How to properly measure OT inductance??

[Wombaticus]

Hi All,

A got myself a birthday present in the form of a Tonghui TH2822A LCR meter. These can be had now for under $200 and are very similar indeed to the well known BK Precision 879B. Anyway for a unit in this price range it seemingly packs a lot of punch: L / C / R / Z; secondary parameters / Q / θ / ESR, with 100Hz-10KHz measurement frequencies.

Of course, I immediately set out to measure all of the caps and inductors that come to hand, and quickly came to the conclusion that correctly using these LCR meters is rather more nuanced than it appears at first blush.

One of the first things I checked was an Allen Amps TO-26 output transformer. The OEM is Heyboer, Heyboer part HTS-9483

According to the Allen amps website this is nominally 7000K into 8R or 16R, intended for 2x6V6 in push-pull operation.

What follows is a description of some readings I took, and there are a few things lurking there that I hope the brain trust here can help me figure out.

First of all, I took measurements of the inductance of the primary and secondary impedances with taps open... the first mystery is that the pri/sec inductance ratio doesn't look anything like the 7K:16R impedance ratio suggested by the spec. It's much lower, with about a 336:1 ratio (18.3:1 turns ratio) instead of the 438:1 ratio (20.9:1 turns ratio) I thought I would see. The inductance ratio suggests a transformer that is more like 5.4K primary Z.

Now, this also turned out to be true of the Z measurements I took. The priZ/secZ ratio was about 315:1, again well short of the 438:1 ratio I expected.

However, when I soldered an 8.2R resistor across the 16R tap, I read a Z (at 1KHz test frequency) of 4.130 K.... Normalizing a bit, 4.130K:8.3 = 7.963: 16R, so this last figure suggests an OT that is behaving more like an 8K:16R.

So I have a discrepancy between the ratio suggested by the unloaded inductances and the unloaded impedances with the ratio suggested by measuring the impedance of the primary under load. And neither of these match the nominal specification of the transformer (although I have written to David Allen to confirm that it is not, in fact, supposed to be an 8K pri Z transformer, as this is pretty common for 6V6 amps).

The problem is that I don't have enough experience to know whether this is something that would be anticipated, either because of the way that I'm using the meter, or because I haven't properly accounted for the influence of parasitics, or if the meter is simply reading inductance incorrectly. The meter is calibrated to account for leads, but I don't have a good external standard inductor to check it against.

What do you think? Hopefully somebody that really understands magnetics will chime in.

EDIT: The inductance and impedance ratios measured at 120Hz are also off.. 8.887H pri/34.8mH sec.. so actually even more wonky at about 255:1, suggesting an ~4K pri. Z. These are big errors!

--------------------------------------------


Primary leads: Blue-Red(CT)-Brown
Secondary leads: Black(common) - Green - Yellow
Measurements are from TH2822A at 1KHz, serial mode, except for DCR from an Ideal 61-312 DMM.

DCR:
Brown-Blue: 249.9R
Blue-Red: 130.8R
Brown-Red:118.4R
Black-Yellow: 0.9R*
Black-Yellow: 0.9R*
Green-Yellow: 1.9R*
* DCR measurements in this range should be regarded as approximate only

Primaries, with secondaries open:
Blue-Brown: 7.064H, Q 5.07, 45.28K Z
Brown-Red: 1.7976H, Q 4.755, 11.53K Z
Blue-Red: 1.807H, Q 4.694, 11.58K Z

Secondaries:
Black-Green (8R tap): 10.776mH, Q 2.625, 72.51R Z
Black-Yellow16R tap): 21.041mH, Q 2.634, 143.97R Z
Green-Yellow: 1.957mH, Q 2.684, 13.152R Z

Primaries with secondaries shorted (all three secondary leads soldered together, inductance values reflect leakage inductance):
Blue-Brown: 36.434 mH, Q 0.4208
Blue-Red: 13.396mH
Brown-Red: 10.037 mH
(interestingly if I solder together only the full tap black-yellow secondary leads I get more like 85mH - I don't know why?)

[Alan0354]

I am just jointing in as I am studying OT at the moment. Voltage ratio and resistance ratio is mainly on the turn ratio where

Rpri = (N1/N2)^2 RL, and Vpri =(N1/N2)VL. Which you measured close.

Inductance L=u N^2 A where A is the cross sectional area. Even if you have the correct turn ratio as given above. The inductance is not necessary a linear relation. Think of how you wind the transformer, the inner turns have much smaller cross sectional area. As you wind the coil, it gets larger and larger, the cross sectional area gets larger. The inductance per turn in the outer layer is going to have more inductance per turn than the each turn in the inner layer close to the core. Depend on how the OT being interleave, you have different inductance. I opened a few cheap OT long time ago, the secondary were wounded at the outer layer on top of the primary, the inductance would be artificially higher because the cross sectional area is larger at the output layer, so you see a lower ratio if you do the open circuit inductance ratio.

Maybe that's the reason no books talk about open terminal inductance ratio of an OT, rather talk about turn ratio.

I don't know how the meeter measure, if you put a resistive load, most of the impedance reflected back to the primary is resistance in series with the Leakage inductance. The leakage inductance is measured by shorting the secondary then measuring the primary.

[Wombaticus]

Maybe that's the reason no books talk about open terminal inductance ratio of an OT, rather talk about turn ratio.

At first blush that seems reasonable, but after a moment's reflection I'm now more confused than ever. Because, after all, it is the *inductive* phenomenon that ultimately gives rise to transformer action, not the turn ratio or any of that other stuff.. The turns ratio is logically just something that emerges from that basic starting place. Doesn't it stand to reason that if I measure the real physical inductors that I would be measuring the property that actually matters?

[Alan0354]

At first blush that seems reasonable, but after a moment's reflection I'm now more confused than ever. Because, after all, it is the *inductive* phenomenon that ultimately gives rise to transformer action, not the turn ratio or any of that other stuff.. The turns ratio is logically just something that emerges from that basic starting place. Doesn't it stand to reason that if I measure the real physical inductors that I would be measuring the property that actually matters?

Not from the books and articles I am reading. voltage and impedance are governed by turn ratio only. Yes you have core loss, but mainly is the turn ratio. Inductance is very clearly defined in physics. The formula I gave is for a coil which the OT is.

[jazbo8]

Check these out, guys! Z changes with V (and f), so I think the little LCR meter is not driving the transformer hard enough for you to get relevant readings (unless you plan to operate it at such low voltage).



Here is the suggested test rig:



More here => http://www.dalitech.com/Resources/Me...erformance.pdf

Jaz

[defaced]

I don't know if this will be applicable, but there was an old thread where someone took many measurements of a stock OT. Primary Inductance, and a link to a page for measuring it, was given. http://music-electronics-forum.com/t6176/

[Steve Conner]

Due to non-linearity of the iron core, the inductance of an iron cored coil varies depending on the size of the signal you use to measure it. That's why your readings all come out different. The amplitude of the test signal out of the meter might be the same, but the amount of core flux it generates depends on the number of turns in the winding you connected it to.

I tried explaining this to Alan, but I guess he never got it, as it can't be expressed in the form of an equation.

The proper way to measure inductance of an iron cored coil is with a signal of the same amplitude and frequency as it'll actually see in use.

[Wombaticus]

Due to non-linearity of the iron core, the inductance of an iron cored coil varies depending on the size of the signal you use to measure it. That's why your readings all come out different. The amplitude of the test signal out of the meter might be the same, but the amount of core flux it generates depends on the number of turns in the winding you connected it to.

I tried explaining this to Alan, but I guess he never got it, as it can't be expressed in the form of an equation.

The proper way to measure inductance of an iron cored coil is with a signal of the same amplitude and frequency as it'll actually see in use.

While I understand that there will always be core losses from eddy currents and hysteresis, I'm having a hard time squaring what you are saying with the function of an audio transformer. Admittedly there must be very pronounced non-linearity when an iron core approaches and enters saturation. But if there were huge non-linearities before that, to the point that inductances were markedly dependent on frequency and signal size, then it follows that the transformer would be nothing more than a huge distortion machine, completely unsuited to the task at hand.

Any thoughts as to the validity of the other test I tried - soldering in a load across the secondaries and measuring the impedance across the primaries?

[Steve Conner]

Transformers are huge distortion machines. The permeability is small for small signals, increases by a factor of about 2-5 for medium-sized ones and decreases again towards saturation.

The distortion isn't as bad as you'd expect because the reactance of the magnetising inductance is large compared to the load impedance over most of the audio band. Things get interesting in the bottom octave, and that's where much of "transformer sound" comes from.

If you measure one winding with another loaded, you see a combination of the leakage inductance, magnetising inductance and reflected load impedance. Maybe your LCR meter can help you tease those apart.

The classic way to test a transformer is: First an open-circuit test at rated voltage and frequency to measure the magnetising inductance and core loss resistance. In an audio transformer this would be the lowest frequency in its range at its rated output power.

Second a short-circuit test at rated current (and whatever voltage it takes to drive that current) This gives you the leakage inductance and winding resistance (copper losses). For an audio transformer this might be done at a quite high frequency.

