Goofy transformer measurements

Nectarpuke

This morning I tried my hand at measuring the inductance of a 10VA toroid to be used as an audio output transformer. Using this method with two multiturn potentiometers in series. The measurements were taken off the secondary with a 4 vpp signal generated from my laptop. The first signal was 70Hz and the measured resistance was 155.2 ohms. Using the formula, this equates to 352.8 mH. Multiplying by the turns ratio squared gives 186H across the primary. This doesn't seem too far fetched from what I have been able to gather from others experience with toroids on this forum. The problem is when I measured at 22Hz, the half voltage was with 136.6 ohms, leading to a calculation of 988.2mH across the secondary and 522H on the primary. This doesn't seem at all right as I was expecting the inductance to go down with the lower frequency. I am trying to understand what I am missing... any help would be appreciated

Attachments

• IndMesurements-1.jpg
40.1 KB · Views: 219

Mark Tillotson

You seem to be using a 330k shunt resistor which makes no sense.

Elvee

Even if the resistor has a sensible value like 330 ohm, the results are worthless because they assume a perfect inductor, which for a 50Hz Xformer couldn't further from the reality: windings have a relatively huge series resistance, and the core has poor properties, meaning eddy currents losses, hysteresis, magnetic viscosity and a large difference between initial, incremental and amplitude permeability.
This means that to make meaningful measurements, you need to make a vector measurement, taking into account the reactive portion of the impedance only, and the amplitude of the stimulus needs to be similar to the actual voltage seen in use.
Using a soundcard, you can make vector measurements, but to have the amplitude right, you also need a suitable lab amplifier.

Mark Tillotson

I'd measure the DC resistance of the winding, then for each frequency measure the shunt voltage and the winding voltage too. Vector subtract the DC resistance from the AC impedance for an estimate of the inductance.

Nectarpuke

The schematic is just an example of the layout. I used a 500 ohm and a 100 ohm trim pot in series for the measurements. The secondary DCR is 1 ohm. I don't understand what a vector measurement is, but I will look into it straight away

Mark Tillotson

Well vector means phasor in ac circuit jargon, or you could say complex impedance from a more mathematical approach. But really its just measuring impedance, since impedance is complex.

If you use a low value shunt (1 ohm is a good choice), then the voltage loss across the shunt won't affect the voltage across the winding significantly and you won't need to measure them both. All assuming the signal generator has a low output impedance (not common, most are 600 or 50 ohms output).

The safest approach is to measure both voltages across shunt and winding so you aren't affected by changes in oscillator output voltage or the shunt voltage. Given a low DC resistance you ought to be able to assume the impedance of the winding is just inductive to a good approximation.

trobbins

I agree with Mark - use a low value current sense device such as a 1 ohm resistor and a uV or mV rms voltage meter, or use a uA or mA rms meter, and use a separate Vrms meter across the winding under test. Preferably use at least 5Vrms across the winding so that the inductance result can be compared to other OPTs (as inductance various significantly with applied voltage, and so comparing inductance levels needs the test voltage to be stated).

If you want to try and reduce error by subtracting out the resistive part of the winding and test arranegement, and can't do the maths but have MS Excel on you PC then link below may help:
http://www.dalmura.com.au/projects/OT%20calcs.xls

Last edited:

PRR

Paid Member
...I was expecting the inductance to go down with the lower frequency....

Why?

Iron-core coils start falling above 200Hz and can still be rising below 50Hz.

JMFahey

If you use a low value shunt (1 ohm is a good choice), then the voltage loss across the shunt won't affect the voltage across the winding significantly and you won't need to measure them both. All assuming the signal generator has a low output impedance (not common, most are 600 or 50 ohms output).
Why would you do that?
The method shown is a classic, tried and true High School (or old style) Science Lab method, and is precisely based on reaching same voltage across each of the circuit elements.

Emphasis on across, you need a floating voltage meter to connect first across
the resistor, then across inductor.

One of them may have a grounded leg, the other by definition can not, in fact ground is irrelevant..

As drawn that is not made clear.

No, you can not simply measure full applied voltage and then expect to find half that at the midpoint, it does not work that way.

