GSM-noise "buzz" issue

[Joerg Reisenweber]

Am So  21. September 2008 schrieb Uwe Klein:
> On 9/21/08, Werner Almesberger <werner at openmoko.org> wrote:
> > Andy Green wrote:
> >  > One interesting thing would be stick a tracking generator on the GSM
> >  > antenna, put it near the headphone socket in a repeatable way, stick
> >  > spectrum analyzer probe on the victim net, and sweep them to get a
> >  > display of coupling efficiency vs frequency.
> >
> >
> > Do you think this will work ? I thought of using a tracking generator
> >  when I wrote that mail lamenting that Joerg didn't have a signal
> >  source, but then I realized that almost all tracking generators have
> >  their maximum output at about 0dBm, while our troublemaker should be
> >  more like 30dBm.
> 
> I had thought about using the FR as an energy sink.
> Using a 4pole networkanalyzer and a pair of antenna
> in a grid dip meter style arrangement.
> This should indicate all absorbtive elements.

A very cute test setup.
Anyway I don't exactly see what we have learnt once we know all dips by name, 
err frequency and dB. I'd guess we get a lot of dips, and we can't tell what 
element caused them, and especially we don't learn anything about audio 
contamination path. Even less we learn something *new*.
Werner's original proposal seems somewhat more result oriented, alas all this 
fails when attaching probe to the "victim net", due to total change of 
RF-properties (probe has some pF! need to open can, etc). So probably the 
most helpful test would be to use a +36dBm generator at 850/900/.... 
and "record" buzz by arecord piped to ssh-stdout to external PC, so you can 
level up generator and see what dBm you need at different frequencies to 
start buzz. Basically exactly the tests I already did in TPE, though I had to 
use a second phone for generator purpose, and this was rather clumsy and not 
very reproduceable.

Also direct coupling of RF to JK4401:4 at a moderate level might be a way to 
go, then check with scope/spectralizer all the pins and pads inside can, and 
get an idea of where RF jumps to from primary path (if any). Again we won't 
learn anything about how we could improve things, other than KEEP RF OUT of 
CAN, i.e. away from non-linear audio-circuits.

Anyway, I'm looking forward to read your report on test results.

cheers
jOERG
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 194 bytes
Desc: This is a digitally signed message part.
Url : http://lists.openmoko.org/pipermail/hardware/attachments/20081001/af241b80/attachment.pgp