Shorting transformer windings for PI, why is it done?

[SecondGen]

Had an interesting discussion on the job today about shorting together transformer windings when doing a PI test. Every technical manual I've come across tells you to do it but I can't seem to find a clear answer why.

I've asked around and get different answers. Some guys argue it reduces stress on the winding and others have said you do it to reduce induction, which opens up a whole new topic of discussion.

Personally, I do it because that's the way I was always taught, but I do see how you can argue that its unnecessary because transformer windings are normally connected together anyway. I also know that using jumpers can reduce your actual reading because I've seen it happen.

What do you normally do in the field? Looking for opinions on whether or not you use jumpers and why.

[BigJohn]

I've never seen that windings should be shorted for DC tests, and we never do.

I could be mistaken, but that sounds like a misunderstanding of the requirement to short windings when doing AC tests like power factor or AC hipot. In those cases, you do it to null transforming effects.

I agree with you, that's not a problem with static DC.

[SecondGen]

I've never seen that windings should be shorted for DC tests, and we never do.

I could be mistaken, but that sounds like a misunderstanding of the requirement to short windings when doing AC tests like power factor or AC hipot. In those cases, you do it to null transforming effects.

I agree with you, that's not a problem with static DC.


I don't think its a misunderstanding, I've seen it in a lot of literature. This is the procedure Paul Gill lays out in his book:

224

I will post some more if I get a chance to dig through some megger manuals, I may have also seen it in "A stitch in time."

[Relay1]

I don't think its a misunderstanding, I've seen it in a lot of literature. This is the procedure Paul Gill lays out in his book:

224

I will post some more if I get a chance to dig through some megger manuals, I may have also seen it in "A stitch in time."

The tests are to measure resistance between the windings and ground, so would you not want to apply the test voltage equally across the device?

[ElectricalTestTech]

The tests are to measure resistance between the windings and ground, so would you not want to apply the test voltage equally across the device?

Well, the argument against using jumpers has to do with the interconnections of the windings. Transformers are wired to where one winding jumps to another. If you look at a delta or why configuration, everything is already tied together. So some people have made the argument that you don’t need to use jumpers when testing because they are already electrically continuous.

But Paul gill and Megger and doble are not stupid people. Their procedure specifically spells out to jumper the windings. I just don’t know the exact reason why

[Relay1]

Well, the argument against using jumpers has to do with the interconnections of the windings. Transformers are wired to where one winding jumps to another. If you look at a delta or why configuration, everything is already tied together. So some people have made the argument that you don’t need to use jumpers when testing because they are already electrically continuous.

But Paul gill and Megger and doble are not stupid people. Their procedure specifically spells out to jumper the windings. I just don’t know the exact reason why

I am running a transformer resistance check, measuring High to Low winding. If I apply 10kV on bushing H1, what is the voltage at H2 and H3? By bonding the bushings together, I know 10kV is applied to the entire H winding. The same goes for bonding the X windings together.

[ElectricalTestTech]

I am running a transformer resistance check, measuring High to Low winding. If I apply 10kV on bushing H1, what is the voltage at H2 and H3? By bonding the bushings together, I know 10kV is applied to the entire H winding. The same goes for bonding the X windings together.

I agree with that. But some can argue that a primary winding impedance of milliohms produces how much loss? Apply 10kv to H1 and read 9.999kv on h2 & h3. Don’t get me wrong, I’m not disagreeing with you. But this is what was argued by others. I always apply jumpers because I want to apply the same voltage evenly to the windings.

I just feel there is a more technical answer to why you jumper them other than being absolutely certain that voltage is applied evenly. I remember speaking to Paul Gil and he gave me a technical response. I just don’t remember and I absolutely don’t want to paraphrase and be incorrect

[Relay1]

I agree with that. But some can argue that a primary winding impedance of milliohms produces how much loss? Apply 10kv to H1 and read 9.999kv on h2 & h3. Don’t get me wrong, I’m not disagreeing with you. But this is what was argued by others. I always apply jumpers because I want to apply the same voltage evenly to the windings.

I just feel there is a more technical answer to why you jumper them other than being absolutely certain that voltage is applied evenly. I remember speaking to Paul Gil and he gave me a technical response. I just don’t remember and I absolutely don’t want to paraphrase and be incorrect

Hello ETT. Consistent is the term I would use. If you tested a device 5 years ago and I now have that task, if we are not being consistent in our connections and methods of testing, we may be comparing apples to oranges!

[ElectricalTestTech]

Hello ETT. Consistent is the term I would use. If you tested a device 5 years ago and I now have that task, if we are not being consistent in our connections and methods of testing, we may be comparing apples to oranges!

That is another excellent point! But there still could be many other variables that won’t allow you to compare apples to apples.

