Register to reply

Transmission line attenuation

by Simon.T
Tags: attenuation, line, transmission
Share this thread:
Simon.T
#1
Oct6-09, 09:03 AM
P: 15
Dear All

I am hoping for some insight into a transmission line problem I have been experiencing. I am quite inexperienced and unqualified in this area so if I am not clear please let me know.


In wiring loom assemblies my employer offers there is a twisted transmission pair. The conductors are terminated at either end into a soldered electrical insert. The assemblies come in various lengths (75, 50, 15 and 2.7m).

The specification of the un-terminated transmission wire (as quoted from the manufacturer data sheet) is <6dB/150m @ 20MHz. “Scaling” of this value seems particularly intuitive for determining testing limits for verification.


What we have noticed is that in the 75m loom, the measured attenuation is approximately half of the scaled manufacturer limit. In the shorter sections the measured attenuation is borderline or a failure; as I mentioned we are scaling the limits so that 15m testing limit is 1/5th of the 75m limit for example.

When we mocked up a very short loom (approximately 10cm) and terminated them into connectors, the loss through that assembly was MORE than the 2.7m terminated section.


It appears that the loss contributed by the connector inserts is not static. In fact, it seems the measured attenuation is a function of the ratio of “connector medium” to “transmission line medium”; the shorter the loom the more of an impact the connector has on the attenuation.


Is what we are seeing representative of typical transmission line characteristics? How can this behaviour be expressed/calculated?

From my studies I am familiar with the concept of standing waves, reflections and "ringing" due to impedance mismatching and I am aware there are a number of standard transmission line equations but I’m not sure how those apply here (see the first paragraph :)).


Absolutely any help would be appreciated. We are attempting to negotiate more lenient tolerances for the shorter looms that keep failing but we would like some basis for our recommendations.



Best regards,

Simon Thomas
Phys.Org News Partner Engineering news on Phys.org
Printing the metals of the future
New gadget helps the vision-impaired to read graphs
3D printing helps designers build a better brick
vk6kro
#2
Oct6-09, 09:29 AM
Sci Advisor
P: 4,016
Are you saying the cables are not terminated with their characteristic impedance?

If you did that, you would certainly get standing wave effects that would mask any attenuation effects due to losses in the wire.
A quarter wave at 20 MHz in free space is about 3.75 Meters.

What is the nominal characteristic impedance of this cable? Is it shielded at all or just a twisted pair?

What is a "soldered electrical insert"?
Simon.T
#3
Oct6-09, 10:21 AM
P: 15
Quote Quote by vk6kro View Post
Are you saying the cables are not terminated with their characteristic impedance?

If you did that, you would certainly get standing wave effects that would mask any attenuation effects due to losses in the wire.
A quarter wave at 20 MHz in free space is about 3.75 Meters.

What is the nominal characteristic impedance of this cable? Is it shielded at all or just a twisted pair?

What is a "soldered electrical insert"?
Hi, thank you for the reply.

The transmission line we are using is this http://www.gore.com/MungoBlobs/26/32..._PLFWI1279.pdf

The wire is not shielded.

During testing the transmission line is connected to a signal generator with a 132R output impedance and terminated into 132R at the other end which should match the transmission line nicely; I believe the loss is not resulting from this per se - when we look at the frequency response from 1 to 20MHz it is quite a smooth plot with little ringing.

The connector I am refering to is a gold plated brass/copper (not sure - I will find out) pinned, much like a Cannon/military style if you have seen them (http://www.connect-solutions.com/itt...nnector-10.jpg), with solder buckets on the reverse which the stripped and tinned transmission line is soldered in to.

I believe the things we are seeing is due to the transition between the transmission line and connector pins and the impedance mismatch between the mediums, perhaps you can confirm my assumptions.



Thanks again for the reply,

Simon

vk6kro
#4
Oct6-09, 10:49 AM
Sci Advisor
P: 4,016
Transmission line attenuation

So, at the far end of this two wire cable you have a multipin plug and also a matching resistor of some kind? At the same time?

It seems like one of these is faulty doesn't it?

At 20 MHz a resistor can be quite inductive. Do you know what sort of resistor it is?

have you tried testing for loss at a lower frequency, say 5 MHz? This would give you an idea of whether the loss is frequency related and if so, how quickly it happens as you change frequency.

How are you measuring the output voltage across the load resistor? With an Oscilloscope?
Simon.T
#5
Oct6-09, 11:35 AM
P: 15
Hi Vk6kro, thanks for the reply again :)

Forgive me if it seems I am witholding crucial information, I did not want to post a huge wall of text in the beginning.


The testing is done with an automated test computer. A signal generator with an output impedance of 50R is connected to a ferrite bead auto-transformer (to transform the output resistance to 200R), this is adjusted down to 132R with a resistor in parrellel (390R off the top of my head).

