GSM power management improvements

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

GSM power management improvements

Mychaela Falconia
Hello what's left of Openmoko community,

If there is anyone still using their FreeRunner (or GTA01) as a phone
with working GSM (perhaps on a private island whose owner likes GSM
and is committed to keeping it forever), there is a modem firmware
update which you might find interesting:

ftp://ftp.freecalypso.org/pub/GSM/GTA02/gsm-fw/moko-new-fw-20190128.tar.bz2

The primary diff from previous versions consists of a couple of sleep
mode improvements, i.e., improved ability to go into sleep modes which
draw less power from the battery:

* During those time windows in which the modem is disallowed from going
into deep sleep by the UART activity timer (10 s after each transmission
from the AP host to the modem on the AT command UART), previous fw
versions needlessly suppressed big sleep in addition to deep sleep,
allowing only small sleep.  The present version goes into big sleep
during these time windows, saving more power.

* Some Openmoko devices suffer from a hardware defect that requires
disabling deep sleep - the infamous bug #1024.  Previous fw versions
did not provide a sleep mode configuration that allows big and small
sleep, but not deep sleep, forcing the user to choose between small
sleep only or big sleep only on no-deep-sleep hardware.  OM AP software
distros have been using the big sleep only AT%SLEEP=2 config in these
circumstances.  Our new fw offers a new AT%SLEEP=5 option that allows
big and small sleep, but not deep sleep, which should be ideal for
deep-sleep-deprived OM hardware.  The default is still AT%SLEEP=4
allowing all 3 sleep modes (small, big and deep sleep), just like
before.

* Most of the Calypso chip's GPIO and multifunction pins are unused
and unconnected on Openmoko devices.  All fw versions for this modem
released prior to 2017, including all legacy mokoN versions produced
by the now-defunct original manufacturer of the hw, contain a bug in
this regard: they configure many of these unused and unconnected GPIO
and multifunction pins as inputs, causing them to float.  The general
dictum in digital hw design is that CMOS floating inputs are bad, and
they can sometimes cause increased power draw as a result of current
flowing through both transistors of the CMOS input structure when they
are partially open.  This bug has been fixed in the newer FreeCalypso
fw releases since 2017: the correct way to handle unused and unconnected
GPIO and multifunction pins is to configure them as dummy outputs with
a constant value, which is what our current fw does.

This new firmware release is brought to you by Falconia Partners LLC,
a manufacturer of new cellular modems under the FreeCalypso brand,
currently for GSM/2G but perhaps some day for UMTS/3G as well.  At the
present time we don't do any work specifically for legacy Openmoko
devices (there is no business case for it), but our own modem product
is very similar to OM's, thus whenever we make significant improvements
to the firmware for our own modem hw, it costs us nothing to compile
and put out a new fw image for Openmoko's old modem as well.

Please note, however, that even though our new FreeCalypso modem hw is
very similar to OM's old modem, it is not identical, and the firmware
images built for the two respective targets are NOT interchangeable!
We build all of them from the same source tree, thus functional
improvements made for one target automatically benefit the others as
well, but a few hardware configuration and external interface bits are
conditionally compiled.  If you take a firmware image built for our
FCDEV3B and flash it into an Openmoko device, it may corrupt your FFS,
so don't do it - please be sure to only flash fw images which were
built specifically for whichever hardware you have.

Happy 2019, and enjoy improved power management in the modem if you
still have a free-running Openmoko device.

M~

P.S. In case anyone is concerned about the legality and safety of
using new GSM products from Falconia Partners LLC, whether our new hw
or our new fw for legacy hw from OM, please note that the products in
question (both hw and fw) have been extensively used on public
commercial GSM networks in several countries (at least USA, Canada,
Austria, France and South Africa to my knowledge, plus maybe others I
don't know about) over the course of many years now, without a single
problematic incident anywhere ever.  Getting an official stamp of
regulatory approval is definitely in the plans, but no one has paid
for it yet.  In the meantime, despite having no official rubber stamp
that says so, we already know with almost 100% certainty that our
products (both hw and fw) function 100% correctly on the air, fully
compliant with all of the relevant technical standards.

