Raspberry Pi - Hardware Flaws and Fixups?

Pete: great to hear you are back from a well-earned vacation! Hope to hear more soon once you've rested up from the trip. 

I think as long as the date and the version are clearly marked on the schematic, that's all you should need to do for now.  A "wish list" for the next board spin is fine, but not as high a priority for those of us on the outside just now, in my opinion.

As you'll see from the RasPi forums, seems like a lot of issues which may be related to cheap USB cables having high resistance (would you believe up to 4 ohms? that would drop your 5V down to 3V, at 500 mA). Also a lot of SD card questions which in some cases have been pilot error (incorrect use of 'dd' utility) and sometimes apparently incompatible cards. And if the below claim is true, it certainly doesn't help finding a reliable card:

http://www.petapixel.com/2011/05/20/one-third-of-the-sandisk-memory-cards-on-earth-are-counterfeit/

Hi Hans, Yes still living in Delft. Yes I'm working on I2C / SPI right now, but have I done that before? Hmm.

Anway, looking at the chip on my RPI, I noticed that it had 64 pins instead of the 48 I was expecting. I apparently misread the datasheet. Happens every now and then. :-(

Re: SD cards:  I have a 4G kingston card, which apparently locks up every now and then. Things stop for 5 seconds and then get going again (the kernel prints something about a timeout).

I'm hoping that the fix by Russel King for this works.

Pete,

R34, the current limiting resistor for the power Led, is too small. It hurts my eyes. In the old days you'd use something like 220 or 330 ohms. Nowadays you'd take say 1k. (which it probably is). I decided to go for 10k on one of my designs, but didn't realize that there were 15k resistors in my storage box labeled 10k (which is short for 10-20k), So I soldered on 15k, and I like that....

Pete, are these resistors 0402 or 0603? I'll measure when I get a chance. Maybe I'll swap it out one day if it irritates me enough.... But most likely not before I can manage to have a spare RPI on hand.

The one on the yellow led has the same problem, but not as dramatic. 2k2 or 3k3 would do if it is now 1k.

Hi  All

Schematics - Yes we have had a discussion on this and there are a set in the release pipe and they reflect the current status of the delivered boards as of today.  Will post back here as soon as they are ready for download on the raspberrypi main site ( I know Liz and Eben are still running round like mad things).

The Pi board has a set of 4 hard links that can be read by the firmware so it knows which version of the board it has and can configure accordingly.  However, anyone who is doing something more than just software needs to know what may change (e.g. GPIO) so they can plan for it. We have a candidate list and this will be published soon but now separately from schematics.

Bright LEDs - They are, we will look at dimming them down but some people who are designing cases want to use light pipes of some sort and this makes them harder to see if they are not 'bright'.  Guess we could save some power?

Cables - we have not seen a problem with any of the ones we have been using - or reports from developers, BUT now we have a population of more than a few hundred I can see issues like this appearing.  We just need to work together to pass around knowledge.

More later - need to dash.

Pete

Roger,

  That's an interesting observation that the micro USB connector may have been

substituted for a cheaper surface-mount one.  Paul Maunders, with beta board #8,

commented on how rugged the power connector felt:

but GizmoB73, with a production road-test unit commented just the opposite.

Now that apparent conflict makes sense.

After the ethernet jack issue, it makes you wonder if any other substitutions were made.

FYI, I've powercycled my RPI about 100 times yesterday, but I always use the connector on the other side of the power cable. This way my RPI doesn't wear down, and I can start over with a new $3 USB-power-cable after the second insertion on the RPI....

I power cycle mine a lot too, and I really wouldn't want to put all that strain on the Pi's power connector either.  I always disconnect the power further down the line.

Pete, I guess the schematics would tell me this, but do you know if the Pi supports the type of high current connections that Ipod/Ipad do?  i.e. a resistive connection to the D+ and D- pins?  If so, there might not be so much of a voltage drop if we have current going through all four wire (albeit smaller on the D- and D+ wires).

Gizmo,

Those voltages on D+ and D- are not power-carrying.
Apple wanted to make a powersupply that provided 1A of charging current, but didn't want to be breaking any 500mA standard-USB power-sources. So they decided that 2.0V on D+ and 2.7V on D- will indicate that this is a 1A capable apple charger.

This is an easy and cheap way to identify the charger. It's just identification.

The D+ and D- lines on the Micro USB connector are not connected to anything on the RPI.

-------------------

Pete, Not sure if you've heard about this, but some people are a bit disappointed that I2S has been dropped. So those are planning on using those strapped pins as access points for the I2S hardware on the SOC.

If another "board revision" possibility presents itself, that might lead to less problems for those people.

I was thinking about: The GPU already knows its serial number/MAC address. So could that be used? You guys have 1M mac addresses, right? So if you reserve the top two bits for board revision, you're leaving some 150k mac addresses unused if the current "batch" of 100k boards are rev 0, and the next will be rev 1.

Or if that sounds wasteful, the mechanism for "personalizing" the SOC that is used for the MAC address can also be used for passing BOARDREV=XXX on the kernel commandline.

On the other hand, that would require a ROM firmware upgrade to the SOC, right? Is that really ROM? Flash? OTP?

Thanks for the update Pete. Can you provide some detail about what changes are required besides the HDMI configuration in firmware ?

Do these changes require a pcb layout redesign ? (Perhaps your opportunity to move that eletrolytic)

Also, does anybody has any piece of concrete information about how many boards have been actually manufactured and shipped out from the assembly house in China?

Thanks & Regards

I've seen a few reports now of power issues that could be cable or power supply, but have an interesting time dependence. I'm just curious what the tolerance limits are on the 5V 700 mA (?) polyfuse, and if there might have been an unauthorized substitution of some lower-spec part there also. The "voltage slowly creeping up" report sounds like it could be a resetting polyfuse to me.

"I left the RPi off and completely unplugged for about half an hour there (which it has been a few times already) and after nothing appearing to be going on again… it's back to working apparently at random."

... from http://www.raspberrypi.org/forum/troubleshooting/networkrandom-problem-with-pi-that-was-fine-yesterday#p66129

"Stuck with the Blackberry 700mah charger with just the keyboard/mouse in via powered hub but for some strange reason over the past hour the voltage has been slowly creeping up

Originally at idle it was 4.86v but it is currently sitting at 5.05v, when i plug the ethernet it it drops slightly and seems to work until i do anything involving data transfer or CPU usage (so basicly anything)"

... from http://www.raspberrypi.org/forum/troubleshooting/usb-bus-seems-to-be-shutting-down-after-30seconds/page-3

jbeale,

  reminds me that you never find just one cockroach or termite.

Liz said on the RPi forum on March 8 in response to a question about substitutions:

"I don’t think we’d necessarily have minded too much if the “alternative” had been something that worked!"

which could be interpreted as not being particularly vigilant about substitutions.