a whole.

My guess is that RPF wants to leave open the possibility of a RasPi Tablet with CSI/DSI providing direct interfaces to camera and display.  Regarding cost, my understanding is that the CSI/DSI, GPIO, and JTAG connectors weren't supposed to be populated in production boards.  They were left on the production BOM by mistake and taking them off wouldn't affect BOM cost significantly.

John Beetem wrote:

 

Regarding cost, my understanding is that the CSI/DSI, GPIO, and JTAG connectors weren't supposed to be populated in production boards.  They were left on the production BOM by mistake and taking them off wouldn't affect BOM cost significantly.

I've heard that said about the GPIO header only, not about the rest of the items you mention.  Interesting.

But the point still stands.  The cost of that decision is being born by everyone, because so much more could have been crammed into the board if CSI and DSI were omitted from the layout and replaced by something more generally useful ... like proper power management, say.