Joerg Reisenweber | 28 Aug 19:27 2008

Re: USB-OVP

Am Do  28. August 2008 schrieb Andy Green:
> Somebody in the thread at some point said:
> 
> | Charger Overshoot Voltage VCHGR_OVRSHT 6.0 V max.
> | -----
> |
> | Though we couldn't even attach a USB-charger compliant to this spec,
> as it has
> | to be equipped with micro-USB plug, we should expect to see voltages on
> | USB-receptacle of up to 6.0V AT LEAST.
> 
> Why?  Our charger doesn't follow that spec that mentions 6.0V.
> 
> | I think it's a strong argument for additional OVP extending our actual
> | abs.max.rating of 5.5V to limits safely beyond 6.0V, like we are actually
> | about to do for GTA03.
> 
> Regulator and cap in the charger should mitigate this.  Why don't we
> measure our actual charger performance in this case, shouldn't be too hard.

For one simple reason: USB-charging was meant to work with *any* USB-charger 
(otherwise we would have used ID-pin for VDD of charger ;-) and Neo wouldn't 
take any power from USB-VBUS, thus no issue with OVP), and for sure we don't 
want to restrict customers to usage of OM-wallwart-powersupply only. We 
wouldn't even get away with this, legally. As long as our device has a 
USB-receptacle, it has to cope with insertion of what a customer can 
reasonably assume should work when attached to it.

Then we also got the *wide* field of externally powered hubs. Ever had a look 
at the way those are providing power to downstream-connectors? And at the 
power-supplies that are (or are not!) shipped with these hubs.
"Dear customer! There are a plethora of powered USB-hubs out there. We 
strongly recommend NOT to attach Neo to any of those, as we didn't care to 
protect the device against voltages that may be found on a lot of the cheaper 
ones you might incidentally use. Please connect Neo to the original charger 
or certified good hosts only!" :-(
Do we really need more reasons to care about that issue?

cheers
jOERG
_______________________________________________
hardware mailing list
hardware@...
http://lists.openmoko.org/mailman/listinfo/hardware

Gmane