Humidity could be different causing lower readings. I have yet to see any correction factors for humidity. And I think we all have to agree that humidity is a major factor.

Was the transformer turned off the night before you arrived? Or the morning of? We’re there any delays in getting started with your work or mine for that fact? Temperature of windings may not have been properly documented on my part or yours.

But I do like your response. For some reason I am looking for a technical answer. But you could be 100% right with your responses. Everybody seems to look at technicality for reasoning. If you give a technical answer that makes sense, it’s easier for someone to buy into it. Rather than for you to give a basic response and people blow you off.

I want there to be a technical answer, but am afraid there might not be one.

Thanks for contributing to discussion. Not many people offer much feedback, more like a 1 response and done. Hopefully we can change that

[Relay1]

That is another excellent point! But there still could be many other variables that won’t allow you to compare apples to apples.

Humidity could be different causing lower readings. I have yet to see any correction factors for humidity. And I think we all have to agree that humidity is a major factor.

Was the transformer turned off the night before you arrived? Or the morning of? We’re there any delays in getting started with your work or mine for that fact? Temperature of windings may not have been properly documented on my part or yours.

But I do like your response. For some reason I am looking for a technical answer. But you could be 100% right with your responses. Everybody seems to look at technicality for reasoning. If you give a technical answer that makes sense, it’s easier for someone to buy into it. Rather than for you to give a basic response and people blow you off.

I want there to be a technical answer, but am afraid there might not be one.

Thanks for contributing to discussion. Not many people offer much feedback, more like a 1 response and done. Hopefully we can change that

Sometimes the best we can do is follow our training and "Best-Practice Methodology" in testing equipment. We have no control over how it was done before, so we need to do all we can to insure our measurements can stand up to the scrutiny of others. Connections, temperature correction factors and weather all play a part in our accuracy.

[JoshuaSWaller79]

The tests are to measure resistance between the windings and ground, so would you not want to apply the test voltage equally across the device?

Shorting the winding to itself (H1-H2, X1-X2) ensures that there is no difference of potential (voltage) between one end of the winding to the other. With no difference of potential this helps prevent any voltage being transformed to the other winding during the initial application of DC (transients). Shorting the other windings helps ensure that if there is any energy transformed there is no voltage across the windings, and they have to be tested anyway.

Hope this helps.

[SecondGen]

See, I've heard this explination before but I don't understand its effects on the test. Since we are talking DC, I imagine that any transformation in the windings would be very, very brief until the winding under test is polarized. Once voltage is sustained, any induction to the other windings would dissipate almost immediately.

All of this happens during initial charging current, which is usually only a few seconds, so whats the point? I imagine the difference in potential cant be more than a couple volts, so how much induction is actually taking place?

The only thing that instantly comes to mind is preventing voltage spikes that could possibly damage the low side insulation, depending on the rated voltage. Is there something else Im not seeing here?


Shorting the winding to itself (H1-H2, X1-X2) ensures that there is no difference of potential (voltage) between one end of the winding to the other. With no difference of potential this helps prevent any voltage being transformed to the other winding during the initial application of DC (transients). Shorting the other windings helps ensure that if there is any energy transformed there is no voltage across the windings, and they have to be tested anyway.

Hope this helps.

[JoshuaSWaller79]

You bring up a valid point about the voltage spikes potentially damaging the insulation.

The transients would be present during the initial engergization of the tester as well as when it is de-energized. Depending on the ratios involved the transformed voltage could be dangerous. It also seems like a safety precaution to me for the personnel testing the equipment.


See, I've heard this explination before but I don't understand its effects on the test. Since we are talking DC, I imagine that any transformation in the windings would be very, very brief until the winding under test is polarized. Once voltage is sustained, any induction to the other windings would dissipate almost immediately.

All of this happens during initial charging current, which is usually only a few seconds, so whats the point? I imagine the difference in potential cant be more than a couple volts, so how much induction is actually taking place?

The only thing that instantly comes to mind is preventing voltage spikes that could possibly damage the low side insulation, depending on the rated voltage. Is there something else Im not seeing here?

[wooddy]

Well, the argument against using jumpers has to do with the interconnections of the windings. Transformers are wired to where one winding jumps to another. If you look at a delta or why configuration, everything is already tied together. So some people have made the argument that you don’t need to use jumpers when testing because they are already electrically continuous.

But Paul gill and Megger and doble are not stupid people. Their procedure specifically spells out to jumper the windings. I just don’t know the exact reason why

its simply to eliminate winding inductance on the insulation being tested, stresses the winding evenly thus eliminating any winding inductance

[test11]

I am running a transformer resistance check, measuring High to Low winding. If I apply 10kV on bushing H1, what is the voltage at H2 and H3? By bonding the bushings together, I know 10kV is applied to the entire H winding. The same goes for bonding the X windings together.

