[Moon] EME 2016

Leif Asbrink leif at sm5bsz.com
Sat Aug 27 01:20:15 CEST 2016


Hello All,

The EME 2016 event was the best EME conference ever to me 
personally. Interesting results of real measurements were 
presented and results were not really as expected. Experiment
wins above theory - particularly when theory is a sloppy
guess....

Dominique, HB9BBD, wrote (on the MOON reflector)  
> GuysWhat I saw during this event was an IT conference, 
> interrupted by clowneries of a bunch of antenna builders.
He also wrote things I do not want to copy....

I am well aware of Dominiques position, but it is not fair to
say the EME2016 event was digital and antenna building only. 
I do not think Dominique would say "I already knew this" or 
"it is not relevant to CW operators" what I write below.

I will try to explain the implications of two of the papers.
G3WDG - Experiences with Circular polarization on 10 GHz and
G4DDK - An investigation into EME LNA safe operating levels.

It is well known (?) that on 10 GHz with linear polarization
the loss when using the wrong polarization is about 11 dB
http://www.sm5bsz.com/linuxdsp/usage/pol10ghz.htm

Without really thinking I have assumed that the loss when using
the wrong circular would be the same. Charlies lecture shows us
something completely different!!!

On page 101 of the proceedings we see that the signal from 
the Rx port  is 8 dB stronger than the signal from the Tx port.
That is the values obtained from the maxima in figure 6, page
101 in the proceedings. The spectral broadening in the Tx port 
is much wider than the broadening in the Rx port however. The
3 dB points are separated by 110 Hz while the 3 dB points in
the Rx port are separated by 40 Hz only. That would mean that
the power ratio would be 3.7 dB if the curve shapes were the same.
That is not the case, the ratio is a little higher, maybe 4 to 5 dB.
It is clear however that a lot of energy is lost in the 
orthogonal polarization. Charlie suggests "nearly 25% of the 
returned power from the moon is lost into the opposite [circular]
polarization." 

Assuming a ratio of 4 dB (2.5 times) 71% of the power would appear 
in the Rx port and 29% would appear in the Tx port. By putting a
second receive system on the Tx port one would get a second, 
independent signal. Sending both of them to the head-phones
should improve sensitivity significantly. In case the signal is
linearly polarized, using both circiular ports, one to each ear
should improve sensitivity, by how much is unknown to me, but
combining the two signals in the appropriate phase would provide
linear polarisation aligned with the polarization of the incoming 
wave. 

>From figure 8 we learn that circular to circular is about 1 dB
worse than linear to linear. That is a factor of 0.79 not too
far from the guess based on figure 6.

These are the conclusions I make based on Charlies paper:
LP to LP has negligible loss.
CP to CP has 1 dB loss
CP to LP has 2 dB loss relartive CP to CP
CP to opposite CP has 4 dB loss 

It was unexpected to me - but once I know it as a fact I think 
I can understand it. Halfway from the center the moon surface 
slopes by 45 degrees. Vertical (on the moon) surfaces form corner
reflectors that reflect circular back to the earth with the
wrong polarization while linear polarization is changed much 
less.

Combining both circular ports on receive would give a 3 dB
advantage (minus losses) if the other station is using linear
polarization. When the other station is circular, combining
ports incoherently will give a smaller advantage. When
combining two signals S/N and 0.4S/N (power ratios) my guess
is that the improvement would be something like a factor of
1.2 or just a single dB. 

It was pointed out in the discussions that big stations might
prefer linear polarization on 10 GHz in order to be able to 
work terrestrial stations that have power enough, but use
linear polarization. This will soon become far more important
in the near future with the new algorithm developed by Nico
IV3NWV that mostly eliminates the S/N degradation due to the
spectral broadening. This will make 10 GHz EME about 6 dB
easier than it is now with the best digital mode. This is 
very new, it is not in the proceedings, but there is a hint
in the video from his lecture. It seems to me that the clever
method to avoid S/N loss due to spectral broadening can
be applied to CW also. Maybe with a slightly smaller advantage,
but even a 3 dB improvement would be welcome I would think.
Basically one would run many receive channels with a matched
bandwidth of about 18 Hz for CW. The EME signal would 
spread over maybe 10 such channels. One would compute
the probability of key up or key down for all the channels
in a certain time interval and compute the product of all
the probabilities. The probability that goes from 0 to 1
would be used to send something into the loudspeaker.

Dominique and other CW operators might benefit from listening
to lectures done by people who understand information theory.
Most of it is about coding - but there are also other things...

The paper by G4DDK shows us that it does not matter much for the
survival of LNAs whether DC supply is on or off. The limit is
in the order of 15 dBm in both cases. Note that these measurements 
are in a 50 ohm system. There seems to be a small advantage to
have DC supply off.

The advice "it does not matter much, but power off seems
just a little better" is ONLY applicable when a relay is
used to switch in a 50 ohm dummy.

In situations where a relay is used to switch between Rx and 
Tx on the same antenna it is different. The recommendation to 
have DC supply on comes from practical experience. If one connects
a conventional LNA to a conventional shorting relay such as the
CX-520D on 144 MHz directly to a conventional LNA on 432 MHz
one might use the attenuation -53 dB from the data sheet to
compute the maximum allowed power on the Tx side. Based on Sams
paper we might go for a 10 dB safety margin and tolerate 5 dBm
into the LNA. That means one would believe that +58 dBm would be
the maximum safe Tx power. (=630W) This belief is wrong. The output
impedance of the relay is 7nH with a resistive component of about
0.4 ohms. When the CX-520 is used on 432 MHz with 630 W Tx power
it would produce 5 dBm into a 50 ohm load, it means that there
would be 0.4 V rms across a 50 ohm load. If instead we connect
a matched load, 0.4 ohms (plus the appropriate reactive part), 
something we might have in an unpowered LNA, we would have 
0.2 V rms across 0.4 ohms which is +20 dBm (100 mW) and 
definitely more than a LNA would survive. In case we keep the
LNA powered, thi input impedance will stay closer to 50 ohms
and power transfer will be much smaller and the LNA would
survive. I think this is the explanation why practical
experience suggests one should leave the DC voltage on.

If you want to use a relay with poor isolation, look here:
http://www.sm5bsz.com/pindiode.htm
When both the relay and the LNA are very far from 50 ohms
we can select the cable to produce the largest possible
mismatch between the LNA and the relay. That would increase
attenuation dramatically and it works best when the LNA is
unpowered.

It is interesting to note that the impedance of the LNA changes
when the gate current starts to flow. When the supply voltage is zero,
gate current starts to flow at modest power levels. The gate current 
will change the impedance at the gate from very high to very low.
One has to select the cable length for minimum gate current
at power levels high enough to produce gate current.
(If there is never any gate current - isolation is high
enough and DC on or off would not matter.)

Conclusion: The best advice is to make sure the LNA always has
50 ohms on its input. Then one can trust Sams study and not
worry about whether the LNA is powered or not.

Under certain circumstances it is however possible to get
an advantage. The extra relay is likely to introduce more losses
than a cable that maximizes the mismatch between the relay
and the LNA. Nominal attenuation in a 50 ohm system does not 
have to be particularly high to make an unpowered LNA survive.



More information about the Moon mailing list