Wifi / let's have RF stuff powered with high current source all the time because...

Andy Green andy at openmoko.com
Mon Aug 25 16:17:12 CEST 2008

Hash: SHA1

Somebody in the thread at some point said:
| On Wed, Aug 20, 2008 at 09:39:14AM +0100, Andy Green wrote:
|> IMO it doesn't make any harm to see MOSFET switches on the power rails
|> to whole GSM side so there is no question.
| no, it actually causes problems

I heard from Tony having a current limit on the switch caused problems,
nothing else.

|> This was done on a previous prototype I saw in .tw, but the low Rdson
|> they used had built-in currently limit (1.6A IIRC?) that got
triggered at low
|> temperature and highest TX setting destroying TX action.  So they
removed it
|> from PA side... maybe we can bring this back in the future.
| to the best of my knowledge it is industry standard practise to have
the GSM PA
| cunnected directly to the battery. the RdsON of the mosfet switches
(even of
| those optimized for very low RdsON) is not constant but current-dependent.
| This causes spectral impurity and can easily have (and actually has in
| practise) effects on violating the regulatory requirements on the
| strict regulations on GSM/GPRS/3G signals.

Sounds like problems with the PA design, but it is hard to argue about
since it's hearsay.

| So I really vote strongly against tryign to do this for no apparent

There's a very good obvious reason to stop a wideband RF amp from having
autonomous power supply even when the device is "off".  It's such a
bizarre design that doing it is the thing that needs the special
pleading, not the insertion of control over its power which is normal
design action.

| The GSM PA is an analog circuit without any digital logic.  Thus, it
can never
| lock up internally.  It has an input pin that switches it off/on
depending on
| the logic state, i.e. a built-in power switch.  This is what should be
| (and is used in GTA02) to ensure acutal power-off.

No I disagree -- removing the power ensures "actual power off".
Everything else is something different than "ensure actual power off"
but "ensure powered but allegedly inactive while we keep that enable
level right".

- -Andy
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org


More information about the hardware mailing list