Any physical cable has low-pass characteristics due to the intrinsic resistance, capacitance and inductance of the cable. Thus you should expect higher attentuation for higher frequencies.

Hope this helps...

Thanks for your reply.

But would I be seeing a linear trend or a logarithmic trend?  

Srinidhi,

But what a wonderful situation you are in!

Take time to reflect on this.

You are gathering firsthand empirical knowledge about a process - it's now up to you to identify a plausible model that might account for whatever shape the curve of loss versus distance takes!

This is science in action - you have a hunch, perform a quick test to check whether what you are doing is rational, build a hypothesis, and then conduct an experiment to see if the data support your hypothesis or not!

So.

What data do you have so far? Do you think that a pulse would be different from a simple sine wave?

(think about what a pulse is actually made of - think about its spectrum)

And when you do have a fall-off curve (loss vs frequency) have a think about what that means in terms of how much of the conductor's cross-section is available to carry the signal at a given frequency.

I found some attenuation values for a rather special coax cable for you: http://friedrich-kusch.de/Koaxkabel/Aircell-7.htm  (sorry - the page is in German. But the values are international.)

Just plot and see :)

Thanks for the reply U. Dreher,

I got the following plot (attached). I see that the attenuation increases with the increase in frequency which i pretty much whats happening in my measurement results too.

I now this is a completely different question now, but do you have any idea about the attenuation was measurement and what was the setup?

Thanks.

Hello,

regarding the measurement setup: I really do not know, but can wage an educated guess.

Whether the measurements are made with a limited number of frequencies or by sweeping the whole spectrum is not clear, but not too relevant and mainly depends on the equipment available. As the attenuation curve has some characteristic form, both methods deliver similar results.

Regards

BTW: plotting the attenuation vs. a logarithmic frequency axis might provide additional insight: classical low-passes have a constant attenuation per octave resp. decade. A log-log plot would allow for a quick assessment on how similar to a low-pass the cable behaves.

Regards

Hey Dreher,

Yes I used a setup some in the pic and observed the amplitudes for frequencies: 1-15MeV and got a linear plot when I plotted Loss(dB) vs Freq(MHz).

Does it make sense?

Thanks.

May I interpret 'CRO' as oscilloscope? This might explain the 'jittery' curve. A HF wattmeter might come up with more precise values.

In the mean time I found a page showing the attenuation curve of a Cat. 5 cable (sorry - German): http://www.hbernstaedt.de/knowhow/steuersignale/steuersignale_04/kabel.htm

Comparing vs. this plot, your cable seems to be significantly lower specification. Anyway, seems you are on the right track...

Regards

Hi Dreher,

Thanks for that link.

Yes by CRO i meant an oscilloscope. Can watt-meter we used to measure the amplitude (or the power) of a sine wave? 

Yes I went through the line-damping section of the link you sent and it does look linear from range 1-15MHz similar to what I see on my plot.

Thanks.

Hi James,

Thank you very much for that insightful reply. 

Research has always been fascinating to me. Although it may be embarrassing to go back to the basics when at this stage at the end it helps me to work out the situation myself and use this strategy in another doubt I have. 

After this question, I went back to read about cable, their properties and their behavior which I may have skipped if I had no problem with my setup and if got the results in the first go. 

I figured out that a square wave is nothing but an infinite summation of sinusoidal waves and also am looking at how the cross section of a conductor decreases at high frequency and most imp why? 

I am sure in the next 2 years of my phd tenure and even thereafter I will be able to discover more answers which are there but not looking at them or which need a bit of stepping back 2 steps before going ahead 2 steps :)

Thanks very much again! 

Hi Srinidhi,

there are some differences between the Cat.5 'reference plot, and your diagram:

I agree: even such 'simple' things like cables bear a lot of challenges. Including the skin effect...

Regards

Hi Dreher,

I have attached the log plot and also the data-sheet of the cable i am using. I am not sure what category cable is it.

Thanks. 

Considering the datasheet, it could be a cable following Cat.7 - though with significantly more than the usual 4 pairs. I currently have no reference attenuation values at hand, and 'standard' Cat.7 uses AWG22 conductors (which I'd expect to have less attenuation), but overall... (The denomination AWG24/7 in the datasheet seems to be a bit 'unusual'.)

Hi Dreher,

I guess CAT7 is correct as it is also mentioned in the datasheet in the attenuation section.  Its a 18pair cable with outer shielding. 

The attenuaion mentioned in the datasheet is 1.94dB/10m hence 21.34dB for 110m at 100Mhz. If I use the slope obtained from my results to calculate the attenuation at 100Mhz it comes out to be 61.954dB which is 3 times more. 

What could be the reason for this? Is it because of the resistors I have added in the setup?

Please help.

Thanks. 

The resistors should not be a reason if you monitor the sine at input and output for resonance effects. I'd guess that the oscilloscope "measurements" contribute to bad results as well as the fact that your cable is so short that setup errors (eg. the scope wiring etc.) may add attenuation that is high with respect to the effects of the (comparably) short cable tested. There 8s good reason why the values are normally given for 100 m lengths.

Another reason may or may not be the fact that you are only driving one wire of the pair while standard testing might excite the pair differentially (as it is thought to be operated). I have no idea how much this might affect the attenuation figures.

BTW (as I'm constantly learning): do you have access to a network analyzer to perform the measurements in a more professional manner? This might give better and more reliable values.

Regards

1. Yes, since its a 110m long cable I have it coiled up on a table while i do measurements on the either ends. However I did read up on the inductance offered by the coil and calculated it using an online calculator (pic attached). How do I add this to my measurements (plot) and justify the attenuation?

2. Yes, currently I have connected the red crocodile pin from the function generator to one wire of a pair and the black one to the other wire of the same pair. What modification would I need to do? 

3. I will try to get a NA and observe the output on the NA instead of a CRO, it that right?

Thanks.

Regards

1. Does the setup qualify to be a parallel wire setup? The two wire in each pair inside the cable (18 pairs in total) are twisted while pair themselves are straight. So does the "wire" in the online calculator site denote a wire carrying a pair or the wire in each pair? (pic attached)

2. Do I still need the isolated transformer if I used a VNA?

3. I will ask my dept if they have a VNA and then will read about the issue in the manual.

Thanks.