XFMR Phase Angle

I have been messing around with measuring phaseshift in transformers.
This is needed to compare the performance of transformers against each other.
It can be useful in telling whether a DIY trnsformer matches up with the original.
Frequency sweeps are useful but they don't tell the whole story.
The Crowhusrt articles have a slightly more advanced method that requires a lot more hassle, so I am just using this method as it is much easier and faster.
It might not be as accurate as the more complicated method, but I do not care, as I will be using the same method to measure up similar transformers, and as long as I use the same method for both, I will get a good comparison, albeit the degrees of phaseshift might not be exact.

Here is the setup: Same as the inductance setup. I am using about 150 ohms with a 100 ohm pot to match up the signals, not ten ohms..





First I find the frequency where there is close to zero shift. Then I get the two signals, one on the resistor, and the other across the winding, matched up voltage wise:



Then I hit the "XY" button on the scope.
If there is zero shift, I should get a 45 degree line, as the x and y signals are the same. This is a non time based setup, so I do not have to adjust the scope for sweep frequency. Cnannel 1 is the X deflection, and channel 2 is the Y deflection.



Now, I switch back to the normal mode and adjust the signal generator up in frequency until I get some phaseshift:



Now I switch back to "xy" mode. If everything is working right, I should start to see an ellipse:



Here is the formula for getting the degrees shift off the ellipse.
You want to center the ellipse, which can be done with the xy position controls:




As more shift is introduced by cranking up the frequency, the ellipse changes shape.
The flatter the ellipse, the more phase shift:



Even more phase shift.
About 1.4 divisions divided by about 1.5 equals 0.933.
To get the degrees, take the anti sin of 0.933 on your calculator, and you get about 68 plus degrees phaseshift:



We can check our method by going back to the standard input mode. This scope is cool, in that it has cursors that you can line up on the wave, and it spits out the mathematical values, which can be used to compute phase angle. It will tell you how many seconds difference there is between the waveforms, and that can be used to compute the degrees. Without boring you with the derivation, the formula is 360 times frequency times the time difference.
Here I am lining up the cursors on the zero crossing. You can also use the delta frequency to get the value. You would divide the actual frequency by the delta frequency and multiply by 360.



This ellipse is denoting a phase angle very close to 90 degrees. If there was a straight line, you would be at exactly 90:



OK, any questions?
Class dismissed!

see ya!
cj
:guinness:

CJ,

Looks like a sound methodology.

I'm spoiled though...phase angle is a direct measurement on the AP. :green: If you ever need exact values...PM me.

Peace!
Charlie

Class is still in session.

You get an ellipse only because you have decreasing response on the vertical channel as frequency is changed. If both channels had equal response, you would get a perfect circle at 90 degrees shift.

The caption on your last photo should read that the ellipse will be horizontal at 90 degrees. It will be a straight line only if the vertical signal goes to zero. Turn up the input sensitivity on the vertical channel so that you can make a good measurement. The ratio is unchanged when you change the gain.

Darnit Dale, you beat me to it!

:green:

Keith

errr, well I, ah....umm...ok...but....maybe.......see.......you......whatever.....
:shock:



so....it.........my......off....years ago........me........so..........stupid.......?
:twisted:

:razz:

Could Crowhurst and Atwood both be wrong?
Yes, you do have to adjust for equal amplitude when doing this test.
Yes, this means self res should be taken care of as much as possible by way of secondary termination resistor.
I ran out of time last night, I was about to say this, but.... :guinness:

What they say and show is quite correct, including the part about adjusting the amplitudes to be the same. That implies "at each test frequency" not set once and forget as your pictures indicated.

The lower picture from the reference is what you will see when operating at frequencies and drive levels that cause some magnetic saturation. For an audio transformer, that would be at the lowest intended frequency and highest anticipated level. This is a very useful test.

FWIW, frequency sweeps give you phase information too....it's just that most folks ignore it!

Cheers,

Kris

Assuming you have a minimum phase system, and not something like a tape recorder.

I think the lad deserves a prize - but it doesn't really get us anywhere - or does it?

I am using this test for a private reverse job. Someone sent me three very important transformers to reverse.
There is a professional company that is going to build the transformers, so I have to gleen as much info as possible to give them in order that we get the closets match possible. If we suceed, I am sure everybody here will be able to buy these famous transformers. So yes, this will help you guys!
If any of you know of any other tests (yes, distortion is one I am going to do also) then let me know.
Dale just gave me a new idea, so you see, sometimes a little bickering can be a good thing!

I have to write down the size and type of every piece of insulation, turns per layer, a lot more intensive than my normal reverse job, as well as aall the mechanical details, but it should be worth it in the end. I know we can't get the exaxt core material and wire as the ol days, but we should get pretty darn close.
My brother works for Varian. I am trying to get him to analyze the copper and iron in one of those million dollar mass spectrometers. (funny story, he took a bunch of different types of beer to work and had the guy run them thru the machine! You can see the individual spikes from all the impurities, as well as the alchohol content. Weird, but some of the cheaper beer, Bud for instance, came out a lot cleaner than Heiniken!)

:sam:

There is a sequence of tests that will help you determine the equivalent circuit of a transformer, including the primary, secondary and mutual inductances. The tests involve shorting one port and measuring the other port, then swapping ends and repeating the test.

Unfortunately you are dealing with a 3-dimensional problem with parameters that vary with both frequency and level. I would be inclined to want swept curves of frequency response, distortion and phase at various levels and frequencies. Lots of data, but a transformer is a really complicated device.

Tell me about it!
I have ten pages already of just freq graphs. One pri vs two secondaries, etc.
Different levels, different terminations,
And that's just for one of the transformers.
Yes, leakage front and back, weight of the steel, weight of the copper,
On and on. But picking up speed.
Oh, and lets not forget listening tests!

Whats weird about the above test is that we are only looking at the primary phase shift.
I am wondering what would happen if I were to do the same test on the secondary.
I guess Crowhurst figures that it will be the same as everything is reflected back. I will do the same tests on the secondary.
Hmmmm
Dale.
1000 Oaks.
Couple of famous xfmr companies down there! I wonder if....
I use to talk to this Dale guy on the phone a couple of years ago,
Nahh, couldn't be!
:shock:
I was born in Northridge. Is that place burning up yet?
Earthquakes, fires, traffic, and the prettiest girls in the world!
(If you leave out Costa Rica)
I was just another freak in the freak kingdom!

:guinness:

Yes indeedie... when you introduce a fixed resistance in series with a reactive load, the variation in transformer reactance/impedance with frequency will affect the proportional distribution between the two.

A long as any 'clone' also does the same thing to the same degree at the same frequencies, it's also another way to validate the accuracy of the copy...

but for a simple phase measurement (which -as a measurement in isolation- is independant of amplitude) it can be disregarded. -If you're going to live by the phrase 'a perfect circle' as representing 90° phase shift at a given frequency, then you'll likely have to adjust the amplitude to get it. -Otherwise an elipse stretched vertically or horizontally will be displayed... both also indicate exactly 90°, and can occur only at ±90°shift. Any angular deviation from that (in otherwords a 'tilted' elipse) represents another figure of shift.

...at least that's how I read it!

A reactive shift with a non-constant impedance is not a bad thing, but it would be a great thing to note the amouont of "uncal" correction to be made at specific frequencies, as another way to compare the behaviour of a copy, or 'clone'. :thumb:

Keith

Nice scope!

Beer analysis using a varian Gas Chromatography-Mass Spectrometer! I used to have one on my bench at work before the turbo pump died for the n'th time (where n is a very large number).

I think it would be an ICP-MS for getting elemental composition (inductively couple plasma mass spec). Sounds cool to me!

Or an electron microscopy or some other gadget,

just a note, I trimmed the series resistor with a 100 ohm pot with a 56 ohm resistor on the wiper.

at high freqs where Xl is like 30,000 ohms, the 100 ohms really dosen't matter.
but at the low end, it comes into play a lot more.

from CJ:

Weird, but some of the cheaper beer, Bud for instance, came out a lot cleaner than Heiniken!)

Click to expand...

To be honest, here in the Netherlands it keeps amazing a lot of people why 'our' Heineken is so popular outside the country. It must be the marketing then, since ehh, 'a bit chemical' & 'Heineken' are synonyms imho. :grin:

Can you get Grolsch over there ? Especially 'Het Kanon', it tastes good and is efficient (11.6%).

Peter

I sent an e-mail to Bill Whitlock at Jensen Transformer to see if he could recommend a book that covers transformer testing and characterization in detail. Don't know if he knows or will respond, but it is worth a shot!

