Assuming you are measuring a carrier and not a satellite originated CW beacon.
The downlink C/Ndown will improve by 3.5 dB but the overall C/Ntotal will not improve by that full amount.
The C/Ntotal you measure on the spectrum analyser is the summation of C/Nup + C/N intermodulation (if multicarriers in the transponder) + C/N down.
For example:
C/Nup= 25 dB
C/Nintermodulation = 21 dB
C/Ndown = 15.5 dB.
C/Ntotal = 25dB + 21 dB + 15.5 dB = 14 dB
If you improve the C/N down by 3.5 dB so C/N down = 19 dB the C/N total becomes 16.2 dB, which is only 2.2 dB improvement.
To do the calculation yourself do the following with the input C/N values:
divide by 10, change sign, 10^X. Write down the amount.
Do this for each value and add them all up.
Process the total thus: log, then times 10.
e.g.
25 dB > 0.00316
21 dB > 0.00794
15.5 dB > 0.02818
total = 0.03928 > 14 dB
25 dB > 0.00316
21 dB > 0.00794
19 dB > 0.01259
total = 0.02369 > 16.2 dB
Typically a VSAT outlink carrier or TV carrier from a large hub dish to the remote sites will operate with a high uplink C/Nup and the C/Ntotal at the remote sites will be closely associated with the VSAT dish size. The high uplink C/Nup is achieved using a large dish, high power amplifier and a low gain setting on the transponder.
Return links from remote sites are likely to have their C/Ntotal constrained by the small VSAT size and transmit power. The satellite transponder has to operate at a high gain step setting. The result is that the C/Ntotal depends mainly on the uplink C/Nup and increasing the hub receive dish size from say 6m to 13m to improve the dowlink C/Ndown makes no significant improvement overall.
Play here:
https://www.satsig.net/linkbugt.htmwxw
Best regards, Eric.
* The above analysis is simplified as it does not mention the uplink and downlink interference entries from cross-pol and adjacent satellites, which might typically be 25 and 25 dB. These figures vary significantly.