Bench LCR Meter Observation

[mawyatt]

While making some measurements of low value power resistors we observed an interesting characteristic with our TH2830 and IM3536 Lab Grade Bench LCR Meters.

https://www.eevblog.com/forum/projects/power-resistor-characterization/

For very low impedance levels with Kelvin Clips that were properly Open/Short Calibrated the results were agreeable for equivalent series R with a direct connected leaded fixture (B&K TL89F1) also properly calibrated EXCEPT the equivalent series inductance was often incorrect, being negative at even lower frequency measurements.

This erroneous result was traced back to the actual LCR meter's ability to INCLUDE the fixture Cable Length Compensation which with the Kelvin Clips is ~1 Meter.

The Tonghui TH2830 has the parameter in the setup but its not activated, the Hioki IM3536 also has this parameter and its selectively activated.

The Hioki gave proper results throughout the ranges, including using the Kelvin Clips beyond their specified range of 100KHz, used to 1MHz with surprisingly good agreement with the direct TL89F1 fixture results.

To prove to ourselves this Cable Length Compensation was the root cause and not something else, we purposely calibrated the Hioki IM3536 with the Cable Length Compensation turned OFF. It barked at us, stating the Calibration was in question after completing such, so the IM3536 was monitoring the phase shift and noted the shift excessive and not compensated, smart meter

When we made low resistance Kelvin measurements with the IM3536 with Cable Compensation Off, the results exhibited the negative inductance (capacitance) as "seen" with the Tonghui TH2830 (no Cable Compensation available) when also using the same Kelvin Clips.

The math behind this is within the actual computations as the meters don't actually measure R, C, or L, they measure Vector DUT Voltages and Currents and compute all the displayed parameters, and part of these computations involves calibration corrections which include Cable Length Corrections in some LCR Meters.

Anyway, the Cable Length Compensation is an important parameter for LCR meters that are expected to return accurate results with low impedances with Kelvin Clip type cables.

BTW, know someone is going to say, "RTFM" but we are too lazy, as for sure it's somewhere in some manual

Best,

[KungFuJosh]

Thanks for RTFM so I don't have to. 😉

[Martin72]

I would replace "Observation" with "Cable Length Compensation" in the title.
But yes, you shouldn't underestimate it.
Interestingly, compensation only plays a role if the LCR also has this option; the other LCRs probably have a fixed value in the calculation.

[Sensorcat]

Anyway, the Cable Length Compensation is an important parameter for LCR meters that are expected to return accurate results with low impedances with Kelvin Clip type cables.

The same applies to anything connected to the LCR meter port with a cable, not only Kelvin Clip type cables. For instance, tweezers, port extensions like the HP/Agilent/Keysight 16048E, 16451B Dielectric test fixture and more exotic stuff. Instruments like the E4990A have a compensation for which the user has to specify the cable length, and how to perform calibration.

This is a RTFM corner case, as the instructions how to do this are usually deeply buried in the manual, which has, for the E4990A, 685 pages.

But it is worth to memorize when using such an instrument that they all directly implement Ohm's law Z = U/I, with all three quantities being complex. Since a complex equation is worth two real equations, two independent real parameters can be calculated from the result. With a properly selected device model, useful estimations for L, C, R, Q, etc. are displayed. But these are only true if the model is appropriate. The measured quantitiy is always Z = |Z| · exp(jφ).

[mawyatt]

Don't think the actual complex Z is measured as magnitude and angle, but more the complex voltage directly across the DUT and the complex current directly thru the DUT.

This scheme doesn't actually measure the DUT current but the result of the DUT current being "forced" thru a scaling transimpedance amplifier to "convert" the DUT current into a measured voltage which represents the current with appropriate scaling factor.

How the next part works to create the desired result is uncertain/questionable, at least for us.

Multiplying these pair of voltages by +- 1 in a Bi-Phase modulation scheme to "Direct Down-Convert" the results to Baseband as In-Phase and Quadrature which contains all the information needed to mathematically create the desired parameters after Low Pass Filtering seems logical. This popular communications method (Bi-Phase Modulation/Demodulation) has the advantage of excellent noise and interference immunity, and can be easily synced with the stimulus signal which likely is created with a DDS.

Another similar scheme would use actual Sine and Cosine wave-forms as the multiplicands rather than +-1

We are not sure "how" these schemes are implemented tho, could be in hardware, or software, or a combination for both types, or not at all!!

If we had a schematic of a modern Lab Grade LCR meter, the operation could be traced out with some certainty.