_______________________________________________
Openmoko community mailing list
[hidden email]
http://lists.openmoko.org/mailman/listinfo/community
Reply | Threaded
Open this post in threaded view
|

Re: GSM power management improvements

Joerg Reisenweber
On Tuesday, 5 February 2019 02:27:10 CET Mychaela Falconia wrote:

> Hello what's left of Openmoko community,
>
> If there is anyone still using their FreeRunner (or GTA01) as a phone
> with working GSM (perhaps on a private island whose owner likes GSM
> and is committed to keeping it forever), there is a modem firmware
> update which you might find interesting:
>
> ftp://ftp.freecalypso.org/pub/GSM/GTA02/gsm-fw/moko-new-fw-20190128.tar.bz2
>
> The primary diff from previous versions consists of a couple of sleep
> mode improvements, i.e., improved ability to go into sleep modes which
> draw less power from the battery:
>
> * During those time windows in which the modem is disallowed from going
> into deep sleep by the UART activity timer (10 s after each transmission
> from the AP host to the modem on the AT command UART), previous fw
> versions needlessly suppressed big sleep in addition to deep sleep,
> allowing only small sleep.  The present version goes into big sleep
> during these time windows, saving more power.
>
> * Some Openmoko devices suffer from a hardware defect that requires
> disabling deep sleep - the infamous bug #1024.  Previous fw versions
> did not provide a sleep mode configuration that allows big and small
> sleep, but not deep sleep, forcing the user to choose between small
> sleep only or big sleep only on no-deep-sleep hardware.  OM AP software
> distros have been using the big sleep only AT%SLEEP=2 config in these
> circumstances.  Our new fw offers a new AT%SLEEP=5 option that allows
> big and small sleep, but not deep sleep, which should be ideal for
> deep-sleep-deprived OM hardware.  The default is still AT%SLEEP=4
> allowing all 3 sleep modes (small, big and deep sleep), just like
> before.
>
> * Most of the Calypso chip's GPIO and multifunction pins are unused
> and unconnected on Openmoko devices.  All fw versions for this modem
> released prior to 2017, including all legacy mokoN versions produced
> by the now-defunct original manufacturer of the hw, contain a bug in
> this regard: they configure many of these unused and unconnected GPIO
> and multifunction pins as inputs, causing them to float.  The general
> dictum in digital hw design is that CMOS floating inputs are bad, and
> they can sometimes cause increased power draw as a result of current
> flowing through both transistors of the CMOS input structure when they
> are partially open.  This bug has been fixed in the newer FreeCalypso
> fw releases since 2017: the correct way to handle unused and unconnected
> GPIO and multifunction pins is to configure them as dummy outputs with
> a constant value, which is what our current fw does.
>
> This new firmware release is brought to you by Falconia Partners LLC,
> a manufacturer of new cellular modems under the FreeCalypso brand,
> currently for GSM/2G but perhaps some day for UMTS/3G as well.  At the
> present time we don't do any work specifically for legacy Openmoko
> devices (there is no business case for it), but our own modem product
> is very similar to OM's, thus whenever we make significant improvements
> to the firmware for our own modem hw, it costs us nothing to compile
> and put out a new fw image for Openmoko's old modem as well.
>
> Please note, however, that even though our new FreeCalypso modem hw is
> very similar to OM's old modem, it is not identical, and the firmware
> images built for the two respective targets are NOT interchangeable!
> We build all of them from the same source tree, thus functional
> improvements made for one target automatically benefit the others as
> well, but a few hardware configuration and external interface bits are
> conditionally compiled.  If you take a firmware image built for our
> FCDEV3B and flash it into an Openmoko device, it may corrupt your FFS,
> so don't do it - please be sure to only flash fw images which were
> built specifically for whichever hardware you have.
>
> Happy 2019, and enjoy improved power management in the modem if you
> still have a free-running Openmoko device.
>
> M~
>
> P.S. In case anyone is concerned about the legality and safety of
> using new GSM products from Falconia Partners LLC, whether our new hw
> or our new fw for legacy hw from OM, please note that the products in
> question (both hw and fw) have been extensively used on public
> commercial GSM networks in several countries (at least USA, Canada,
> Austria, France and South Africa to my knowledge, plus maybe others I
> don't know about) over the course of many years now, without a single
> problematic incident anywhere ever.  Getting an official stamp of
> regulatory approval is definitely in the plans, but no one has paid
> for it yet.  In the meantime, despite having no official rubber stamp
> that says so, we already know with almost 100% certainty that our
> products (both hw and fw) function 100% correctly on the air, fully
> compliant with all of the relevant technical standards.
>
> _______________________________________________
> Openmoko community mailing list
> [hidden email]
> http://lists.openmoko.org/mailman/listinfo/community


Reviewed and considered 100% plausible and without any obvious issues.
Not tested on real hardware though by me.

Many thanks for providing this fine bugfix / improvement firmware

cheers
jOERG
(OM senior EE, hw & synergy)



_______________________________________________
Openmoko community mailing list
[hidden email]
http://lists.openmoko.org/mailman/listinfo/community