Re: Compact Fluorescent Noise

["Jeff Volp" <JeffVolp@xxxxxxx>]

"Joerg" <notthisjoergsch@xxxxxxxxxxxxxxxxxxxxx> wrote in message
news:0Fb2h.26109$7I1.14085@xxxxxxxxxxxxxxxxxxxxxxxxxxxxx
> Hello Jeff,
>
> > There doesn't appear to be an effective way to deal with the randomness
of
> > this kind of noise on a cycle-by-cycle basis.  But then a longer term
> > average has its own set of problems.  Eliminating this kind of noise at
the
> > source may be the best answer.
>
> Have you tried narrowing the LC filter? Several kHz wide might still be
> ok from a ringing point of view.

I've been back and forth on LC ratios and damping.  After countless
simulation runs and many hours of testing actual hardware, what is in there
now seems to provide the best  tradeoff.

> The other trick is what radio engineers refer to as a "noise blanker".
> If the spike is short enough versus the channel information a riding
> threshold detector is set. This detects sharp peaks above a slower
> signal and then notches out that portion of the receive phase. The good
> ones work remarkably well in muffling irregular spike noise.
>
> While the LC filter should be an easy one the noise blanker does tend to
> increase uC workload to the point where one has to tap the cost brakes.

I used that technique to blank impulse noise back in my ham days, but that
would not help here.  The waveform I captured this morning showed beating CF
noise almost mimicking X10 data.

The PIC was added to the original XTB design to just gate off the
superfluous 3-phase bursts so transmitted energy is concentrated in the
essential zero crossing burst.  Once the PIC was on board I decided to
include TW523 emulation for little additional cost.  Then I added complete
error detection, the ability to receive sequential dims, extended messages,
and now AGC.  Lurking in the future may be even a repeater capability built
into the XTB-II.  This is WAY past what the hardware was originally designed
to do.

Since most X10 transmitters cannot develop the signal levels produced by the
XTB, I made the receive channel as sensitive as possible.  Unfortunately,
that gets it down into the noise level.  The simple solution is to just
stick thumbs in its ears and ignore low-level signals.  AGC gives it the
ability to recognize low-level signals in a low noise environment while
still allowing it to function well in a high noise environment.
Unfortunately, there will always be noise sources that can cause errors.

Jeff