joerg at openmoko.org
Tue Aug 5 15:29:21 CEST 2008
Am Di 5. August 2008 schrieb Ole Kliemann:
> On Tue, Aug 05, 2008 at 07:49:51PM +0800, Joerg Reisenweber wrote:
> > Am Di 5. August 2008 schrieb Jeffrey Ratcliffe:
> > > 2008/8/5 Andy Green <andy at openmoko.com>:
> > > > Like Ole says when I started looking at it, by calling a landline in
> > > > same room and listening to its receiver lying on the desk, it varied
> > > > tremendously and not in a repeatable way. Eg, it appeared to vary by
> > > > orientation of the phone, but when I traced the path backwards, the
> > > > did not return.
> > >
> > > OK. Confirming this - the buzzing seems to come and go in waves of
> > > varying frequency. Even without touching the phone, the buzzing just
> > > comes and goes.
> > I noticed the buzz come and go in discrete steps. I guess that's
> > sending PCL commands to mobile, to level up/down the TX-power.
> > BS decides on this depending on signal-quality of MS as BS sees it.
> > It would be *very* helpful to confirm this, by using some RF-meter (e.g.
> > microwave leakage tester?), and/or reading the battery current, while
> > observing the noise come and go-
> I can only observe: I did this test some weeks ago at my home. The FR
> has signaficantly better signal when held out of the window.
> When holding out of the window during a call, it takes a short moment
> then buzzing disappears quite suddenly. Going back into the room again
> after a moment the buzzing returns suddenly.
> So it comes and goes in discrete steps and it lags behind the change of
That's exactly the behaviour we would expect when noise is related to
MS-TXPower controlled by a PCL command from BS. BS isn't responding to any
momentary change in signal-strength of inbound MS-signal, instead BS does
some averaging on signal-quality and then decides to send a PCL to MS
There are a maximum of 32 PCL steps, actually not all of them are valid for
type of FR GSM-chipset.
Remark: FR immediately tore down the connection, when I sent a PCL command to
level up to an (invalid?) high powerlevel (PCL0. On "CMU200 universal radio
communication tester", kind of a test-BS). But aside from that we still try
to reproduce the noise under controlled test environment. Strange enough, so
far there seems to be no noise when on CMU200. Anyway tests are not completed
So it's very important to get as much detail as possible on situations where
noise actually is seen. Such as battery current, GSM-signal level, frequency
Quotes from R&S CMU200 manual:
Power control level (PCL)
Dynamic power control is used in GSM networks to reduce
the output power of the mobile station as far as possible. In
practice the base station sets the mobile power on a dimen-
sionless scale of power control levels (PCL) ranging from 0 to
31. In GSM900, PCL 0 corresponds to the largest nominal
output power (39 dBm), power control levels between 16 and
31 can be set for phase II mobiles only.
In contrast to the PCL the power class characterizes the
nominal maximum output power of the mobile. Depending on
the power class of the mobile the range of possible PCL set-
tings may be restricted.
GSM mobile phones are divided into different power classes according to their
maximum output power:
Power class Nominal maximum output power in dBm
GSM400 GSM1800 GSM1900
1 – 30 30
2 39 24 24
3 37 36 33
Besides, a dimensionless scale of power control levels (PCL) is determined for
dynamic control of the mobile power:
GSM Power control levels (PCL)
Power control level, PCL Nominal output power in dBm
GSM400/ GSM1800 GSM1900
0 39 30 30
1 39 28 28
2 39 26 26
3 37 24 24
4 35 22 22
5 33 20 20
6 31 18 18
7 29 16 16
8 27 14 14
9 25 12 12
10 23 10 10
11 21 8 8
12 19 6 6
13 17 4 4
14 15 2 2
15 13 0 0
16 11 0 0
17 9 0 0
18 7 0 0
19 - 28 5 0 0
29 5 36 36
30 5 34 33
31 5 32 32
PCL 16 to 31 for PCL 11 to 15, 29 PCL 11 to 15,
phase 2 only to 31 for phase 2 29 to 31 for
only phase 2 only
> BTW: Not that I would know anything about these things. But would it be
> possible to lower the TX-power of the FR to see how far down you can go
> without loosing connection?
The maxmum TX-level is a parameter that can be set in firmware (see power
class above). So we could patch the firmware for GSM-chipset to disallow
levels of TXPower, that are causing problems with GSM-noise. Anyway hat
probably would cause a reduced connectivity, means FR isn't able to place a
call at all in situations where before we had the possibility to have a call
with noise issue. Fore sure there is a wide range of signal levels where the
BS would like to increase the MS-TXpower, but still is able to keep the
connection even when we refuse to power up. These situations probably could
be augmented by choosng a different power class for FR.
Alas it seems there is no AT-command or other means to this from userspace, I
guess the only way to do this is by flashing new firmware to the calypso.
> I have active speakers in my room that catch GSM signals. I noticed
> several times that the noise in the speakers when the phone is
> registering is louder with the FR than with a Motorola V3x with the same
Hmmm, maybe source side signal quality of FR (the generated RF-signal) is
inferior compared to the other phone, so BS would compensate for this by
pushing up FR's TXPower. :-/
There's a lot of tests we have to do, and a lot of questions not answered yet.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 194 bytes
Desc: This is a digitally signed message part.
Url : http://lists.openmoko.org/pipermail/support/attachments/20080805/d4a53a2a/attachment.pgp
More information about the support