[TowerTalk] fading and the value of 1 dB

[Michael Tope]

I've thought about this before. You could develop an experiment whereby 
your contest logging software could control a 1dB attenuator and switch 
between power levels on a short-term random basis descretized to ~1 
minute, 1 QSO, or 1 CQ. The software would have to make sure that the 
average time spent at the each of the two power levels was the same. The 
equipment would need to be such that it would not give subtle cues to 
the operator as to the state of the transmit attenuator (i.e. the 
experiment would need to be blind).  The attenuator would only be in the 
transmit path, so the receiver gain would not be effected by the 
attenuation state.

Over the course of a contest, you could compare the QSO rates and 
overall score of the Pout state versus the rates of the Pout + 1dB 
state.  You could run an experimental control where the attenuator value 
was 0dB to gain some insight into the noise floor of the experiment. You 
could also run the experiment multiple times with different attenuation 
values to see what difference in dB was required for the effect to show 
up over the course of a single contest. The value might turn out to be 
mode dependent.

It would be a ton of work and would require multiple participants, but 
the results could be fascinating.

73, Mike W4EF...........


On 5/18/2022 9:59 AM, Lux, Jim wrote:
>
> Here's some references that are interesting...
>
> https://www.isode.com/whitepapers/skywave.html  talks about variations 
> over 10 seconds to a few minutes.
>
> http://tracebase.nmsu.edu/hf/reports/walnut.pdf describes a model of 
> HF links.
>
> And gives some statistics. D (decile value of variability) = 
> 1.28*SigmaN (SigmaN = SD of noise power)
>
> 3 MHz D is about 8 dB during stable hours and ranges from 10-17 during 
> transition
>
> 10MHz D is 5 dB during stable hours, 7-11 transition.
>
>
>
> Shorter term variations (Rayleigh fading) - those are modeled by 
> what's called the Watterson model.
>