Open circuit means no voltage drop. The entire high side is whatever you put in. You can prove it to yourself by taking a meter across the winding. It will read zero because there is not a load, therefore, no voltage drop.

[Relay1]

Open circuit means no voltage drop. The entire high side is whatever you put in. You can prove it to yourself by taking a meter across the winding. It will read zero because there is not a load, therefore, no voltage drop.

Take a two winding xfmr, 14400/120-240 center tapped. All bushings floating, ground removed. Energize H1 at 10kv and measure X1. Measure X2. Measure X3. Now bond H1-H2 and bond X1-X2-X3. Energize H1 at 10kv and Measure X1. You should see a difference in your readings. It may not be much of a difference and not enough for you to be concerned with but if I'm paying the bill I would want to know the procedures used are the same as the previous tests and will match the future test procedures as well.
If this was an open circuit there would be no current flow, no voltage drop and an infinite amount of resistance. What we are measuring is the current flow between the windings to develop our resistance reading. Since we have current flow from the energized winding would we not also have a voltage drop?

[test11]

Take a two winding xfmr, 14400/120-240 center tapped. All bushings floating, ground removed. Energize H1 at 10kv and measure X1. Measure X2. Measure X3. Now bond H1-H2 and bond X1-X2-X3. Energize H1 at 10kv and Measure X1. You should see a difference in your readings. It may not be much of a difference and not enough for you to be concerned with but if I'm paying the bill I would want to know the procedures used are the same as the previous tests and will match the future test procedures as well.
If this was an open circuit there would be no current flow, no voltage drop and an infinite amount of resistance. What we are measuring is the current flow between the windings to develop our resistance reading. Since we have current flow from the energized winding would we not also have a voltage drop?

The current flow is from the winding as a whole to ground or other winding with an infinite number of parallel paths of varying resistance. Definite voltage drop of 99.999 percent across the open you're reading unless the insulation is exceptionally poor(weak spot or spots) which would read well below any published specification anyway. There is no current flow through the traditional current path so the voltage difference in the energized winding would be zero. You may see micro or milivolt differences which could easily be noise already present.

As for getting different readings, you could not change a thing, retest, and still get a slightly different reading(if it doesn't peg high). The measurement is just so low to begin with(machine usually reading nanoamps). Not to mention if you're in a humid environment you now have a different surface area of bare energized parts leaking from atmosphere to ground.

Bottom line: do what you want. You said it yourself it was negligible.

[Samuel]

Well, the argument against using jumpers has to do with the interconnections of the windings. Transformers are wired to where one winding jumps to another. If you look at a delta or why configuration, everything is already tied together. So some people have made the argument that you don’t need to use jumpers when testing because they are already electrically continuous.

But Paul gill and Megger and doble are not stupid people. Their procedure specifically spells out to jumper the windings. I just don’t know the exact reason why

I expect this is to keep your DC pulse from inducing a proportional voltage pulse on the other winding. Remember DC is only DC at steady state and in the beginning of application it still creates a changing magnetic field.

[TestTim]

See, I've heard this explination before but I don't understand its effects on the test. Since we are talking DC, I imagine that any transformation in the windings would be very, very brief until the winding under test is polarized. Once voltage is sustained, any induction to the other windings would dissipate almost immediately.

All of this happens during initial charging current, which is usually only a few seconds, so whats the point? I imagine the difference in potential cant be more than a couple volts, so how much induction is actually taking place?

The only thing that instantly comes to mind is preventing voltage spikes that could possibly damage the low side insulation, depending on the rated voltage. Is there something else Im not seeing here?

I got stuck at this problem only because some guy in my team wanted to skip shorting windings of transformer. I was happy to the discussion here. I found some useful information a HV testing technique book which was published in 1970. The reason why we consider shorting winding on each side is that the impulse voltage which caused by IR/PI testing will create voltage differece and current between three phases, which might cause potectial damage to insulation or saftey hazards.318

[Primepower1]

I don't think its a misunderstanding, I've seen it in a lot of literature. This is the procedure Paul Gill lays out in his book:

224

I will post some more if I get a chance to dig through some megger manuals, I may have also seen it in "A stitch in time."

But, this article does not explain why you jumper the poles. I was told it was to equalize the potential at the ends of all bushings. Why you would do that I don't know. It may effect the DAR reading due to current flowing through the winding connections. I have never experimented to look at the differences.

[test11]

I got stuck at this problem only because some guy in my team wanted to skip shorting windings of transformer. I was happy to the discussion here. I found some useful information a HV testing technique book which was published in 1970. The reason why we consider shorting winding on each side is that the impulse voltage which caused by IR/PI testing will create voltage differece and current between three phases, which might cause potectial damage to insulation or saftey hazards.318

It states for rapidly changing voltages. This would definitely apply to power factor testing, but not a stable DC voltage after a few miliseconds.

