Re: [Discuss-gnuradio] working AM radio - critique

[Dave Emery]

On Tue, Aug 24, 2004 at 03:55:39PM -0700, Ettus, Matt wrote:
> Broadcast AM radio uses a max of 100% modulation, I believe.  

        That is correct, though in fact there may be some upward peaks
greater than 2X the carrier.   Dropping down to actually 0 signal
produces splatter (transient energy outside the nominal bandwidth of the
signal - sometimes for tens to hundreds of khz around the signal) so AM
stations are not actually supposed to go beyond 95% in the negative
direction (it has been more than 40 years since I got my FCC First Class
Radiotelephone license back when I was in the 10th grade, so the precise
rules may have changed a bit since I learned them).

        Most broadcast stations use peak limiters (these days based on
DSP) to prevent overmodulation in the negative direction.   FCC rules
require them in fact.

        However, one point that very much needs to be made is that AM
Shortwave signals (as opposed to AM broadcast band signals) are subject
to something called selective fading (due to interference between signal
energy propagated by different paths in the ionosphere). This can result
in substantial differences in level between the carrier and sideband
components and can produce signals that appear to have substantially
more than 100 % modulation as the carrier fades more than the sidebands.
        
        And it also needs to pointed out that real world signals often
include significant impulse noise (static crashes, car ignitions, light
switches being turned on and off, etc..) so scaling them so a 100% 
modulated signal is full scale guarantees clipping and saturation on
noise spikes which may not be what you want.   Clipped signals have very
sharp edges and lots of high frequency transient energy (as in the case
of splatter in an AM transmitter) which may cause problems in subsequent
processing or reproduction.   At the very least they can cause overflow
errors in carelessly written code.

        Thus some headroom is desirable and customary...

> You can get the exact carrier level by doing a lowpass filter.  A simple one
> is a moving average or a recursive averager.  You'll want a cutoff frequency
> lower than any audio components you expect to receive.  I think 100 Hz would
> work, but I'm not sure of the lower cutoff used on AM broadcast stations.

        Try around 20 hz...  though that varies with the transmitter
used and how serious the station is about sounding good.
-- 
   Dave Emery N1PRE,  address@hidden  DIE Consulting, Weston, Mass 02493