[Elecraft] unequal power output per tone in FSK

[Kok Chen]

On Jan 10, 2011, at 3:35 PM, Wayne Burdick wrote:

> The effect on transmission bandwidth is  negligible, and it's also extremely unlikely to affect copy.

It depends.

Consider an RTTY demodulator that includes automatic threshold correction (ATC).  It will treat such a signal as one with 1 dB worth of constant, selective fading.  This will affect the SNR of the whole scheme. I.e., when you apply a threshold correction the noise goes up by the same amount.  To boost the lowered tone, the noise in that channel is also boosted.  You can see the effect exposed here

http://homepage.mac.com/chen/Technical/FSK/ATC/

So, unless you apply the more complex circuit (I don't know anyone else than myself who uses it), the character error rate will be affected by the amount that is equivalent to 1 dB more SNR degradation.  

How much is it?  With a weak but stable conditions (no fading, no flutter), a 1 dB imbalance can double the character error rate.  If you have fluttery condition, it is true that 1 dB of degraded SNR won't do much additional harm.

You can apply the same analysis for "FM" demodulators and should see something similar.  Some modems in the early days use hard limited "FM" demodulators instead of an ATC to fight selective fading because of the Page Communications patent 2,999,925.

Some of us RTTY nuts tweak for the last decibel :-).

73
Chen, W7AY