The classic method shown works always, and is based on "first principles", which let you measure using the most basic circuits and techniques; no Arduino needed

Forget modern conveniences, think how did Maxwell, Hertz. Hall, even Ohm, (and hundreds others) measure their experiments which allowed them to CREATE all of this wonderful modern Science.

In this case, trying to achieve two equal values, and nothing else, makes it quite independent of meter sensitivity (as long as it can show *something*), linearity, frequency response, etc.

Last edited:

Nectarpuke

Why?

Iron-core coils start falling above 200Hz and can still be rising below 50Hz.

This statement is the first I have encountered that suggests otherwise. 522 Henries across a primary seems a bit ridiculous. That is the highest inductance I have heard of by a large margin. I am no electrical engineer, merely a machinist with a decent grasp on electrical theory... such values raise quite a bit of doubt. I am not yet done tackling this. I will dig in deeper in the morning

PRR

Paid Member
Yes, 500H in a 10VA core (what nominal voltage???) seems unlikely even at 20Hz.

Such errors are usually un-monitored voltage sag in he source. Hundreds of Ohms is a very heavy load (for high accuracy) from a laptop. If the output is capacitor coupled it is even worse. And as said, the technique shown should measure BOTH V(L) and V(R) at the same time with floating meters.

Nectarpuke

Yes, 500H in a 10VA core (what nominal voltage???) seems unlikely even at 20Hz.

Such errors are usually un-monitored voltage sag in he source. Hundreds of Ohms is a very heavy load (for high accuracy) from a laptop. If the output is capacitor coupled it is even worse. And as said, the technique shown should measure BOTH V(L) and V(R) at the same time with floating meters.

I did miss the measuring of VL and VR with seperate meters initially, I will admit. After it was corrected, my numbers came out even higher. It might be helpful to mention that the measurements are taken on an oscilloscope. The transformer is a 115+115:5+5 and rated for 50Hz. Primaries are in series, as are the secondaries. I will look into slinging together a suitable buffer... although the source didn't appear to be struggling. I appreciate everyone's experience and input.

Last edited:

trobbins

You can use a simple step-down transformer as the signal source - eg, a 5Vrms, 6.3Vrms or 12Vrms secondary winding.

Nectarpuke

You can use a simple step-down transformer as the signal source - eg, a 5Vrms, 6.3Vrms or 12Vrms secondary winding.

Indeed... but I intend to measure at more than one frequency if you are implying using 60Hz from the mains. Definitely an idea worth trying though as I have a few filament transformers and it would be quick to throw in.

trobbins

Is there a reason to measure inductance at multiple frequencies?

OPT's are often characterised by a low frequency inductance, and sometimes inductance variation with signal voltage, and with DC bias. And apart from winding resistance, the other characteristic parameters usually relate to high-frequency performance.

Nectarpuke

Mostly a learning exercise stemmed from curiosity about sub 40 Hz performance. I intend to make other measurements, namely parasitic capacitance... but one obstacle at a time. I know how it sounds in a variety of different circuits. But I don't know the numbers

trobbins

Have you read through some of the seminal texts such as:
Norman Partridge's two series of articles - distortion in transformer cores from 1939, and harmonic distortion in a-f transformers from 1942.
Whitlock's book/chapter on Audio Transformers
And a smattering of other articles, such as from Menno van der Veen.

Nectarpuke

I have not read any formal papers on transformers.

davidsrsb

Mains transformer primaries have huge inductance. There are a lot of turns on a big iron core

Nectarpuke

Ok... so I got around to trying again. This time I used a 12.6 volt transformer with a variac dialed in to 10 volts rms as the source, and two floating meters instead of a scope. The 50% division occured at 721 ohms equating to 1.91H at 60Hz. The turns ratio is 23 giving 1.011 kilo Henries! These numbers seem absurd, but that's what I keep getting. Is it really measuring this high or am I totally botching this? I am at a loss. I will send out an email to Antek and see what they have to say. I am baffled. I tried the resonance method, but my function generator isn't very functional at the moment... perhaps with these numbers I can work out something within my laptop's bandwidth.

Last edited: