Differential coax cable vs 2 single ended coax cables

[Novicent]

Does anyone have recommendations for a 100 ohm differential coax cable? Can I use two matched single ended 50 ohm cables instead of the grounds are connected in the same area and ground plane? Also any recommendations for pcb mount differential connectors? Thanks.

 

[crutschow]

Normally a shielded twisted-pair is used for differential signals.
High frequency twisted-pair have a characteristic impedance of around 100 ohms.
Ethernet cable is 100Ω twisted-pair, for example, and you could use ethernet connectors.

I wouldn't recommend trying to use two coax for that.

 

[Novicent]

What SNR performance would you expect from shielded twisted pair? What about bandwidth for short cable lengths (<7”). What would be the concern with 50ohm single ended coax if the grounds of both cables are tied to the same nodes at both ends? Is the issue the tolerance on the impedance ? I’m looking for high performance (>70dB SNDR) over the 7”length of cable for frequencies <100MHz. Apologies for the basic questions. Thanks.

 

[crutschow]

What SNR performance would you expect from shielded twisted pair?

The cable will introduce little noise.
What is the nature of the signal and its level?

What would be the concern with 50ohm single ended coax if the grounds of both cables are tied to the same nodes at both ends?

It's just more complex, and any differences in the two cables will introduce differential distortion and noise into the signal.

If you have a differential signal, then you should use a single differential pair cable.

What are you using to detect the differential signal?

 

[Novicent]

This is for evaluation purposes. Signal will be 2.5vpp into 100 ohm differential. I would like a relatively flatband from 5MHz to around 100MHz. The differential signal will be sampled by another board with LPF and ADC. Would shielded twisted pair give a flatband response over that frequency band. What about loss at the connector? Are twinax cables/connector a better option in terms of better contact at the connectors? I don’t want the cable or connectors to degrade the performance. Thanks again

 

[crutschow]

This is for evaluation purposes. Signal will be 2.5vpp into 100 ohm differential. I would like a relatively flatband from 5MHz to around 100MHz.

For a flat band at that frequency you would likely want high speed shielded, twisted-pair and connectors such as these.

If obtaining those types are a problem, you could try the two coaxes and see how it works, if you have that option.
If so, make sure the two coax cable are as identical as possible, and both the cables, and connectors are 50Ω, terminated with 100Ω between the two cable outputs.
Physically tie the two cables together at frequent intervals with cable ties or lacing cord so there is little gap between them.

 

[Novicent]

Thank you. This is very helpful. What would be the purpose of tying the two coax cables together at frequent intervals? Since coax is shielded does it matter that the pairs are routed together in this manner?

 

[crutschow]

Since coax is shielded does it matter that the pairs are routed together in this manner?

If there is external radiated noise, it could affect the differential signal differently if the two cables are not close together.
The difference likely would be small, but keeping the cables close together will minimize it.