[Arm-netbook] power consumption?
luke.leighton at gmail.com
Sat Jan 14 01:58:56 GMT 2012
On Sat, Jan 14, 2012 at 1:36 AM, Bari Ari <bari at onelabs.com> wrote:
> On 01/13/2012 07:10 PM, lkcl luke wrote:
>> hmmm... it might be worthwhile deliberately having the case at 8mm
>> (Type III PCMCIA) so that it doesn't fit into the smaller sockets.
>> that would make it possible to still fit the smaller cards into the
>> (larger) sockets. other remaining issue: power. PCMCIA's 68-pin
>> connector pins are limited to 0.5A each. if we made one of them
>> negotiable up to 12v (after power-up and reading the I2C EEPROM), that
>> would give almost 8 watts (12*0.5 + 5*0.5). do you think that would be
> I started looking at the power pins more closely today. There are a few
> more grounds than V+ currently. Part of this is for accommodating
> separate analog and digital grounds.
initally i added them more for separating the high-frequency signals
from each other. i'm taking a bit of a flyer by having the SATA pins
at the end, figured there'd be a shield nearby. also i used the two
5V lines to separate I2C from GPIO, on the perhaps naive assumption
that a) I2C is slow and wouldn't result in cross-talk b) GPIOs 0 and 1
would also be low-speed.
>We could swap a ground for 5V or 12V.
would that affect the stability of that power input? if not i'd
recommend either pin 20 or pin 54. or both.
> The GPIO's could also be configured at plugin to have a few extra 5V or
> 12V pins by reading an eeprom or pinstraps on the card.
oh duh of course, forgot about that. hmmm...
> The larger case for Type III would keep them from getting plugged into
> the smaller 5mm thick slots will work here again for EOMA68.
> Type II cards may work in either Type II or Type III motherboards or
> Type III motherboards or devices could default to Type II mode at
[initially] not sure why you'd do that... [but i do now, below].
> Type III cards won't fit in Type II devices so that problem is solved.
> Type III devices could be detected by the Type III mainboard or device
> at reset/power-on by a pinstrap or eeprom on the module. The mainboard
> could have a cheap 8 bit micro to read the pinstrap or eeprom and
> configure a couple GPIO for use as power vs GPIO.
ah. right, ok now i see what you're on about [above]
of course. darn. that's where i'd prefer it wasn't optional, and we
picked either pin 20 or pin 54 or perhaps both to be 5V
(non-optionally), and that was the end of it.
if it was both, that would make for 10 watts. which is pretty damn good.
btw i'm not keen on 12V as part of the standard, *unless* you can
show me that absolutely all EOMA68 CPU cards at the 8mm height will,
without fail, require 12V power.
i only suggested it as an idea to get extra wattage (and then
down-convert) but if you think we could get away with converting those
2 pins to 5V power that's much preferable, i think.
i'm not a great fan of "optionitis" in hardware, because it costs money :)
More information about the arm-netbook