How to choose the right frequency when measuring coil inductance with LCR meter

I would like to drive a 12V DC hydraulic valve coil using a high side driver (VN7140AJ). The driver datasheet specifies the maximum demagnetization energy (page 24), and I would like to make sure that I can drive the coil without damaging the driver.

The datasheet for the valve does not specify its inductance and the manufacturer doesn't seem to know what I am talking about, so I decided to measure the coil inductance myself. I have a DE-5000 LCR meter, but depending on the frequency I select, the measured inductance changes:

• 100 kHz → 4.681 mH
• 10 kHz → 7.578 mH
• 1 kHz → 21.79 mH
• 120 Hz → 52.44 mH
• 100 Hz → 55.71 mH

Knowing that I'll be driving the coil at < 1 Hz, what frequency setting should I choose on the LCR meter, to obtain a value that can be checked against the demagnetization graph in the driver datasheet?

• My limited experience with this kind of thing is that coils in valves are poorly approximated by the most basic "LR" model. In particular, the "inductive" part of the impedance curve deviates significantly from 90deg phase and 20dB/decade slope. So I would measure it at multiple frequencies. (if you keep going, eventually you may find the resonance after which it will be capacitive, and that freq. will change with DC load, but this freq. range is usually not relevant to valve applications) Commented Mar 5, 2021 at 20:28
• Probably with a frequency close to that that it will drive the coil. Also I do think you have a bar graph that shows the measurement accuracy on DER-DE. Bigger inductance needs lower frequency, small inductance higher frequency. Commented Mar 5, 2021 at 20:32
• The valve will be activated at random and very low frequencies (< 1Hz). I measured the inductance at 100 Hz, which is the lowest the meter can do, and got 50 mH. After checking the DER-EE accuracy table, the 200mH range is indeed most accurate at 100 Hz or 10 KHz.
– Toun
Commented Mar 5, 2021 at 20:53
• So what discrepancy you have at 100Hz and 10Khz? Commented Mar 5, 2021 at 21:02
• The solenoid is most likely iron cored and may not have laminations so it will get lossy at the higher frequencies. Probably best to test below 1kHz or use a pulse test to measure the inductance Commented Mar 5, 2021 at 22:40

simulate this circuit – Schematic created using CircuitLab

You have a large coil with lots of turns, each turn has its resistance and inductance. Between each turn there is also a capacitance, this is because the wires are close with each other, separated with a thin layer of insulator.

The LCR meter measures the phase shift voltage VS. current. At high frequency the capacitance becomes more and more dominant, so your LCR meter "thinks" the inductance is decaying.

You can see those effects, described in this paper.

That's a very similar scenario as yours. At low frequency, the inductance is dominant up to resonance point, then the capacitance is dominant. You can see also the resistance increase due to skin effect.

• Thank you. So, if my coil behaves the same as this one, as long as I'm measuring at a frequency lower than the resonance frequency, the measured inductance should be roughly representative (at least not off by an order of magnitude). Is that a safe assumption?
– Toun
Commented Mar 14, 2021 at 12:10
• @Marko - this would suggest his resonance is less than 1kHz? for a 50mH coil? I wonder if there is something else here... cap between layers of windings or magnetic losses? i would guess phase probably significantly under 90deg at f well below resonance. (i have observed this in valve solenoid coils but do not know how it translates into total stored energy, for this question) ... would be interesting to measure. Any thoughts? Commented Mar 14, 2021 at 13:49
• @Toun It depends on how DER-DE is interpreting these results, you may also look for Theta angle and other features of the instrument - Ls, Rs, Cp. But a 50mH coil is a huge one, lots of numbers of turns, tiny wire,... so definitely it has a big self capacitance and the readings are correct. Commented Mar 14, 2021 at 14:27
• Excuse me. In the linked paper there is a function -27.95dB =20 log(200mApp / 5Vpp). I have a really hard time understanding why is it referenced at 5Vpp. Commented Jun 24, 2023 at 13:36