A multi-winding transformer can have a different leakage inductance for every coupled pair of windings, so you would repeat the short-circuit test n-1 times. The open-circuit test only needs done once, because there is only one magnetising inductance, it being a property of the core. The different windings just reflect it according to (the square of) their number of turns.

[Wombaticus]

Hmmm. So I just bought an LCR meter that turns out to be completely unsuitable for one of its intended applications. I can't tell you how much that sucks. I mean, WTF is it good for then? The upper range of inductance measurements supposedly goes to 1000H. But I can't imagine that very many inductors of even more than a few Henries *aren't* iron cored.

I guess this explains another observation.. measuring the inductance of a Hammond 156G 40mA choke (nominally 9H) with this meter gives me only about 6.6H (and 300ohms DCR).

This choke is designed to be operated with a DC bias, and the nominal inductance stated with that 40mA bias current. But I was expecting that the measured inductance would be much HIGHER than the nominal, since it should increase with decreasing bias current. But of course, that's not what I saw. So it looks like I can't even measure a simple choke with this damn thing.

[Wombaticus]

Sigh, there's obviously even more about this that I'm not understanding.

I noticed that the primary inductance measurements I was taking were slowly drifting down over time. So last night I left the meter attached to the OT primaries overnight (secondaries open), and the value seemed to pretty much asymptotically stabilize. The weird thing is... after leaving it attached like that overnight, if I then short the secondaries, instead of the value falling to a few tens of millihenries, as it did the other day, it now falls only to something like 6.4H. What have I done? Have I magnetized the core or something by leaving it attached for so long?

EDIT: falls to 6.4H if I short the 16R tap. But actually *increases* slightly if I short the common, 8R, and 16R! I'm going now to sacrifice a chicken and do some voodoo rituals. I'm beginning to suspect that this may be the only way to way I'm ever going to wrap my head around this stuff.

[R.G.]

There’s a lot here to recap and comment on.
First – any time you get a new instrument, it will have its own quirks and may need to be calibrated. This is true of laboratory instruments and even guitar-instruments.

Second, there are many views to take of what you’re measuring. This is a little like trying to figure out what an elephant looks like by combining the reports of several blind men.

And particularly with iron-cored inductances, you are correct – it’s a more nuanced process than it may appear. As Steve correctly points out, the inductance you measure varies with both frequency and with signal size; this is a consequence of the core material being imperfect and changing its magnetic materials with operating point. Magnetic amplifiers use this phenomena to make amplifiers out of iron cores.

You are correct that the magnetic field inside the core is what matters for what a transformer does; however, the turns of wire are the viewports that let us indirectly see what the M-field is doing. So determining the turns ratio is important. Fortunately, this is easy. For a mid-band frequency, the ratio of primary and secondary voltages is the same as the ratio of the numbers of turns. 500-1000 Hz is usually a good place to start for an audio transformer. Get this number estimated by the ratio of voltages first, then use the voltage ratio and resulting turns ratio to estimate the impedance ratio by squaring the voltage ratio. It’s easier and more accurate to get impedance ratio this way, as it’s a more direct measurement.

I would expect the loaded versus unloaded inductances to maybe not match all that well, just out of experience with transformer internal parameters being wiley and difficult to nail down.

Alan is correct that the impedance will always be measured in series with the phantom leakage inductance. This will appear between the lead you’re measuring with and the reflected load in all cases. The classical tests on transformers are the open circuit primary (or other winding) inductance, and the shorted-secondary inductance, which will be very close to the leakage inductance. During the Golden Age, the ratio of the primary inductance (… at some frequency (?) ) to the leakage inductance was taken as the main figure of merit of the transformer for frequency response. This sometimes was as high as 200,000 to one for carefully constructed hifi transformers.

Alan is mildly incorrect about the inductance per turn being larger for turns further away from the lowest layer of the core. This would be true if there were not a core there; however, the iron core “conducts” M-field something like 10,000 times better than free space does, so to a first approximation, all the M-field is sucked into the core iron by this free pathway for flux.

To a second approximation, there is some flux that does not get into the core, and therefore does not couple higher layer turns to lower layer turns. This is the flux that makes for leakage inductance. It is literally the flux that leaks outside the core.

You’re right that the basic nature of the beast is inductive, so measuring inductance directly seems like a better measurement. What fouls this up is that you can only look through the winding turns to “see” this, and that it’s not one number. It varies with frequency, signal size, residual DC, probably temperature, core magnetization history, even things like whether the core has been pounded on mechanically. So measuring inductance for a signal transformer is a slippery affair.

