Measuring inductance: HELP!! I'm confused.

[Mark Hammer]

I bought myself a new meter recently, since the old one was....old....and not behaving well.

The new one was reasonably priced, and came with a bunch of features I'd been yearning for, one of them being inductance measurement.

So help me out here.  I take a little Mouser-style transformer, and I measure the inductance on one side that has a centre-tap.  Fine.  Then I measure the inductance from the centre-tap to each outside lead.  Why doesn't the "halfway" point from the centre-tap equal half the total inductance?

If I had never seen people discuss the centre-tap as representing half the inductance (see the EPFM Passive tone control), I wouldn't be quite so baffled.  But I was expecting something very different than what I got.  On the plus side, it was nice to take a little ferrite rod, wind myself an inductor, and be able to measure it.

[alanlan]

I think this is only true for uncoupled coils.  Try measuring 2 separate coils and then join them with a piece of wire so that they are not coupled i.e. far enough apart not to produce any significant coupling, and them measure both together.

[R.G.]

I bought myself a new meter recently, since the old one was....old....and not behaving well.
The new one was reasonably priced, and came with a bunch of features I'd been yearning for, one of them being inductance measurement.

New tools are incredibly fun!

So help me out here.  I take a little Mouser-style transformer, and I measure the inductance on one side that has a centre-tap.  Fine.  Then I measure the inductance from the centre-tap to each outside lead.  Why doesn't the "halfway" point from the centre-tap equal half the total inductance?

It probably measures about 1/4 of the end-to-end inductance. Measuring across end to end, you are measuring across N turns. Measuring from center tap to one end, you're measuring across N/2 turns, and inductance is proportional to N². Half the turns, 1/4 the inductance.

[R.G.]

Forgot  - for even more fun, measure the inductance end to end on the primary, being careful to make sure all other leads are open circuit.

Then short the secondary and measure the same two primary wires. You are measuring the primary inductance in the first case, and the primary to secondary leakage inductance in the second case. For audio transformers, the ratio of the two is a figure of merit. You would like the primary inductance to be as much bigger than the secondary as you can reasonably make it. Good hifi output transformers have a ratio of 100,000 to 1 for this number, some even more.

[Mark Hammer]

Okay!

Mouser 42TM018, a 10k/10k used here for things like the Tim Escobedo Jawari.

Primary outside leads: 2.55H (secondary unshorted)

Primary center-to-outside:  0.63H   Exactly as predicted!

Secondary: 2.61H

Secondary: 0.64H

Primary outside leads with secondary outside leads shorted together: Here it gets weird....
    on 20mh range: 9.55mh
    on 200mh range: 49.4mh
    on 2h range: 70mh
    on 20h range: 70mh

So the ratio is something ranging between 273:1 and 37:1.  Not exactly "hi-fi."     But it's starting to make more sense now.  Thanks.

[R.G.]

It is tough getting solid measurements on a transformer. The actual inductance is both variable and nonlinear, depending on the size of the signal swing, the test frequency, presence or absence of DC and how much DC. We used to swear it varied with what we had for lunch, and yesterday's lunch if you measured before noon. Like hfe, the measurements are all exactly correct - for the full set of conditions that happened at the instant you took the measurements.

While I'm thinking about it: The open circuit test (i.e. across a winding, with the other windings all open) and the short circuit test (across one winding, short another winding) are the fundamental tests on transformers other than simple voltage measurements. They tell you the inductance of that winding, which is of course different on each winding, and whether there is an internal short. If your transformer ought to have, say, 1H of inductance on a winding, and it's instead 1mH, and no other winding is shorted, the transformer must be shorted internally. The short circuit tests tell you how much leakage inductance there is, and this is a measure of how well coupled the windings are. The coupling coefficient figures into a lot of filter and signal-transformer work.

If you measure the open circuit current for a given excitation voltage, you can very sensitively pick up when the current starts to increase more than linearly with voltage: that's the onset of saturation at that frequency. For measuring power capability, you measure the DC resistance (as best you can; it's tough measuring low DC values), then run the transformer for five time constants; thermal time constant is proportional to mass. Then after that time, you unhook the transformer and measure DC again. The difference in resistance tells you the internal temperature rise because copper has a positive thermal coefficient of resistance.

[Mark Hammer]

What you describe maps onto my measuring yesterday perfectly.  I'd hold the probe tips onto the transformer leads, and the least significant digit of the inductance would keep changing.

Of course, that is the world of transformers.  Should I assume that a simple modest inductor, wound around a simple ferrite rod , behaves in a slightly different manner, with respect to the stability of readings?

[R.G.]

Actually every digital meter can show a flickering least-significant digit. It's the nature of the beast to flicker either when what is being measured dithers a bit or if it's rock solid, but halfway between digits. Each conversion cycle it tries to figure out if that last digit is a 3 or a 4, for instance, and about half the time it picks one or the other. On a static test, same meter, no DC excitation, no changing frequency, I'd guess (and it is a guess) that the dithering is from the meter, not the inductor.

You'll likely get the same results from a simple inductor with that meter.

Brings up a fine point of using meters. You probably already knew this, either explicitly from instruction or implicitly from using meters, but one should always measure things with a meter scale that makes the measured quantity be in the top half of the meter scale on that range. The reading is more accurate because the meter inaccuracies and dither are less significant if the actual reading is up out of that last digit, or in between two scale markings for analog meters.

[Mark Hammer]

Understood.  Actually, what I was getting was a progressively descending least significant digit.  Flicker, I'm used to.  This was something more on the order of what you see when measuring the hfe of a germanium transistor you're holding between your thumb and index.

[R.G.]

Progressively descending is another issue. That's likely to be the result of a DC offset in the meter testing slowly ramping the inductor DC current.

There are several techniques for measuring inductance. The better ones keep DC out of it, and use true AC and not pulses all the same direction.