[Primepower1]

It states for rapidly changing voltages. This would definitely apply to power factor testing, but not a stable DC voltage after a few miliseconds.

So, I agree, you do not need to short the X and H windings for a DC PI test, but it is very important to short them for an AC Power Factor Test.

[ibeam01]

This was an interesting read. I am new to the field, and we do not short the windings when doing a 10kV megger/PI test. Just one H to one X.

We do "jumper" the unused windings to ground when doing AC hi-pot tests.

I hadn't put much thought into the "WHY" though.

[mfndoc]

Had an interesting discussion on the job today about shorting together transformer windings when doing a PI test. Every technical manual I've come across tells you to do it but I can't seem to find a clear answer why.

I've asked around and get different answers. Some guys argue it reduces stress on the winding and others have said you do it to reduce induction, which opens up a whole new topic of discussion.

Personally, I do it because that's the way I was always taught, but I do see how you can argue that its unnecessary because transformer windings are normally connected together anyway. I also know that using jumpers can reduce your actual reading because I've seen it happen.

What do you normally do in the field? Looking for opinions on whether or not you use jumpers and why.


I would say that your megger has an AC power supply, no inverter makes a perfect DC signal, it is normally more of a ripple DC, which is enough to introduce an AC component, which would make inductive reactance a factor, all be it probably a small one, but a factor none the less.

[testguy]

Responses from our LinkedIn Page:


Frank says: "Been a while but back in the day we were taught that “shorting” the windings “ nulled out the resistance of the windings because you were only interested in the insulation resistance. Consider a single phase power transformer instead of a 3 phase internal connected. Megger instructions still have you short the single winding. Old sets even had a null switch somewhat like a Doble UST. With the windings shorted the test voltage is applied across the insulation and not the windings. Just a side note don’t ever try to measure the resistance of a winding with a AC DLRO just to test the theory. It can’t handle the field collapse and will let the smoke out while your hand is still on the test switch. Use a winding resistance set instead.
Happy Testing"

https://www.linkedin.com/feed/update/urn:li:activity:6630979620424548353?commentUrn=urn %3Ali%3Acomment%3A%28activity%3A663097962042454835 3%2C6631236447695355905%29


Jeff says: "Glad to see this debate. I too short the windings because the guys who had been doing it longer than me taught me that way. Likewise, that's how I teach it at PHe Services but I do wonder. I've seen it done both ways. Given what we know about multiple resistors in parallel, I find it’s easy to affect the test results negatively using jumpers if your not careful."

https://www.linkedin.com/feed/update/urn:li:activity:6630979620424548353?commentUrn=urn %3Ali%3Acomment%3A%28activity%3A663097962042454835 3%2C6632262519127896064%29

[backpack96]

Same potential at each point. BTW a PI on oil filled equipment is useless no matter what NETA states. In the instruction manual from Megger (Biddle was the expert on DC testing) states that a PI on an oil filled is useless. Dry type equipment such as motors, generators, dry type transformers, cables and etc. is a very good test to perform before applying a high voltage.

[cycle61alt]

Same potential at each point. BTW a PI on oil filled equipment is useless no matter what NETA states. In the instruction manual from Megger (Biddle was the expert on DC testing) states that a PI on an oil filled is useless. Dry type equipment such as motors, generators, dry type transformers, cables and etc. is a very good test to perform before applying a high voltage.

I was curious about this, and have performed PI tests on many, many oil filled transformers, so I went looking for a source. Found the following quote in Megger's "A Guide to Diagnostic Insulation Testing above 1KV"


It is also interesting to note that many people have tried
to use the PI test on oil-filled transformers and cannot
understand why a known good transformer gives them
results close to 1. The answer is simple. PI testing is not
appropriate for oil-filled transformers. The concept
depends on the relatively rigid structures of solid
insulating materials, where absorption energy is required
to reconfigure the electronic structure of comparatively
fixed molecules against the applied voltage field.
Because this process can go to a theoretical state of
completion (at “infinite time,” which obviously cannot
be achieved in the practical field, but can be reasonably
approximated), the result is a steady diminution of
current as molecules reach their “final” alignment.
Because the PI test is defined by this phenomenon, it
cannot be successfully applied to fluid materials since
the passage of test current through an oil-filled sample
creates convection currents that continually swirl the oil,
resulting in a chaotic lack of structure that opposes the
basic premise upon which the PI test rests.

[RedBrick]

I was curious about this, and have performed PI tests on many, many oil filled transformers, so I went looking for a source. