ET432 LCR Meter - Wild Results - Range Function?

[killingtime]

This is depressing.   I thought I had a decent brand/model picked out....

Not so fast ...

I remember reading some posts on a guitar amp forum about valve output audio transformers. They love to measure them because the output inductance on the amp forms a low pass filter in the audio chain, and that affects the way it sounds. Conclusion was the level a transformer is driven at alters its electrical properties.

I was taught that as the AC voltage on an unloaded transformer primary (or inductor) increases for a given frequency, the core becomes more saturated with magnetic flux, and this affects the permeability of the core (assuming it's not air). As the permeability goes down, so does the inductance as the two are directly related. This means less reactance (Ohms) and the transformer will pull more current as you increase the driving voltage on the primary, even if the secondary is unloaded.

Spent last night searching the web for some charts on transformer\inductor driving voltage and inductance but couldn't find anything. This topic isn't discussed much, but it's important for LCR meters as you'll see. So I ran my own tests today.

Results and data attached.

I'm driving the original transformer (unloaded) with a variac, all the way from 300mV to 250V AC RMS. Recording the voltage and current using two DMMs. Reactance (XL) calculated using Ohms law as we know the current through the Xfmr and the voltage across it. Inductance then calculated using L = XL/2(pi)F.

The inductance starts off low and increases, peaks then decreases as the core starts to saturate. Did the same test again for an MOT I had lying around, as I know these saturate easily.

I don't know what's causing the low inductance at low voltages, but the figures don't lie.

I also ran 600mV AC RMS from the variac as this is used by the ET432 LCR meter (see post above), and got 0.68H. That's not too far off the 0.528H I got in AUTO on the LCR meter. A 22% difference, which sounds poor but when you look at the data I collected the inductance hits 7H @ 100V, so about 10x that.

What does all this mean? You need to drive an inductor or transformer at the voltage it will be used at to determine its actual inductance.

This brings LCR meters back into focus. They only output about a volt, so they're not going to tell us a whole lot unless the target circuit runs at that voltage as well. I'd hold off on that 1K bench LCR meter for now.

Incidentally, capacitors have a similar issue with voltage as well, but depends on the dielectric used. As you increase the bias voltage, the capacitance goes down, so you need to test them at working voltage.

[Randy222]

This is often done to reduce thermal EMF effects.

Best,

The tiny handheld, DC near 200mV, will run into a thermal EMF effect issue?
2nd, we know the meter cant do 200mv when measuring 1m-ohm, I just don't see how it can drum up 200A.

[mawyatt]

This is often done to reduce thermal EMF effects.

Best,

The tiny handheld, DC near 200mV, will run into a thermal EMF effect issue?
2nd, we know the meter cant do 200mv when measuring 1m-ohm, I just don't see how it can drum up 200A.

As said this is often done, whether this LCR meter does this we don't know, as we don't have one.

The lower the impedance measurement the more influence Thermal EMF has on the measurement. Thermal EMFs can easily produce many microvolts which for example measuring 1m-ohm with 10ma current (what our Hioki IM3536 can produce), that's only 10uV across the DUT, so a few microvolts due to Thermal EMF can introduce significant errors..

Obviously the LCR meter measuring 1m-ohm can't produce 200amps, what if you short the test leads!!!

Seems you need to study up on how these LCT LCR meters actually work, how measurements are made, and how these relate to the parameters displayed on these LCR meters.

Best,

[csuhi17]

Has anyone figured out how to properly calibrate this device?
That is, I mean the correct use of the CAL button.

[Phil1977]

Has anyone figured out how to properly calibrate this device?
That is, I mean the correct use of the CAL button.

I get the best results if I use the calibration function twice. Once with the short circuit bridge and once without.

I also have no idea why the instrument uses this square wave for the DCR measurement. I´ve never noticed these problems before I´ve read about them in this forum. So far my ET432 did and does a good job for characterizing caps and inductors over the frequency range. But the more I read about it the more difficult it gets to interprete its results 

[csuhi17]

Thanks,
I noticed the above-mentioned problems because I measure a component at multiple frequencies.
Also, before I start using an instrument, I make several test measurements. It's good to know when it's lying.

It is possible that I make a mistake during the measurement, or maybe my instrument is faulty.
As I can see there are many versions.

I understood my question to mean that if:
I use the kelvin clip, then I do the calibration with the kelvin clip.
I use the crocodile clip, then I do the calibration with the crocodile clip.
I insert the component to be measured directly into the two grooves, then I do the calibration with the added U plate?

Or

in any case, do I have to use it with the supplied U copper plate?

My measurements were very different from the tolerances in the manual.

I just ordered a UT622E because of its special price, when it arrives I will compare the two instruments and keep the more accurate one and pass on the other...

If the result is not correct due to my fault, and I measure something wrong, it will become clear whether it is my fault.

[Phil1977]

Please report about your findings - it´s always good to get experiences from other users.

I always do some plausability checks with my unit. I have a few high quality foil capacitors, air coils and metal film resistors that measure quite stable over the frequencies available at the ET432. As long as I get consistent results there, I can trust the instrument that it finds frequency dependent variations at other components or measurement setups. That´s what I need the LCR meter for and for this purpuse I think the device is performing quite well.

And regarding the calibration: Personally I always calibrate without test leads, using only the provided U-Plate. If I test then with leads, I expect the results to have an error due to the electric properties of the leads. If I need more precise results, I desolder the component and test it in the grooves, or I attach leads as short as possible.

But principally the way you describe (calibrate with the used test leads) should work too - at least at the lower frequencies. Especially at 100kHz the test leads can easily resonate, and then I think even the most expensive LCR meter can not be precise.

[mawyatt]

Has anyone figured out how to properly calibrate this device?
That is, I mean the correct use of the CAL button.

I get the best results if I use the calibration function twice. Once with the short circuit bridge and once without.

I also have no idea why the instrument uses this square wave for the DCR measurement. I´ve never noticed these problems before I´ve read about them in this forum. So far my ET432 did and does a good job for characterizing caps and inductors over the frequency range. But the more I read about it the more difficult it gets to interprete its results 

As an example to help understand the use of a precision squarewave to measure a low value resistor with an LCR meter in DCR mode.

Consider a DUT resistor of 10 milliohms (0.010 ohms) and the LCR meter produces a current of 10ma to measure the DUT in DCR with a unipolar drive. The test leads have a Copper to Nickel interface somewhere throughout the measurement "loop" that experiences a 1C temperature gradient across the connection. Then this produces a Thermal EMF of 10uV due to the Seebeck effect for CuNi of 10uV/C.

The LCR meter reads a voltage across it's inputs of 10ma * 0.010 ohms (DUT) + 10uV (Thermal EMF), or 110uV and computes the DUT value as 110uV/10ma, or 11 milliohms.

The LCR displayed the measurement as 11 milliohms, which has a 10% error due to Thermal EMF

Now the user switches the LCR meter to use a Bipolar Squarewave excitation current of 10ma peak. The meter subtracts the negative voltage readings from the positive voltage readings across the DUT resistor and averages the two readings.

So the meter reads for positive DUT voltage +10ma (positive excitation) * 0.010 ohms (DUT) + 10uV (Thermal EMF), or +110uV as before, then the excitation current polarity is switched (squarewave) and the meter repeats the DUT voltage reading of -10ma (negative excitation) * 0.010ohms (DUT) + 10uV (Thermal EMF) or -90uV.

Now the LCR meter subtracts the negative excitation reading from the positive reading and divides by two to average the readings as {-(-90uV) + (110uV)}/2, or 100uV.

The LCR displays the measurement as 100uV/10ma, or 10 milliohms without the Thermal EMF induced error

Please note the polarity of the Thermal EMF makes no difference in the final rendered LCR displayed measurement, nor the magnitude of such!! Multiple Thermal EMFs can occur throughout the measurement "loop" due to various conductor interfaces, but ideally their effects are neutralized by this technique. Also note the Thermal Induced error tends to become smaller as the LCR measurement voltage increases, either with higher DUT currents and/or higher resistance valued DUTs.

Edit: Should also add that with low resistance measurements the higher excitation DUT currents cause self heating due to I*I*R effects, and this alone produces Thermal Gradients across the DUT connections and other connections throughout the measurement "loop", and the Bipolar Squarewave Technique also helps with these self induced thermal effect measurement errors.

Anyway, hopes this helps and shows the value of using a Bipolar Squarewave DUT excitation for low DCR measurements. Believe most quality LCR meters offer this DCR mode, in addition to the typical fixed polarity DCR measurement mode.

Best,

[Phil1977]

Of course, for precision milliohm-meters this approach sounds good, but the ET432 is a low-end LCR-Bridge with an add-on-DCR-function. Yes it provides 4-wire-measurement, but beside this I do not see any special point that makes it better for that than a jellybean multimeter.

And on the other hand, in the case of my device it´s not even a bipolar squarewave but switching between around -0.2V and -1.0V. Of course, if you calculate the differential resistance over two voltages the thermal EMF will cancel out, but if your development target would really be to measure low resistances with low currents, then you should obviously use a bipolar wave as described by mawyatt.

Anyhow, there must be *some* reason it is designed that way, so thanks again for the explanation!