Differential lines termination

[MuroSamuro]

Hello there,
I have yet to design a differential pair TL, so I came up with this fundamental question:

Say you design a single ended TL that is connected at its both ends to coax cables which are connected to a calibrated vna. To avoid reflections, the characteristic impedance of the TL should be the same as the coax cable.

However, in a differential pair you design for a specific differential impedance, which could be acheived for many values of isolated characteristic impedances (typically marked as Z0).
My queation is, should I design the isolated characteristic impedance (Z0) to be equal to the coax cable impedance to avoid reflections ? Or deaigning the differential impedance alone is enough?


Thanks

 

[KlausST]

Hi,

I don´t understand.

You say differential pair ... and you say coaxial. How do both match?

Klaus

 

[MuroSamuro]

In some youtube videos they show how to measure the differential impedance when the four terminals are connected to vna

 

[D.A.(Tony)Stewart]

This only makes a difference when measuring CMRR with CM or ground noise or Transfer Impedance from outside shield degradation but for a pure TL. I think you use 50+50 to match a 100 Ohm differential Zin when the differential has a null virtual ground between Vin+ and Vin-

I think.....

But if doing something like this for Differential 50 Ohm impedance then you ignore single ended impedance.

 

[BigBoss]

There is no such Isolated Characteristic Impedance.
Differential Lines have Odd Mode and Even Mode Characteristic Impedances and the Geometric Mean of These Impedances will give you Characteristic Impedance. Also, You have to use a Differential to Single Ended Converter to connect these Differential Lines to Coaxial Cable. This converter might be a transformer, balun or any other similar element.

\[ Z_{0}=\sqrt{Z_{odd}Z_{even}} \]

 

[volker@muehlhaus]

My queation is, should I design the isolated characteristic impedance (Z0) to be equal to the coax cable impedance to avoid reflections ?

Yes, exactly. If the coupling between the lines decreases, we converge towards Zeven=Zodd=Z0

For your differential line on PCB, you can increase the gap towards the two separate connectors, and adjust the line widths accordingly. You start from a 100 Ohm differential line, and end with two lines of 50 Ohm impedance and no coupling at the connectors.

 

[crutschow]

My queation is, should I design the isolated characteristic impedance (Z0) to be equal to the coax cable impedance to avoid reflections ?

Don't really understand the question.
Are you planning on connecting the differential cable to the coax cable?
Otherwise how would there be reflections?

 

[PlanarMetamaterials]

Due to field asymmetries, a three-conductor TL cannot interface its differential mode directly with a single coaxial mode; this is why often 4-port setups are used instead to extract the relevant modal quantities in post-processing.

If you are interfacing the TL with a 4-port 50-ohm impedance VNA, then the differential mode impedance must be 100 ohms to minimize impedance mismatches, since the two VNA ports at each end of the line are essentially in series.

The "isolated characteristic impedance" is irrelevant if the lines are coupled. You may choose to define a "Z0" for each of the individual lines, but these are physically meaningless since they do not correspond to actual propagation modes.