Measuring Inductance

[mendip_discovery]

So I have an ELC Inductance Decade Box DL07 and it has a calibration certificate with it.

When measuring Inductance what things do I need to be wary of. For example, is it affected by temperature, magnetic fields etc.

The reason is that several meters I get into the lab have the ability to measure inductance and I am looking to build an Uncertainty budget for it and currently I just have the 5% spec for the inductors and the imported uncertainty. I haven't had much time to play with how test leads or even environmental things affect the measurements. During use to inductors stay fairly stable or should I expect them to move about much like resistors do as they get older?

[TimFox]

One imprtant thing to worry about is parasitic capacitance which will have a frequency-dependent effect on the measured inductance.
Another important thing is non-linear ferromagnetic material:  if the coils are wound on a ferromagnetic core, the inductance will be affected by the DC current flowing through the coil.  This effect is smaller if the coil has an air gap, for example a toroid with a small spacer or an E-I core with a paper spacer in the flux path.

[mendip_discovery]

One imprtant thing to worry about is parasitic capacitance which will have a frequency-dependent effect on the measured inductance.
Another important thing is non-linear ferromagnetic material:  if the coils are wound on a ferromagnetic core, the inductance will be affected by the DC current flowing through the coil.  This effect is smaller if the coil has an air gap, for example a toroid with a small spacer or an E-I core with a paper spacer in the flux path.

With regards to frequency causing a difference in the readings is that quantifiable in a formula?

What I am aiming for is reducing the errors in the measurement of these "known" inductors I have. I notice the certificate that I have has them measured at 1 kHz and almost any certificate I see states 1 kHz as the measured frequency.

The thing is are inductors known to drift from year to year? So is the 5% tolerance just what it is from nominal or its potential yearly drift as well. Because I could reduce the errors and refer to the certificate for the inductors. So far what I have seen from Schedules is labs often refer to the measurement capability rather than generation, suspect this is because they measure the inductance of the standards they use before using them.

Just reading up and I spotted this from UKAS,

In inductance measurements it is necessary to define connecting lead configuration and be conscious of the possible effects of an earth plane or adjacent ferromagnetic material.

[Kleinstein]

For the parasitic capacitance effect one usually has to get an equivalent model for the inductor. So one would not have just a number for the inductance, but also prallel capacitance and damping of some way (e.g. series resistance or parallel resistance). So ideally the calibration would be with more parameters, not just 1. The model circuit would give the inductance for at least some frequency range.

Things can also depend on the range: the smaller inductors would be without a core and thus not effected by a magnetic field, but possibly with iron near the inductor. To reduce this effect I would suspect reference inductors to come in a relatively large case, maybe with iron shielding.
Large inductance values (e.g. > 1 mH) would kind of need an iron-core. This would add errors and sensitivity to external fields and maybe hsteresis effects. So it is not just the DC current, but also possible past DC current to have an effect. The amplitude may have an effect too.

The higher end LCR meters can usually measure equivalent circuit parameters and include lead capacitance.

[mendip_discovery]

ELC DL 07 Datasheet

• Each decade consists of inductors wound around a ferrite pot core, which gives a Q factor between 55 and 100.
• Like all inductors of this type, the applied voltage mostly depends on the operating frequency. The voltage must be limited to avoid core saturation and overheating.
• Professional quality switches with gilt contacts.
• Screen outputs (front panel and case) on safety earth socket.

[jonpaul]

Inductance is a function of DC bias (if any), excitation (Ma or V from bridge) and especially Frequency.

ANy inductor measurement needs to at least sp0ec the freq and voltage applied.

External fields are normally 60 Hz or DC and do not affect bridges.

Finally the smaller inductors must have the test lead inductance zeroed out.  ~ 1 uH per sq ft area.

The precision standards are usually measured at 1 kHz and 100 mV ..1V.

