Topband: question about RX splitter : Conclusions

[VE2TZT]

Hello Folks,



Following the promised feedback:



After studying all the stuff you sent me it appeared that I had to design 
and realise by myself the needed splitter. The post of W5XZ Dan was the one 
which inspired me the most to start this design.



After many tries, the better splitter I was able to realise for my need is 
made of a floating (no need to be connected to the ground) primary of 7 
turns with a secondary of 4+4 turns on the well known BN 73-202 binocular (I 
use a old spool of what I believe to be some #27 transformer wire for all 3 
wires) . That gives the 1:1.15 turn ration needed to transform the 75 ohms 
of the primary to the 100 ohms of the secondary. The common point of the two 
half secondaries is connected to common ground via a 25 ohms resistor (yes 
twenty five) and the opposite sides are the 50 ohms outputs which in this 
particular configuration are 180 degrees out of phase. This 180 out of phase 
is of no importance for just feeding two different radios.

I achieved some very good performances measured with the VNA 2180 : from 1 
to 30 Mhz, SWR better than 1.2 on 75 ohms input and better than 1.3 on 50 
ohms outputs and less than 1.1 below 8 Mhz for input and outputs. The 
rejection between output ports is an impressive -39.5 dB and even better 
over 30 Mhz.

One of the prototypes with an additional parallel trimmer capacitor on the 
primary gave -41 dB on 160m going down to -45dB on 20m, -53 dB at 30 Mhz and 
ending back to -42dB at 50 Mhz...



Except for the additional to -3dB input to output loss which is slightly 
higher than found in literature (I got -0.4 dB on 160m and -1dB on 20m) this 
splitter while simpler is for my particular need far better than all those 
tested or documented and it gives the additional feature of 75 to 50 ohms 
impedance conversion.

During my experiments, I have seen that the key for a good rejection between 
output ports is the optimal coupling  between the 3 wires : the primary and 
the two secondaries. For that, it is very important to twist the 3 wires (1 
wire is 35cm long the 2 others 25 cm long) together with 1 turn every 10 mm 
long plus or minus 1mm before winding in the binocular. If less or more than 
this turn per 10 mm, there is a degradation of the performance. Twisting and 
winding only the two secondaries together and passing the primary later 
results in poor performances. For I think the same coupling reasons, I did 
not get such good results with usual FT50 cores (tried FT50-43 and FT50-75). 
A strange point is that modifying the same transformers, I got 
reproductively 1 dB of better port isolation with all outputs on the same 
side of the binocular than with the primary outputs at one side ant the 
secondary outputs at the other side.



The consequence of that important point about coupling is that because that 
strong coupling is increasing stray capacitance between primary and 
secondary, a good splitter cannot have a very good common mode rejection, 
then a separate 1:1 transformer must be added at the input to achieve this 
function.

Conclusion : I will need only 2 cores in place of 4 to realise the project 
described in my first post.


Thanks very much to the many participants, each answer has brought me a 
piece of the answer.

73 de Gilles VE2TZT