USB-OVP

[Joerg Reisenweber]

Am Fr  29. August 2008 schrieb Andy Green:
> Somebody in the thread at some point said:
> | Joerg Reisenweber wrote:
> |> Do we really need more reasons to care about that issue?
> |
> | How about Postel's law ?
> |
> | http://en.wikipedia.org/wiki/Robustness_Principle
> |
> | Sounds like a good enough rule to follow at any interface, not just
> | software interfaces.
> 
> ''Be conservative in what you do, be liberal in what you accept from
> others'' is a nice general principle.
> 
> But then where does one stop, accepting mains power directly, automotive
> 12V, 12V charger connection, broken chargers...  
I gave some sort of spec in start of this thread, to give you an idea where 
*not* to stop, and this is 5.5V abs.max.rating of PMU. But seems you are 
considering to better not start at all on improving this situation, as *you* 
seem to fear *you* don't see a point to stop at. Whereas we started to think 
about it and when we see we landed at 25V OVP with our solution at 
(virtually) no additional costs, *we* all knew we did it and it's much better 
now.

> it is a guiding 
> principle to be taken into account at specification time.  Every new
> thing we accept to support has costs.
False. Obviously, evidentally *not*, as you notice yourself in last line of 
your post.

> 
> What I like though is the "Be conservative in what you do" part of it.
> Because we have not seen failures in the field from GTA02 arrangements,
False. You don't stop to allege this, but actually there *are* some devices 
with obviously broken PMU... -->

> I am having a hard time accepting we need to change anything from proven
> GTA02 situation.
... so there is no such thing as a proven (good) GTA02 situation. Only proven 
thing is even USB charger specs allow for voltage transients we can not 
handle.

> If someone can actually show that GTA02 style 
> arrangement leads to product failure in normal circumstances then it
> makes it clear we need to do better.  But it seems thousands of users
> are proving it's robust enough already.
False. And false. See above. The possibility to connect random power supplies 
to powered hubs is proof enough, it's self-evident.



> Why add reverse voltage 
> protection when nobody seems to have reversed the voltage on their USB
> connector to date?
False. I read a report of a user thanking the hw-EE for designing such a 
robust device, as he actually applied voltage reverse and device survived. 
Dunno why it did, you are free to ask Wolfgang for 100 devices to do a mass 
test and prove we don't need additional protection against that. But wait, I 
think 100 broken devices are more expensive than decent reverse protection 
for 250.000 to 500.000 devices, and even the time I spent for these messages 
is worth protection for another several 1000 probably. So why not just let's 
do it?

> 
> Still one part of what Joerg has been talking about he characterized as
> different component selection so it was more robust, that sounds fine if
> it is not adding expense / size.
Though you were bitching at it from very beginning, and it never would have 
happened if we listened to you. And, strange enough, all of your posts sound 
like you never had a chance to look at the GTA03 schematics and changes in 
there by yourself. :-/
Was the original USB-design yours, so you feel you have to fight for it with 
claws and teeth, despite it has ABS.MAX(!!!).ratings obviously much too close 
to normal operating conditions? That's clearly a faulty design, no EE would 
feel like "good enough" with it, absolutely no matter whether it already 
shows breakage in the field, or just is so close to the edge that it's easy 
to see it very likely will eventually.
BTW: please do the exercise and apply all your basic objections raised in this 
case to your suggested uC. ;-)

cheers
jOERG

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 194 bytes
Desc: This is a digitally signed message part.
Url : http://lists.openmoko.org/pipermail/hardware/attachments/20080829/deb936bd/attachment.pgp