Tonghui TH2821A LCR Meter

Aargh! Make that .078A or 78mA. Sorry!

Actually that's pretty good.

I have an office flourescent with two large tubes mounted in the ceiling about 3 ft above the shelf the Tonghui is on. It doesn't seem to have affected the readings on the 10uF cap, but I'll run a test to make sure.

I'll also try to find a variable cap in your range and see if I can duplicate your results. I also have a smaller desk flourescent lamp that I can move much closer.

The leads to the Kelvin clips are shielded, but it would be nice to know if there's any leakage.

I'm quite interested in this topic - any sensitivity to noise could have undesired results. Thanks for the headsup.

OK, I was thinking the discrepancy went the other direction. That is a curious effect. I have lots of loose transformers. I'll check a couple and see if I get similar results.

Aren't you thoughtful!

It's interesting and well worth pursuing. I'll try some checks when I'm waiting for some other process to finish.

Thanks, John. Very good info.

Mike

Indeed! No complaints from this end.

I've done a couple of investigations since yesterday to try to get to the bottom of this.

First, I checked my Tonghui data against an old HP4260A Universal Bridge, bought on eek-Bay and long out of calibration. I can produce the data if you wish, but the crux of it is that they agree with each other.

Well, I tell myself, the transformer must be producing unexpected data.

First, I checked the Tonghui for output voltage when the transformer was attached. The lower frequency settings (exclude the 10kHz setting) produced 98 to 141mV RMS.

I then connected the transformer to an HP audio oscillator set for 60Hz and an HP3400A RMS meter. The results were not only surprising to me, but they were good support for the believability of the data produced by the Tonghui.

E(source) V(L) XL(calc) L(calc) 5 3 450 1.2H 3 1.5 350 .92H .1 .038 247 .66H

Wow! The inductance goes down as the excitation voltage decreases?

Remember, the Tonghui measured .52H even on the 1kHz setting. That's plenty good for me.

What this tells me is that I cannot trust *any* RLC meter to give me the inductance of a 60Hz piece of magnetics at low excitation voltage.

I have a lot more to learn about magnetics. But, the Tonghui is not at fault and I need more education.

Cheers, John S

That's very good news. Thanks for the info.

That was my feeling also. I was concerned there might be a bit of residual magnetism left after running on 120VAC. I thought that might bias the transformer partway up the BH curve and give spurious results.

My plan was to measure with the Tonghui, measure on 120VAC, then back to the Tonghui again. But too busy at the moment to start it.

Is that measuring the output of the Tonghui across the primary of the transformer?

What is V(L) - the voltage across the primary? Is that using the ouput impedance of the sig gen in series with the primary? What is the output impedance of the sig gen, and the resistance of the primary?

In the old days, we used transformers to couple the output of 50B5's and 6L6's to speakers. These transformers used laminations very similar to power transformers. In fact, if you were in a pinch, you could use a power transformer as the output transformer.

We also used transformers at the input to audio amplifiers to break common-mode hum problems, such as PA systems and intercoms.

So the transformers had to be pretty linear wrt amplitude, else we would see plenty of articles in Electronics World about reducing low-level distortions.

OTOH, today's power transformers may use different steel, which may have nonlinear characteristics or some residual magnetism.

Recall the power transformer in a microwave oven is running pretty close to saturation to minimize the amount of iron needed to reduce cost and weight.

This is becoming a very interesting subject. Thanks for bringing it up and for the good data.

I have studiously avoided learning anything about magnetics for over 60 years. But it looks like my golden days of ignorance are about to come to an end. I had already come to the conclusion it was time to start learning about magnetics since I need to do some work on pwm power supplies for a new product. It looks like you have accelerated the process:)

Thanks John. Very interesting info.

Mike

Yes.

Yes to your first question. Output resistance: 600 ohms. DC primary resistance: 11.5 ohms.

Usually, audio used thinner laminations than 60Hz transformers. The lams could be had much less than .014 inches.

Yes. I think the lams run about .014 thick.

You're welcome. I think this is now outside the Tonghui subject into the subject of magnetics. The lesson, though, is don't trust LCR meters with their low excitation to give valid results on large magnetics.

Cheers, John S

That's good info. Too low to turn on bipolar junctions.

Thanks. Primary resistance is not going to have much effect. But what is the last line showing 247 volts and 0.66H? Is that from the previous 220VAC measurement? If so where do you live that you have 220VAC in a wall socket?

It looks like you have opened a Pandora's box issue. We need to learn more why a low-level measurement doesn't work on laminated cores. What is the mechanism that is causing the discrepancies? Are there any web links that discuss this in greater detail? Does the same thing happen on toroids?

This is a very important subject for the Tonghui. If you want to gap a core, you need to know the inductance you should get after gapping to confirm it is correct. If the Tonghui can't measure it, I think we're in trouble.

Thanks,

Mike

The 10kHz setting produced 217mV while attached to the primary. Still low enough.

