The problem with non-standard USB charging (was: Re: Price of the Freerunner spare parts)

[Michael Shiloh]

Correct me if I'm wrong, but I think the big problem is that there is no 
standard for reporting current capacity other than the USB protocol, 
which is not supported by dumb chargers.

What this means is that the various tricks that we and others play 
(shorting unused data pins, adding resistors of specific values between 
two pins, etc.) are all completely non-standard meaning there is no 
agreement between manufacturers and most importantly NO PROM
iSE THAT FUTURE DEVICES OF THE SAME MODEL NUMBER WILL USE THE SAME 
TECHNIQUE!

In other words we could test a given Motorola charger today, find that 
they have a resistor of value x between two pins, write the code to test 
for this value and then bump up the current draw to the value that we 
know the Motorola charger can provide.

But then next month Motorola might change the value of this resistor. 
They might send a firmware update over the air to all their phones to 
make existing phones compatible with the new charger.

We, of course, do not get this update, and suddenly the charger that we 
had tested no longer works.

For a good example of this problem see Lady Ada's discussion on her 
Minty Boost. You will see a string of reports from people saying they 
needed to modify the resistor value and/or location in order to get 
Minty Boost to work with different models of iPods. And any solution is 
always temporary, until Apple changes something and then the resistor 
must change again.

Thus the only way we can guarantee safe charging at higher than the USB 
maximum of 500mA is with our own charger.

I would love to see a technical report, a white paper, written on this 
issue. Would anyone like to take on the challenge? Winner gets an 
Openmoko t-shirt (http://wiki.openmoko.org/wiki/SWAG)

Michael

Kevin Dean wrote:
> On Thu, Mar 27, 2008 at 8:18 PM, joerg <joerg.twinklephone at gmx.de> wrote:
> 
>>  Alas there is *absolutely* *NO* way for NEO though to tell when your braindead
>>  850mA charger will start to slowly melt down overnight when you set 1A charge
>>  mode via applet. It's very unlikely there will ever be any trouble other than
>>  very rare occasional broken chargers, but we just can't guarantee it's safe,
>>  for nobody knows all the chargers in this world.
> 
> As a hacker poking around with prototype hardware, that's a fine
> answer. For a mass market device, that's suicide. I understand that
> the Neo can't make the determination, but in the long run I think that
> Openmoko MUST. It is physically possible to charge the Neo at it's
> full speed (because it happens with the USB charger) so therefore it
> must be physically POSSIBLE to have an AC or DC based charger that
> does the same thing without requiring a computer tether. If any
> charger on the market can do this I think it would be simple enough
> for Openmoko to "certify" that the MODEL (not each unit, of course) is
> compatible and, unless it malfunctions, won't burst into flame.
> 
> If there exists no such device on the market then perhaps that needs
> to be corrected either directly or indirectly. Put very simple "Car
> charging is a very big deal". Assuming the first (certification) or
> second (fabrication) options can't be met - drop the "fast charge" as
> a feature since anyone unable to fast charge in their car will
> consider their device "broken" at best or sold on a feature that
> doesn't exist.
> 
>>  However it's not our fault when sth really bad is happening, for every
>>  charger/host is supposed to be short-circuit safe (mustn't burn, otherwise
>>  probably mustn't be sold),
> 
> I don't think anybody expects a certification of safety from Openmoko
> (there are voluntary organizations like Underwriter's Laboratories
> here in the US that will certify safety) for a product to go to market
> and even if they DID nobody can rightly expect you'll certify that no
> device will ever malfunction.
> 
>> and it's the USER who has to think *before*
>>  enabling 1A mode. Hey guys, when connecting your 110V device to a 230V outlet
>>  in Europe, whom do you blame for the smoke?!
> 
> The "common user" mentality is one that has, IMO, bitten Free Software
> repeatedly. As a serious (and non-belittling) thought exercise, how
> many people here know what amperage their car charger pushes out? How
> many people here have wives or mothers or nephews that know that
> answer?
> 
> The average person buys a car charger because the box says "Compatible
> with <model>". Some might not even go that far since a commonly held
> assumption is that "if it fits in the hole, it will work". Yeah, kinda
> stupid, but it's also pretty reasonable to expect that this WILL
> happen. It's also reasonable to suspect that at least SOME people will
> take that and have it reflect poorly on Openmoko even if the charger
> was bought third party. It's sometimes easy to dismiss user stupidity
> when you're writing code that you use and the only signs you see are a
> ticking download counter or more hits in a server log, but this is
> something that I think could directly relate to the ability to sell
> SECOND units to people.
> 
>>  bottom line: USB is specified for 500mA at MAX, and even this needs
>>  intelligent host (alias charger). Everything beyond has to have some
>>  intelligence with it, either the one of the user not messing with chargemode,
>>  or the one of OM-designers creating and checking for a proprietary way to
>>  signal >500mA charger capabilities (=48k).
>>  (PS: For the record: If i got this right, the 48k even means "I'm a charger
>>  and I can do *2*(!)A", just GTA02 can't take more than 1A. This might change
>>  for GTA0x!)
>>
>>  HTH
>>  cheers
>>  your friendly engineer jOERG
> 
> _______________________________________________
> Openmoko community mailing list
> community at lists.openmoko.org
> http://lists.openmoko.org/mailman/listinfo/community