Seems like as long as you know the actual value of the sampling resistor to use in your calculations, the absolute value isn't important. Since you are measuring the actual current and voltage of just the transformer, your phase calculations should be correct. Any frequency response plots, however, might be sorely affected by the series resistor.

My list of interesting characteristics would include:
DC resistances
All self, mutual and leakage inductances (will this yield core reluctance?)
All capacitances - between windings, to shields and core and case
Core material BH curves at several levels and frequencies
Electrostatic shielding
Magnetic shielding
Physical construction
Core dimensions
Winding size, placement, tension, type, polarities/end connections
Any magnetic gaps
Wire insulation thickness and shape (round, rectangular, multifilar)
Turns ratio and count for each coil

This should allow the determination of:
Voltage or impedance transformation ratio
Balance of multiple windings
Common mode rejection (vs. frequency)
Distortion characteristics vs. frequency and amplitude
Include changes with DC current, if permitted
Thermal noise
External hum and crosstalk sensitivity
Safe voltage levels
Maximum permissible temperature

I have probably left out quite a bit, but that is the best I can do in 5 minutes off the top of my head. I will be glad to incorporate suggestions from others.

By the way, I am not a 'transformer' person, but my expertise in magnetic recording has many similarities.

Thanks Dale!
Yeah, there is another Dale at Jensen who is very helpful.
I will ask him about some phase test setups.
I can do a phaes plot of pri vs sec. with the dual trace and see how it matches up with the ellipse test.
Crowhusrt mentions something about the current thru the resistor being sinusoidal, and the current thru the transformer not being exactly sinusoidal, or something along those lines,I wish i had the whole book.
Maybe a trip to the Stanford Library is in order.
:thumb:

OK, just some note on the phase shift tests:
the resistor in series with the primary current method compares the signal the transformer is seeing with the magnetizing current.
So in theory, the two signals should be out of phase by 90 degrees if we were testing a perfect transformer. This is what I am seeing.
Also, the ellipse method does match up very closely with the a/b dual trace method that uses the time based cursors to determine the delta f.
The ellipse is better at small degrees of shift, as there is a larger difference between the y crossing and the top of the ellipse.
However, at close to 90 degrees, where the ellipse is more circular, there isin't much difference between the y crossing and the top of the circle. This is where the a/b side by side testing takes over, giving a much more accurate phase shift angle.
So I use the two methods to plot magnetizing current shift, and the straight a/b method to compare the pri. with the sec. shift.
At close to zero phase shift from pri to sec, there is close to 90 degrees shift in the magnatizing current, which makes sense. At at large pri/sec shifts, the mag. current phase angle tends to zero.
And it looks like you will get the same numbers when you check mag. current on either the pri. or secondary. This ,makes sense, as in both methods, we are looking at the current needed to put flux into the core, and it shouldn't matter whether we are using the pri. or sec. to get the mmf going.
Also, as you do a sweep, the phase of the mag current goes from leading to lagging, so you have a plus and minus phase shift. So you have to be careful to note when they cross over the zero phase shift point.
I am using a decade resistor box, so I can note phase shift along with the series resistance used, which has to change to get a good r to Xl ratio.
cj
ps: Ozzfest was killer. No better tone than Iommi's sg on the planet.
Halford sounds better now than he did 20 years ago.
A lot of people leaving early, missing the main acts!
A lot of people forgot to wear sunscreen.
Got to ration yourself out at those 14 hour metal concerts!
Which is why I only caught the last three acts.
Couple of people getting IV's on the stretcher!
:guinness:

I dont mean to steer the transformer talk to beer (why does everything revert to beer around here, keef?) but Clint, I'll never EVER forget the first time I toured with a band in Europe, we got to the Netherlands and in my beer snobbery I kept passing on the Heineken but eventually cracked one open and man, if you ever come to the US, DO NOT drink the heineken, you'll be sooooo bummed. The shit we have here that says heineken on it is NOT AT ALL the same as what you guys drink there, and americans love the stuff here, so bizarre. I also learned on that trip that the american chocolate bars use dutch chocolate, so simple stuff like snickers bars are AMAZING, another food item you can skip when you come to the USA. I also discovered chocomel on that trip, I often have this fantasy of getting rich importing wonderful chocomel into the US, Im pretty sure I could retire off of that program. You cant get a good burrito in europe though, so we have you trumped on that. And you cant get pizza outside of nyc.

CJ, get busy on that analysis, we've got a group order baging down the door!

dave