Unloaded output from the Tonghui:

Hz Vrms

Some loading is to be expected. But, still low enough even at no load. Good!

I think it is due to line wrap with all the >'s building up: Original was...

E(source) V(L) XL(calc) L(calc) 5 3 450 1.2H 3 1.5 350 .92H .1 .038 247 .66H

All good questions. I have my suspicions, but I'll do some research before I present them.

Well, yes, but not just the Tonghui. This seems to apply to any RLC instrument that uses low excitation. At least, so it seems for now (Two in a row). I wouldn't shun further experiments and data.

Thank you, John S

It's all in the "Slope of the BH curve". At low currents the slope of the BH curve is more vertical and turns horizontal at higher current. See the link,

Click it, it opens to a pdf, then it is large enough to see the slope of the initial ramp up, ( the line that starts at 0,0) which is what I think is similar to the slope at the low currents. Mikek

I have seen it before and seen an explanation that seemed reasonable at the time. I was a LOT better with magnetics back then (25-30 years ago), but still no great shakes. I think is may be related to effective air = gap due to the laminations meeting up less than perfectly; not that i = remember any such information.

Maybe someone here who does know something about magnetics can help us = out here.

Magnetic material is also part of the mess, maybe more important than = core shape. Certainly laminated and tape wound cores may be the most problematic.

the

=20

In the hysteresis curves please not that there is a small low slope part just coming off 0,0. That is part of where it is coming from.

?-)

Quite simply, magnetic domains are "sticky". Down at low levels, buried in the hysteresis loop of the material, small-signal loops are flatter because few domains are responding.

I'm sure this effect has a strong tempco, since higher temperatures free more domains (much like dithering a balance to get it to the equilibrium position despite bearing friction). It's present in all metallic materials I know of, though you'll have a hard time measuring it in super-high permeability materials due to the extremely small hysteresis loop, and in powdered materials, since they have a lot of gap in the matrix already.

Still, powdered materials yield a fair amount of change:

I'm trying to remember if I've seen an analogous curve for ferrites. If I have, I don't remember which manufacturer had the curve...

The tempco is more easily observed in ferrites than metals, since the curie temperature is lower (which is like a magnetic "melting point"; it loses all order and ceases to be ferro/ferrimagnetic at that temperature). Permeability almost always rises on approaching the curie point (which, I suppose, is like water becoming less viscous as it approaches the boiling point), which means more domains are being more easily moved; meanwhile, saturation flux goes down, which means fewer and fewer are staying in their polarized positions. At curie, Bsat suddenly drops to air-cored levels.

Well, a gapped core doesn't depend on the core very much, so you're okay on that one. I'd worry more about the expansion tempco of whatever the core spacing material is!

Tim

Sorry, can you rephrase that, I don't understand what you wanted to say. Mikek

Thanks for the links, Mikek, and for your comments, josephkk.

I spent all morning researching the question. The best answer I could find was here:

Note that the AC permeability (page 1) increases with increasing flux density. A crude estimate of the B involved with my measurements gives about 60 gauss (.1V @ 100Hz) for the Tonghui and about 600 gauss when the HP audio oscillator was set for 5V.

The AC permeability would therefore go from about 6000 to about 30,000 for about a 5 to 1 increase in inductance. My actual measurements were about 2 to 1, but, as I said, this is a very crude estimate.

This doesn't answer *why* the permeability changes, but does point out that the inductance of cored inductors will change with B. So, I think we can expect usually lower inductance readings from the Tonghui, especially for the power magnetics cases.

Here is another shot showing powdered iron core changes of inductance with flux density:

Cheers, John S

the

at

of

part

Tim Williams explained it soo much better. At very low flux levels there is a lower proportion of domains responding (per Ampere Turn) to the impressed flux, thus lower effective Mu(r) and lower inductance.

?-)

Sort of like a magnetic version of crossover distortion.

Cheers

Phil Hobbs

That's very good. I have ask a couple of times about this phenomena and could not generate any interest. My line of questioning had to do with transformers at the RF input of a radio. The signal can be as low as 1 uV.(less than a picowatt) I'm sure when you look at the spec's for *Mu(r) it was measured at much higher power levels. So the transformers used at these low levels should have many more turns than the spec's would lead you to believe. Yet, I have not seen any references to support this.

Your explanation makes a lot of sense, Tim. Thanks very much.

John S

That's

me the

voltage.

not at

turns

slope of

part

say.

there

Thanks for the new (to me) analogy.

That's

me the

not at

turns

slope of

part

say.

there

=20

=20

Not that i claim to really understand this (let alone well), i am pretty sure that much of this is a material dependant property. Laminated core power transformers would be rather bad about this (they normally would have plenty of magnetizing current). And ferrite RF materials would be very nearly independent of this. But that is just a guess.

Relative permeability, (Mu of the material/Mu of space).