-6dB loss phemonena in differential pair signaling

[yefj]

In differential port 1 we have same power going threw 1 SE and 3 SE. we get the ouput at 2 SE and 4 SE.

suppose 2SE is faulty.What whould be the value of SDD21?how SDD21 in that case? What is the mathematical logic we get -6dB in SDD21? Thanks.

 

[sophiecentaur]

could you clear something p for me please? Is that diagram of what I would have called a 3dB coupler or Hybrid?)- with two quarter wave 3db coupled lines? (1-2 and 3-4 ). Of so, I would expect the inputs for1 and 4 to combine in phase and the output from 1 to be +6dB and the output from 3 to be zero (many negative dB to nearly zero power) when the inputs to 1 and 4 are equal and co-phase. In practice, unbalanced power would emerge from 3.

From what you have said, you seem to be connecting 1 and 3. The device I used to be familiar with should have its inputs at 1 and 4.

Perhaps you could try that and see if the result is more what you want.
It's at least 30 years since I made and used one of these and it was all UHF (below GHz) with clunky micro strip but the theory should still be the same.

 

[Baluncore]

In differential port 1 we have same power going threw 1 SE and 3 SE. we get the ouput at 2 SE and 4 SE.

suppose 2SE is faulty.

Based on so little solid information, you really cannot expect us to guess, but you don't answer questions, so here goes ...

I assume there is a ground plane below the tracks. You have a balanced line between 1 and 3, with an unbalanced line between 2 and 4. Since 2 is faulty, I assume it is shorted to the ground plane. The impedance of the output is half that of the input, so you may have built a balun transmission line transformer with a 100:50 impedance ratio.
I have no idea how you are instrumenting or terminating the lines. If you expect an output voltage that is 1/2 of the input, then that will be 1/4 of the power = -6dB.

If your VNA has balanced 100Ω transmitter inputs, 1 and 3, each of which is a 50Ω coaxial connector, then the VNA may be doing something with the computation since the receiver inputs 2 and 4 are not balanced.

 

[sophiecentaur]

Based on so little solid information, you really cannot expect us to guess, but you don't answer questions, so here goes ...

Little information; yes. I was thinking that the layout is like the well known directional coupler design of which there are millions on micro strip and smaller:

the signal paths are shown to be 'diagonal'., which agrees with my last post Does this fit in with the OP's description? I notice the OP's diagram is a CAD version; what's actually in the real test device and how is it actually being connected? It could affect the result, without anything being actually wrong with the device.

 

[Baluncore]

Does this fit in with the OP's description?

With much closer tracks, above a ground plane, it would form a backward wave coupler, but the tracks are not close coupled, they are widely separated, and remain parallel.

I suspect the OP is testing a 100Ω parallel differential transmission line, formed from two parallel 50Ω tracks, each above a ground plane, with each side of the differential fed through a 50Ω coaxial connector.

The track fault hypothesised could be a coaxial connector series-open, or parallel-short, both of which would have λ dependent reflections.

We will probably never find out if this was an EM simulation, or a prototype under test.