In another installemnt on RFI mitigation, I spent 1 hour today debugging th=
e
cheap PC speakers on my desk which have been whistling/singing pocsag/cell=
=20
tower melodies for 1.5 years now. I had enough and I proceeded to silence t=
hem.

I still do not know what exactly causes them to go blabeepblabeepblabeep al=
l=20
the time even with volume set to zero. Inside, the usual cheap phenolic pcb=
,
containing rectifier bridge, obligatory too low voltage caps for the=20
transformer at idle (changed them), a potentiometer and a TDA2822M dip audi=
o=20
amp from NXP.

And it received the blabeepblabeepblabeep with enthusiasm. 500pF caps on th=
e
input cable, "shielded" low cost cable to PC sound output, did not help. Ge=
tting
my hand closer and further to the bare board made the electronic music get
quieter (!) then louder again. It was almost okay with my hand at about 5cm=
.. I
can't sit like that, I get cramps. I do not know what wavelength it was, bu=
t
5-6cm would be right for roughly lambda/2 at 2.4GHz. Removed the laptop fro=
m the
vicinity, same. Phone removed, same.

Eventually it took 2 ferrite donut cores on which all cables going to the b=
oard
made one turn to get the noise down to half... but it was still there.

So I took a look at what was left: the cca. 10cm (magic number!) long wires=
 from
the board to the speaker which was in the box with the amp, the other one b=
eing
the remote one, and on an already donut equipped wire. After changing the w=
ire
length and sacrificing another donut to the RFI gods (you have to admit tha=
t the
RFI gods have high expectations, putting donuts on 10cm wires inside $9 PC
speakers...), it was QUIET.

So I put it back where it usually stands, and it started singing again. So =
I
grabbed the nearest ferrite objects I could find without waslking away from=
 my
desk, two E halves about 3cm on the long edge, and wrapped as many turns of=
 the
wire between the speakers and of the input audio "shielded" wire around eac=
h,
and held them in place with bread bag ties. It works!

So, to make a low frequency, ~1W, $1 small integrated amp (NXP TDA2822M) wo=
rk in
a soho environment in anno domini 2013, the following supplies were needed:
 - one guy with some experience in RFI gods placation
 - 2 good ferrite toroid cores, about 1.5cm o.d. each, of unknown material
 - 1 good ferrite toroid core, about 5mm o.d., also unknown origin
 - 2 E ferrite cores about 3cm on the long edge, ~6mm thick
 - 2 bread bag zip ties
 - 2 500pF disk caps
 - a small mod to the pcb which had a ground interrupted in some location i=
n the
hope to cut ground loops - in this case it made the ground plane of the pcb=
 a
better antenna so I shunted the gap in the ground with some thick wire.
 - almost 2 hours of work

After this, I can STILL hear the free electronic music faintly in the backg=
round
with the PC off and regardless of set PC speaker volume.

I do live in downtown and there are lots of cell towers and whatnot here, b=
ut
this is going a little too far. I had similar but different adventures with=
 many
other ic's in the past, including with LM386s which liked to oscillate at >=
15MHz
and to rectifiy cell phone pulsed signals.

I do understand that die shrinks are necessary but does ANYONE test for
susceptance after the fact? I mean, I was about to improvise a tin can shie=
ld
for an audio amp?!

Please share your experiences, I believe that the subject is a painful one =
to=20
many.

-- Peter

--=20
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.