This output of the auto-transformer is connected to a female connector, which is connected to a male connector on the transmission line. On the other end, the transmission line is terminated into a female connector which is connected to a male connector on the recieving instrumentation.

The instrument we are using for measurements is an AD606 (http://www.analog.com/static/importe...eets/AD606.pdf) which forms part of the automated system. There is a resistor network on its input which presents a 132R input impedance.

All the boards used in this system is through-hole based; having read up on this it seems this is quite terrible for high frequency measurement instruments. All the resistors used are wirewound or metal film type.


Some history: this system has been used for over 15 years and it is only recently I have been involved in investigating this issue. I am certain the equipment is to blame at least in some part (one of the last EEs to look at this system was almost shocked by the design - in a bad way).

Before I waste your time any further, could you recommend a cost effective way to test raw telemetry wire with a seperate setup? We have a wide collection of lab equipment although unfortunately our Signal Analyser is primarily for acoustic measurements (0-102.4 Khz).

Thanks again for your reply :)
berkeman
#6
Oct6-09, 12:03 PM
Mentor
berkeman's Avatar
P: 40,718
Can you try using Cat-5 cable instead? Change the terminations to 100 Ohms if you try it.

I agree that 20MHz is a pretty high frequency to be using with connectors that do not maintain the cable Zo. If you eliminate the connectors and drive and terminate the wire directly, what do you measure for the attenuation.

Why is 20MHz used?
f95toli
#7
Oct6-09, 01:18 PM
Sci Advisor
PF Gold
f95toli's Avatar
P: 2,241
Also, is the impedance of the cable really independent of frequency? There is always some frequency dependence in real cables, and for something as "crude" as what you are describing the impedance can change quite a lot.
Remember that sqrt(L/C) etc only works for lossless cables, and the skin effect (among other things) give rise to frequency dependent losses that changes the impedance.
berkeman
#8
Oct6-09, 01:33 PM
Mentor
berkeman's Avatar
P: 40,718
I was thinking the same thing initially, but the Zo for most TP cables that I work with is pretty flat above 1MHz. Below 1MHz the Zo rises, and that can be important when terminating lower-frequency communication networks.
vk6kro
#9
Oct6-09, 06:00 PM
Sci Advisor
P: 4,016
Are you using an unbalanced feed on a balanced line?
(Is one side of the feed from the signal generator earthed?)

This would cause an increase in losses due to radiation.

If so, you would need to install a balun optimised for 20 MHz.
berkeman
#10
Oct6-09, 06:53 PM
Mentor
berkeman's Avatar
P: 40,718
Hah! Great point vk6kro!
Simon.T
#11
Oct7-09, 01:58 AM
P: 15
Wow, thank you for all your replies.

The line is being driven by a 1 to 4 balun, with the signal generator being unbalanced.


Forgive my ignorance by if the impedance was changing significantly with frequency we would see more ringing at higher frequencies? We don't see anything like that, the frequency is relatively smooth but with higher attenuation at the upper end.

1 to 20MHz is used for the test because it represents the maximum data transmission rate used by the final application. Our products are used in real time seismic systems which incorporate thousands of multiplexed channels - the bandwidth requirement is massive :).


What I will try today is testing a cat5 twisted pair (if I understand correctly) on our instrumentation and post some screenshots of the results.

I am going to make a very basic testing system independant of our automated equipment and perhaps perform a comparison.

I will report back with my findings, thank you all very much for your feedback and comments :)


Simon
vk6kro
#12
Oct7-09, 08:46 AM
Sci Advisor
P: 4,016
Reading back through your posts, I came across this:
All the resistors used are wirewound or metal film type.

Wirewound resistors are absolutely unsuitable for RF applications.

I set up a 5 watt 100 ohm wire wound resistor on an antenna analyser.
At 15.7 MHz it reached a peak impedance of 138 ohms and started to droop above that as it became capacitive.

I don't have many wire wound resistors but I did have a 47 ohm 5 watt one which I tested the same way and it reached 700 ohms impedance at 20 MHz. Even at 5 MHz it was 230 ohms.
It measured 47 ohms on a multimeter ohms scale.

These huge errors would make your readings meaningless if your resistors are as bad at RF as mine are. They would give you standing wave effects that would be dependent on the length of the cable being tested.


Register to reply

Related Discussions
AC V,I, and P Distribution on a Transmission Line Electrical Engineering 6
Transmission line Electrical Engineering 6
Transmission line termination? Electrical Engineering 11
Need Help on a transmission line problem 1! Engineering, Comp Sci, & Technology Homework 0
Antenna, transmission line Electrical Engineering 4