I need to measure the characteristic impedence of short coax

[sigmanut]

I need to measure the characteristic impedance of some short pieces of coax. I've tried a Ten-Tec product but don't have confidence with it. I've heard of an antenna smith but have not been able to confirm that it will do the job. These pieces of coax are typically about 5' long and are typically AWG 40 (yes really), so making a great connection to the end is not like crimping a TV type coax connector on the end of it. Any advice? Thanks in advance.

 

[richs]

Hiya-

Well, it's a Smith CHART. A drafting tool (chart) used
to "swing" impedance though it's wavelength. Typically
used to include a "stub" impedance (short piece of coax,
waveguide, twisted pair, pieces of copper on a PCB, etc.)
to modify the impedance seen by either a source or a load.

Named after Phillip Hagar Smith, a diagram designed along
these lines was constructed in early 1937. It is
essentially the form still being used today.

Scaler impedance can be measured by taking the root of the
product of the unknown with the opposite end open by the
opposite end shorted. Z = sqrt(Zo * Zs)

Depending upon the frequencies of interest, these pieces
could show the source (or load) anything from a dead short
(AC) to an open impedance. The goal is to "match" the
impedance of the transmission line to the impedance of the
source/load.

Here's one link:

http://www.maxim-ic.com/appnotes.cfm/appnote_number/742

On a google search I ran into the Agilent site which has
the following little phrase:

'"Smith" is a registered trademark of Analog Instruments
Company of New Providence, NJ 07974.'

Hope that this helps getting you started.

Cheers,

Rich S.

 

[MacGyverS2000]

The Smith chart is also trademarked by the executor of Smith's estate. Just a side note...


Dan - Owner

http://www.Hi-TecDesigns.com

 

[jimkirk]

To actually measure the impedance of your cables, a time domain reflectometer is probably your best bet.

Explore

http://www.emcesd.com

for some ideas as well.

 

[Comcokid]

A TDR is what you need. As to the connector, I would just strip one end and connect it to a SMA or smaller connector of the end-launch or PCB type by soldering the exposed ends to the connector with as short a connection as possible. A TDR should allow you to see the impedance Vs time (length) so you just cursor the measurement delay to the cable and avoid your connector/coax splice.

There are general purpose cable TDRs (for field use) and then there are the better ones (for bench use). I've only extensively used the bench type (Tek 11801B w/S?? module), and I know they can easily make the measurement even on a piece of coax a couple inches in length. However, I've not familiar the general purpose TDRs and don't know how accurate they are and how well they can measure a shorter piece of coax. Maybe someone else can recommend a specific general purpose TDR model.

 

[jimkirk]

How accurately do you need to know the impedance?

A quick and dirty TDR can be done with a square wave oscillator with a known output resistance.

http://emcesd.com/50mhzosc.htm

shows an example. Connect it to your cable and observe the resulting steps on the edges. From the amplitude of the steps, the amplitude of the unloaded source, and the value of the source termination resistor, you can calculate the approximate cable characteristic impedance.

If you can adjust the resistance so the over or undershoots disappear, then that value of resistor is equal to the characteristic impedance.

Just remember that for 5 feet of cable, you're talking nanosecond/GHz range stuff, so for example, changing the resistor value as mentioned above means removing and soldering in a small SMT resistor.

I once built a box like that using matched high speed MOSFETs as the resistor elements. There were four devices, with the Vgs all set equal, so one was the source termination, one was the load termination, one was used to control the effective resistance of the others. Had to use fairly small signals so the MOSFET channel resistance was relatively constant.

 

[logbook]

Unless you can make a decent coaxial connection at one end (at the very minimum) you can't nicely measure the cable impedance. What you will measure is the connector/cable interface impedance. Of course if your cable isn't 50 ohms and you connect it to a 50 ohm connector, that would be wrong too! Typically you would measure the dimensions of the cable and get the dielectric constant from the manufacturer to get a starting point.

You can’t actually make a good measurement of anything until you know roughly how much big the answer will be! Usually it is obvious what range the result will be in. Sometimes you may have to make one measurement to get the rough range then make another measurement.

Suppose I asked you to measure a length accurately. Would you use a micrometer, a ruler or a tape measure? Unless I told you the context/application you might use the wrong tool for the job.

 

[sigmanut]

Wow! This is my first post of a question and I didn't expect to get as many answers nor as qualified answers as these. Thanks to all of you.

These wires are in the 40 to 44 ga coax wires in ultrasound probes. They are typically 50 ohm and I suspect some are 75 but some I've checked indicate 96 ohm. I have one wire that I know to be 50 ohm and know the velocity factor for it. So I'm going to test the length first and adjust the velocity factor until I get an accurate length reading. Then I think the impedance should indicate properly. That will allow me to validate the 96 ohm reading.

I may have to look into renting one of the bench top units mentioned or find a testing lab to get me started. (If you know of one let me know please.) Then I would have more samples to validate my ability with the Te-Tec. The Ten-Tec just may not be sophisticated enough to do this job. It's in for a cal check and I've sent a sample wire and question list to them. It does TDR and Smith chart results and I've learned to read the smith charts fairly well.

Thanks again to all of you.