Bon Chance


Jon

[Conrad Hoffman]

IMO, that's a bench type inductor substitution box, not a standard. Inductance standards are usually wound on inert cores, not ferrite. They change very little over time and very little with applied voltage, unlike ferrite cores. For this, I'd just go with the manufacturer's spec of 5 and 10%, unless you want to go through a very involved characterization process. FWIW, I've had similar boxes and have never actually found a use for one because they were incapable of the currents needed or because they couldn't do their job on multiple inches of wire away from the circuit.

[mendip_discovery]

It was bought as we had a job where a customer needed an old Wayne-Kerr unit calibrated ASAP. This was during lockdown etc. It did the job and the unit was all good. No adjustments were needed.

I have since used it a few times for Handheld LCR meters and them Atlas LCR things you see about. But I would like to build a budget and a procedure for calibration using this unit.

I would love to get some calibration standards such as the resistor, inductance and capacitance ones. The problem is getting the money to spend on it all.

[trobbins]

Given all the inductance elements are pot core based, with a nominal specified Q range, I'd anticipate you would need to firstly appreciate where first resonance starts to come in to play for assessing test frequency upper end, and where DCR dominates for test frequency lower end.  The easiest way I know to cheaply do that assessment is with a soundcard and software like REW using an impedance measurement setup to do automated frequency response sweeps to determine the frequency range most suitable for fixed frequency testing of each range of your box.  That type of test can also indicate the variation in response with excitation voltage level changes available from the soundcard.  That type of test can have an inductance accuracy sufficiently under 1% to allow characterisation, as it is based on reference resistor tolerance for calibration.

[1audio]

I would guess the inductance would vary with amplitude and frequency such that you would need a table with a lot of specifics (amplitude, frequency) to have any real confidence. Resistors and capacitors are better standards; fewer things that cause errors. Even passing DC through the inductors could magnetize them and change the inductance.
You can characterize them with an oscillator, a known resistor and a meter. Its pretty tedious to do manually.

[Le_Bassiste]

IMO, that's a bench type inductor substitution box, not a standard. Inductance standards are usually wound on inert cores, not ferrite. They change very little over time and very little with applied voltage, unlike ferrite cores. For this, I'd just go with the manufacturer's spec of 5 and 10%, unless you want to go through a very involved characterization process.

^this.

[jonpaul]

Aircore inductors have no nonlinearity and only the self C and R to contend with.

GenRad (General Radio) Boonton, Yokogawa and other high quality instrument mfg made standard inductors in varios accuracy and ind.

Finally Boonton Radio made Q standards for testing their Q meters,


Bon Chance



Jon

[Conrad Hoffman]

FWIW, General Radio Corp. 1482 standards can usually be found pretty cheap because supply is greater than demand, unlike their good invar capacitance standards.

[mendip_discovery]

IMO, that's a bench type inductor substitution box, not a standard. Inductance standards are usually wound on inert cores, not ferrite. They change very little over time and very little with applied voltage, unlike ferrite cores. For this, I'd just go with the manufacturer's spec of 5 and 10%, unless you want to go through a very involved characterization process. FWIW, I've had similar boxes and have never actually found a use for one because they were incapable of the currents needed or because they couldn't do their job on multiple inches of wire away from the circuit.

TBH this is what I plan to do. I can't spend lots of money on kit, this is just me looking down the rabbit hole.

1 kHz and 1V is what they have been measured at so that for me is what I would use them at. What I was trying to get to was do I have to worry about errors caused by EMF, Cables, Temperature etc. not so much the resonance frequency or effects of voltage.

The problem is when you are just given the 5% (or 10%) tolerance that could mean from nominal or annual drift. If its just the error from Nominal I can just use the last certificate measurement but then I need to account for drift, are these likely to drift by the above percentages? Until I can build up sufficient data I can't just assume drift from the two calibrations I currently have so I am using the last certificates measurement and percentages for the drift. Which once all the things I had added in are accounted for gives me a Unc of ~5.8%/11.5%, I'm not using percent for my actuals though.

[Conrad Hoffman]

The box spans 1 uH to 10 H so the materials and construction will  likely be different for the different switch banks (at least). I'd expect the stability, both long and short term, and tempco, to be quite different for different settings. Obviously cables will be important towards the ends of the range, and temperature towards the top. Much characterization to do!