transmission lossy line for long Cable

[sh-eda]

Hi

I have a general question about the standard model used for cable lossy transmission lines.
I am looking redesigning a product that drives a signals down very long cables, ie could be well over a kilometre and various configurations. (Ultimately this is due to IC obsolecence). The frequency I can use is fixed, around 100-200kHz region, as it has to work with existing system. I have been doing some testing on different types of cables and configurations to work out the various parameters (characteristic impedance and . Unfortunately I do not have access to cables this long. The maximum length I can currently test is the approx 20-30m. So I am looking at using the data I have and simulating the various cables and configurations.
The things I am finding is that the model works fine when looking a frequency response for a sinewave, but doesn't seem to work when looking at cable propagation delays and reflections. i.e for a pulse or squarewave. I am wondering whether this is true or not? or maybe I need to increase the number of inductor capacitor elements?
I am currently using Proteus ISIS, but have also tried multisim.


Thanks.

 

[FvM]

As discussed in various previous threads, the lossy transmission line models provided by most circuit simulators simulate the frequency dependant losses and particularly time domain response rather poorly. The usual RLGC model doesn't model skin effect losses, which are the dominant frequency dependant mechanism correctly.

Aplac is the only tool with an acceptable lossy line model I'm aware of.

I achieved acceptable simulation results in some cases by applying a laplace transformation to a transmission function with √f proportional attenuation in LTspice. There are some limitations of the laplace sources by nature, due to the limited transformation time range, thus it should be handled with care. Supplied unsupported and as is!

https://www.edaboard.com/threads/189432/
https://www.edaboard.com/threads/119198/

P.S.: I don't know about the features of ADS for transmission lines. The distributed circuit components manual doesn't tell clearly about the implemented models.

 

[etmabreu]

That is happening most likely because the cable is a dispersive medium, i.e., the wave speed is different is not equal at all frequencies, but is a function of the frequency.

 

[FvM]

That is happening most likely because the cable is a dispersive medium.

Not the dominant effect I presume. Neither for coaxial cables up to GHz, nor twisted pair telephone cables at lower frequencies.

 

[sh-eda]

Apologies for the delay, thanks for the quick replies, and the links.
I am going to have a look at the software suggested.
I have since found some models for transmission lines for the software I am using. So I will also see how accurate they are.
I guess I was expecting to see
I assume the wave speed varies with frequency because the propagation delay = impedance * capacitance?

Is it possible to model or calculate the amount of noise generated in a long screened cable. (need to do more research). I know about the thermal noise (√4kTBR) but the resistance of cable is not that great.

 

[FvM]

I think, the frequency dependant cable behaviour can be sufficiently described as a variable attenuation. Group delay will be nevertheless affected.

There will be frequency dependant noise according to skin effect related Rac. But I don't think that it's an important problem compared to external interferences.