Maybe folks that actually know (not just a Google search) could comment.

Anyway, glad we found this issue with the cable length, at least for our measurement sanity 

Edit: To put these low impedance measurements into perspective, if we consider a 1 ohm resistor with 10nH series inductance. Measured @ 10KHz, the inductive reactance is 628 micro-ohms, and the impedance Z is 1 + j0.000628ohms, or 1.00000020 ohms @ 0.036 degrees (that's a difficult measurement, see image)!! The cable* introduces ~0.0182 degrees phase delay @ 10KHz to DUT and 1.82 degrees @ 1MHz.

* Assuming a 1 meter cable with 0.66 propagation factor.

Best,

[rf-messkopf]

We are not sure "how" these schemes are implemented tho, could be in hardware, or software, or a combination for both types, or not at all!!

In this video the designer of the old Hameg/R&S HM8012-2 explains the block diagram:

When we made low resistance Kelvin measurements with the IM3536 with Cable Compensation Off, the results exhibited the negative inductance (capacitance) as "seen" with the Tonghui TH2830 (no Cable Compensation available) when also using the same Kelvin Clips.

Have you experimented with a full open/short/load compensation, using a known impedance as a load (which may be difficult to realize)? It is my understanding that at least the R&S HM8118 also compensates for phase delay of cables when open/short/load compensation is used. I just checked the HM8118 manual as well as the manual for the newer LCX meters, however, they are silent on the actual math behind that compensation. No idea if that is equivalent to a full one-port compensation of a VNA with three known standards. I would have to make some experiments...

[gamalot]

My NF ZM2372 has a cable length correction key which can be found on the front panel, so I don't have to RTFM.

It's just a rough compensation, choose one of the fixed lengths of 0/1/2/3/4 meters.

[Sensorcat]

Don't think the actual complex Z is measured as magnitude and angle, but more the complex voltage directly across the DUT and the complex current directly thru the DUT.

Dpends how you define 'measurement'. The LCR meter measures complex Z in the same way a DMM 'measures' resistance. My intention when I wrote that it measures Z was to express that one should not think L, C, R are the primary parameters or that anything in the measurement part of the instrument is altered if you change the display parameters.

This scheme doesn't actually measure the DUT current but the result of the DUT current being "forced" thru a scaling transimpedance amplifier to "convert" the DUT current into a measured voltage which represents the current with appropriate scaling factor.

Like a thermocouple, which does not actually measure temperature, but voltage...

How the next part works to create the desired result is uncertain/questionable, at least for us.

There's no reason to speculate about that, or to regret that we cannot know, because the 'Impedance Measurement Handbook explains' it:

https://www.keysight.com/de/de/assets/7018-06840/application-notes/5950-3000.pdf

It's a relic from the times when HP told almost everything about the internals of their instruments. Nevertheless, Keysight still publishes this precious resource. I discovered it when I had to design my first signal conditioning circuit to interface a gas sensor that has a variable impedance. Long ago.

[rf-messkopf]

https://www.keysight.com/de/de/assets/7018-06840/application-notes/5950-3000.pdf

It's a relic from the times when HP told almost everything about the internals of their instruments. Nevertheless, Keysight still publishes this precious resource. I discovered it when I had to design my first signal conditioning circuit to interface a gas sensor that has a variable impedance. Long ago.

Thank you. That is a really good resource. It also confirms that a full open/short/load compensation will account for any linear circuit between terminals and measurement plane. At least if it is implemented properly in the firmware.

[mawyatt]

How the next part works to create the desired result is uncertain/questionable, at least for us.

There's no reason to speculate about that, or to regret that we cannot know, because the 'Impedance Measurement Handbook explains' it:

https://www.keysight.com/de/de/assets/7018-06840/application-notes/5950-3000.pdf

It's a relic from the times when HP told almost everything about the internals of their instruments. Nevertheless, Keysight still publishes this precious resource. I discovered it when I had to design my first signal conditioning circuit to interface a gas sensor that has a variable impedance. Long ago.

Thanks, the Keysight Handbook does a good job of discussing the various methods. Wish KS would still publish details about their instruments as in the old HP days, but that's long gone now with the immediate bottom line mentality all OEMs march too

After reading it appears our Hioki IM3536 utilizes the more complex version of the Auto-Balance Bridge technique, and the Tonghui TH2830 may employ the simpler version. Some day we'll probably dive inside these meters to see whats going on!!

Best,