Steve is also correct that the leakages between every winding pair will be different; this stands to reason, since it’s the physical location of the wires that cause different flux couplings per winding pair, and the shared flux coupling is that makes voltages. Also, it’s the change in magnetic field per time that makes the voltage, not the field itself.

And that we know inductances by what they do, not what a meter says.

Your meter is not worthless – it’s a valuable indicator of what may be happening. What you’re going through now is the learning period where you find out what it can and can’t do, what its limits are.

[Alan0354]

Due to non-linearity of the iron core, the inductance of an iron cored coil varies depending on the size of the signal you use to measure it. That's why your readings all come out different. The amplitude of the test signal out of the meter might be the same, but the amount of core flux it generates depends on the number of turns in the winding you connected it to.

I tried explaining this to Alan, but I guess he never got it, as it can't be expressed in the form of an equation.

The proper way to measure inductance of an iron cored coil is with a signal of the same amplitude and frequency as it'll actually see in use.

Not that I don't get it, I want to see it from a book or preferably more than one book . Many people can write articles, I even published papers in American Physics Institution, Review of Scientific Instruments which is journal used by real scientist and under gone detail scrutiny before publishing. Believe me, don't take them too seriously!!!! Most are just opinions with proofs that are quite skewed. The name of the game is just like patent application..........how to get pass the examiner!!! Don't trust what's in an article too much. When people write a book and is used widely, THEN you go through peer reviews, and then people study the book and proof the formulas and challenge the validity like what I am doing. Believe me, I challenged the book in Phase Lock Loop by Roland Best to the point he offered to send me a revised copy.

And even in Books, I am poking holes in RDH4 and Flanagan. How do you expect me to take an article too seriously. That's the reason any subject I studied, I bought at least 5 books to fact check each other. I don't think I'll do this on transformers as I think it deserve only a few more days only.

As I said, I am just starting to study, so far, there is one article from R.G or Jazbo8 about this, both RDH4 and Flanagan that I studied( which is only the small portion) have not mention anything about it and they derive formulas without using the hysteresis. Not saying it is not true, just have not seen it yet.

[jpfamps]

Not that I don't get it, I want to see it from a book or preferably more than one book . Many people can write articles, I even published papers in American Physics Institution, Review of Scientific Instruments which is journal used by real scientist and under gone detail scrutiny before publishing. Believe me, don't take them too seriously!!!! Most are just opinions with proofs that are quite skewed. The name of the game is just like patent application..........how to get pass the examiner!!! Don't trust what's in an article too much. When people write a book and is used widely, THEN you go through peer reviews, and then people study the book and proof the formulas and challenge the validity like what I am doing. Believe me, I challenged the book in Phase Lock Loop by Roland Best to the point he offered to send me a revised copy.

And even in Books, I am poking holes in RDH4 and Flanagan. How do you expect me to take an article too seriously. That's the reason any subject I studied, I bought at least 5 books to fact check each other. I don't think I'll do this on transformers as I think it deserve only a few more days only.

As I said, I am just starting to study, so far, there is one article from R.G or Jazbo8 about this, both RDH4 and Flanagan that I studied( which is only the small portion) have not mention anything about it and they derive formulas without using the hysteresis. Not saying it is not true, just have not seen it yet.

I particularly like this forum because of the number of real experts who take their time (free of charge) contributing to it, and who give great theoretical and practical advice.

I hope this doesn't seem too blunt, but if you don't trust articles written in books or on the internet (and with good reason), then why are you keep asking questions on an internet forum, especially as you don't seem to like the answers?

Furthermore, a lot of your posts seem to be an exercise in showing off your academic credentials rather than contributing anything to the subject being discussed e.g. "Believe me, I challenged the book in Phase Lock Loop by Roland Best to the point he offered to send me a revised copy."

If you want to start a thread "Please list your academic achievements" fine, but I don't think it's helpful sprinkling them liberally throughout rest of the forum.

[Alan0354]

I particularly like this forum because of the number of real experts who take their time (free of charge) contributing to it, and who give great theoretical and practical advice.

I hope this doesn't seem too blunt, but if you don't trust articles written in books or on the internet (and with good reason), then why are you keep asking questions on an internet forum, especially as you don't seem to like the answers?

Furthermore, a lot of your posts seem to be an exercise in showing off your academic credentials rather than contributing anything to the subject being discussed e.g. "Believe me, I challenged the book in Phase Lock Loop by Roland Best to the point he offered to send me a revised copy."

If you want to start a thread "Please list your academic achievements" fine, but I don't think it's helpful sprinkling them liberally throughout rest of the forum.

But a lot of time, it is true, I do find mistakes, you blame me in challenging the articles or books. You study something and you really take all at face value? If you don't challenge, how are you going to learn?