Edward R Cole kl7uw at acsalaska.net
Mon Nov 7 00:13:21 CET 2016

Comments inserted, below:

At 01:30 PM 11/6/2016, Leif Asbrink wrote:
>Hello Serge,
> > 1)  50 ohms dummy load on the IQ+ INPUT
> >
> >
> >     FLOOR NOISE  5DB  ON MAP65
> >
> > CONCLUSIONS: THE 16DB required are barley enough 22db -5db floor
> > noise =16db RF GAIN & very often depending on the AZ/EL it can go
> > down to -19db -5db = 14db RF GAIN well below of what is required.
>What you measure is the right thing, but it is not RF gain.
>It is how much the noise floor changes when you switch on/off the
>LNA. As expected your #1 and #3 gives the same noise floor.
>22-5=17 and that is OK, but you note that this difference can go
>down to 14 dB and that is a bit low. If you add a LNA in front of
>the IQ+ you can expect to gain 0.10 dB. That will be hard to notice
>but your detect rate in map65 should improve by about 10% at
>the detect limit with the FT decoder which shows the steepest
>dependence on S/N. A station that detects with a rate of 50% would
>improve to about 56% if you add a LNA - but that would happen only
>when your antenna points in a direction where your noise floor
>is at 19 dB. The DS decoder is less steep and if you use that you
>can only expect an improvement from 50 to 53%.
>Maybe, if you look at a time when Leo (quiet sky) is in the clear
>over the ocean you might see an even lower noise floor.....
>Anyway, your system is very good (provided your LNA NF is low)
>I suggest you consider how often you work stations at the detect
>limit in a quiet direction. If you make 10 such contacts per year,
>improving detect by 10% from 50% to 55% would make QSOs faster
>but not often.
>On the other hand, if you have a sked and detect only 10% of
>the transmissions with DS, a 0.1 dB improvement on S/N should
>give a detect rate of 11.5%. Instead of detecting one in 10
>transmissions you would detect one in 8.5 transmissions.
>The advantage of adding more gain is small. There is a risk that
>there are strong out of band transmitters that saturate your
>system if you add too much gain.

Serge stated he has a quiet location - not sure if that also 
eliminates strong signals out of band like TV or commercial FM.  But 
understanding is he is remote from all those problems

>There is no such thing as "the required gain" Gain is always
>a compromise. In a very quiet location (like mine was when
>I was QRV) it is satisfying to use very high gain to know that
>one has squeezed the last fraction of a dB from the system.

This topic started with Serge noting a difference in decoding 
threshold between using MAP65 and WSJT.  So maybe this will provide 
that small improvement to have both Rx systems equally 
effective.  This in the -28 to -30 signal range.

>By the way, even if you can not hear it you probably have some
>impulse noise that lifts your noise floor. There is a blanker
>inside map65, but I do not know how good it is.

Not very good from my perspective.  If you apply it lightly to noise 
that runs 6-10 dB higher than your quiet noise floor it helps.  But 
if faced by major noise running 20 to over 30-dB then applying the 
MAP NB aggressively does lower noise but also lowers signal.  The 
effect is raising the decoding threshold from --28 to -5 or worse.  I 
cannot decode HB9Q on 2m when noise is running S8/9 (>30 
dB).  Normally HB9Q is -10 to -4 on JT65.

I have more success using the DSP NB and/or NR in my K3, but only 
usable with WSJT as there is no IQ output from the K3, only monaural 
audio on the data line out.  To use K3/KRX3 dual Rx I have to tap 
into the 1st IF which is sent to two ext SDR which are phase-locked 
which provide two IQ inputs to the delta44.  K3 DSP is located in the 
15-KHz 2nd IF of the radio.  So I can only use the K3 NB/NR with 
single-pol WSJT which works with simple soundcard input.

This is what makes the IQ+ and recent other dual-channel SDR's better 
choices (simpler/cheaper) approaches for using MAP65.

>  You might use
>a sound recorder to make a .wav file from your Delta44 while
>rotating the antenna 360 degrees in azimuth. Noise pulses would
>be visible with an audio editor such as audacity. If you make
>such a recording I would be happy to look at it - and if you do,
>switch off the LNA for 10 seconds at the end.

I sent Leif a wav recording of my noise interference which he was 
able to analyze, saying the NR in Linrad would not be effective for my noise.
I have to locate the source and eliminate it (if legally 
possible).  Using my shotgun as noise blanker is not an option <smile>

73, Ed - KL7UW

73, Ed - KL7UW
     "Kits made by KL7UW"
Dubus Mag Business e-mail:
     dubususa at gmail.com 

More information about the Moon-net mailing list