How to electrically characterize a multi-wire shielded cable

The placement of the wires in the cables may be slightly different. Are the paires of wite withing the cables twisted in the same way? Measure the capacitance between all pairs. Capacitance of each wire to the shield.
What type of signals are you measuring?  DC?  Audio?  RF?  VHF?

Digital? so that bandwidth is important?

How fast are your digital signals?  How are they terminated?

If you put a single shield around multiple conductors you are not shielding the signals from each other.  If there are multiple ground conductors in the bundle, you are not controlling the impedance of the signal lines very well.

An LCR meter is not the best piece of test equipment to diagnose this problem.
A Network Analyzer or Time-Domain Reflectometer would be better.

Looking at the received signals with an oscilloscope would also work if you know what to look for.  You might see excessive crosstalk or ringing.

aburr, you may be right about the placement of the wires within the sheath, i dont have the ability to determine this, however, even if i split them apart.

this has nothing whatsoever to do with digital signals...  sorry if i wasnt clear about their use.

these cables are test cables for in-house test equipment.   they are used to measure impedance of a load, when an arbitrary waveform of approximately 50kHz is applied.    

the problem surfaces during an automated calibration procedure, after these new cables have been installed in this piece of custom test equipment, when the test equipment is check against some reference points to calibrate.....     the baseline reference impedance is *slightly* out of tolerance.  

i can not merely change the boundaries of reference impedance.   previous cables have worked for years, with no problems.    they are, for all we can tell, being manufactured to our own specifications, which only includes a "High-Potential" test (applying 20kV for some time and checking that leakage is less than 'x' amount of milliamps.

SIMPLY PUT, WHAT I NEED TO DO,

is characterize the impedence of this multiwire sheilded cable so that i can compare the complex impedance values of an "old good" cable with a "new bad" cable.

i dont need, or want to get into specifics, just a basic high-level how to...    using an LCR meter, taking a generic shielded cable with 6 separate conductors inside... how to go about characterizing the complex impedance of these conductors in relation to each other, within the overall cable.

Using an LCR meter:

With one connector entirely open:  Measure the impedance of each conductor to the shield and then the impedance between each pair of conductors.

    That should be 6+5+4+3+2+1 = 21 measurements.

The readings should all be capacitive.  
All the conductor/shield readings should be similar.
All the conductor/conductor readings should be similar.

Make the same set of measurement for a couple of new and a couple of old cables.

If you find a single anomaly, there may be a defect of some sort in the new cables.

If all the values have shifted slightly, you may be dealing with process variations:
Different amount of twisting, or different size or type of insulation.

>> would probably be do   to   changes ...
                                   due two changes
                                   dew too changes

thanks, d-glitch.   that's exactly what i'm looking for!

I don't quite understand your question.   Are the cables specified to have a particular surge impedance, and is that what you have been testing?  

Quite often twisted-pair cables have a somewhat loose specification on impedance, as it's hard to control that tightly between twisted pairs.  In a coaxial cable things are more stable and impedances can be controlled to within a percent or two.

But I suspect in your 50KHz system the exact surge impedance is not all that critical.  Perhaps your test equipment is being too finicky.  Impedance is only critical when your cable lengths more than 1/4 wavelength.   1/4 wavelength at 50KHz is about 3.5 miles.   I suspect you may not need to spec the impedance that closely if your cables are shorter than this.

>> "Perhaps your test equipment is being too finicky."

well, that's EXACTLY  the problem, methinks.

but it's not easy to just go and change the calibration program  for a system that has been in place for *years*, in a federally-regulated and traceable manufacturing environment.

here's the exact problem...   we have a complex load impedance in some test equipment that is supposed to be within a certain allowable impedance, Z = R + jX....  and when we go to calibrate our test fixture and measure this test load, we first reference our measurement equipment with  short- and open-circuit tests

the short circuit test notes the cable impedance, and the acceptable resistive component has long ago been declared to be between 0-5 ohms....   and  for these many years, they've always just been on this side of zero ohms.  

now all of a sudden these new cables' resistance are measured by the LCR as being slightly *negative* ohms (but still essentially zero).   this causes the test to fail, because the lower boundary is zero, and doesn't allow an absolute measurement

the question everyone asks is "WHY"... as in WHY are these cables now "FAILING" the test.

I understand that the LCR will measure "negative" ohms or capacitance or inductance in some circumstances, and I don't believe this negative direction is meaningful.   but first i needed to characterize the cables so that i could answer the inevitable question "what is wrong with the cables"

now i've done that.   and the answer, i think, is that "nothing" is "wrong" with the cables.  

but what i still don't exactly have an answer for is "well, then why are they all of a sudden showing negative impedance?   especially since they've "never done that before"

got any suggestions?

never mind about my last post.

i figgered it out.  

it has to do with the LCR meter accuracy, where non-lossy loads have impedanance (Z) error margins that are much MUCH lower than the error margins for the resistive (R) component.

depending on D and Q values, overall error margin for Z can be less than 1%, while the error margin for R can be greater than 200%  (wow)

so when i do a short-circuit reference point, i see now how easily it can be to have +/- 0.5 ohms, and the derived value can easily be less than zero.

the only question i have now, is why did it take 10 years to ever see this happen?

but that's rhetorical.  

i have to give full credit here to d-glitch.   he answered the question exactly as it was posed.

thanks all