Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Circuit to emulate long twisted pair cable (1000m)

Status
Not open for further replies.

zainka

Full Member level 2
Joined
Nov 18, 2004
Messages
135
Helped
5
Reputation
10
Reaction score
3
Trophy points
1,298
Location
No(r)way
Activity points
1,648
In a test setup I need to emulate a long twisted pair cable of 1000m with 6dB loss, 100R /120R/ and 130R char.imp.
Are using a resistor setup today but this is just an very simple approach which only merely meet the attenuating characteristics for a twisted pair cable.
the line is a bidirectional data line sending data at 2048 kb/s.

Code:
----R1-----R1-----
       |
in     R2      out
       |
----R1-----R1-----

For 100R
R1~21.8R
R2~173.5R

**broken link removed** discuss a twisted pair model for 10BASE-T / 100m cable as found in ch.14 figure 14.7 and 14.8
Calculations is not shown and I would like to learn how to do this including compensating for length to be 1000m and not 100m.

Second, I might meet my criterias using a simpler approach but more up to the real cable than the my solution shown above.
Did also take a look at microwaves101 and their calculators, but its quite similar to my existing design which only considers the attenuation.
 

Hi,

I don't know about these things. I do vaguely remember - but not where I'm afraid - pictures of what data lines are composed of r.e. L C R.

Mightn't there be info somewhere on what telecommunications engineers/technicians can use/do to emulate a reel of cable or long distance cables, etc.? Maybe there's a formula to calculate inductance, capacitance and resistance for x metal/alloy and distance.
 

The cable model is designed to represent an empirical measured attenuation in a specific frequency range significant for the 10BASE-T media. It does probably not model the attenuation in the frequency range of your application exactly.

Once you know the frequency characteristic of your cable, you can adjust the model parameters to approximate it.

Alternative to an empirical measurement, you can estimate the frequency dependent attenuation based on skin depth.
 

Have been trying to read me up on the topic the last hours and learned that the design found in the IEEE document is quite limited to 10BASE-T specs only, i.e. similar freq/bandwith etc. Found this in another thread which also is discussing twisted pair equivalent circuits

https://www.edaboard.com/threads/357585/

There is also a lot of information found here: www.radio-electronics.com

However, by studying 10BASE-T solution (which is offcourse twisted and what I kinda want) It seems that to make a more accurate model of the twisted pair cable the task is to use a series of attenuators which also takes into account the characteristic capacitance and inductance of the cable used (Which I soon will have as soon as the manufacturer responds whit these values).

But, another lesson for me is that for a low speed data link, which I have, it might not be worth it to play with more complex models of the twisted pair cable. Thus the simple fixed resitive H-pad attenuator design that I am already using may be sufficient for mu needs. Or what does the expert think about this,,,

Other sources:
https://www.eeweb.com/tools/pi-match
https://www.electronics-tutorials.ws/category/attenuators
 

2 Mbps isn't particularly low speed when transmitted over 1000 m cable. Cat5 spec predicts about 30 dB (at least 20 dB referring to other specs) at 1 MHz which can be considered as fundamental frequency of a 2 Mbps date stream. You might worry even more about waveform distortion according to the frequency dependent attenuation.

It's not modelled by simple cable models, e.g. the lossy SPICE cable model. You need to consider √f proportional skin effect attenuation to get a halfway correct frequency and time domain representation.

Physically correct cable modelling is discussed in some previous Edaboard threads, e.g. this
https://www.edaboard.com/showthread.php?t=189432
 

When you say "emulate"... do you have anything against
picking up a fat roll of the actual cable on eBay? I did so
and now have a 1000-ft roll of CAT5 taking up a few
cubic feet of shelf space, job long since done.

If you're getting paid a lot to noodle over SPICE tline
model quality I believe you'd be well ahead to get some
hardware in the mix - at least then you'll have a point
of reference as to just what kind of wrong, your tline
results are. They're likely to be -some- kind of wrong.
 

I already have a bunch of roll's of cables but there is a need to have a system that could test the product without having to push these cable bundles around. And I am talking 3000ft not 1000ft. or 1000m to be accurate. And yes, I am paid a lot to noodle about this since there is money to spare to do this by emulating the cable rather than pushing 3000ft of real cables around.

Emulate...!?! maybe I should have been using the word "equivalent" in stead. English is not my native language so bear in with me.

Thanks FvM for the link, I will look into that.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top