Hmm, hadn't thought of that. I'll check my generated .WAV files via
the waterfall plot in Spec Lab to make sure the I/Q is the right way
round. Although, I think I did check it once and thought it all
agreed, but another test will help.
Meanwhile, plotting 0.004Hz bin-ize waterfall plots of two PN sequence
generators used as noise source. Paint dries quicker that this
Andy
On 10 December 2015 at 10:59, Paul Nicholson <[email protected]> wrote:
>
> Results of IZ7SLZ 16K21A 6C 2S 2015-12-09
>
> 18:00 no decode
> 19:00 Eb/N0 = 4.4 dB BER = 34.1% rank 0 ph uniform
> 20:00 Eb/N0 = 4.2 dB BER = 34.5% rank 0 ph 0 30 30 60
> 21:00 Eb/N0 = 3.5 dB BER = 35.5% rank 0 ph uniform
> 22:00 Eb/N0 = 7.5 dB BER = 27.8% rank 0 ph uniform
> 23:00 Eb/N0 = 7.1 dB BER = 28.8% rank 0 ph uniform
> 00:00 no further decodes
>
> Easy decodes with such a strong signal, weakest was about -48 dB S/N
> in 2.5kHz. Symbol timing is within 30mS.
>
> Andy wrote:
>
>> I discovered by trial and error The FreqOffset window in
>> EbNaut had ot be set to [minus] -3Hz to compensate for the
>> positive shift on the input - after spending a fruitless
>> few minutes trying at +3Hz . But I suppose this is the
>> same meaning and sign as the start time offset...
>
> It should have needed offset +3Hz. Following the principle
> of least surprise we would expect +3Hz to mean the signal is
> at +3Hz in the input!
>
> Maybe one of us needs to reverse the sign of the Q signal?
> --
> Paul Nicholson
> --
>
|
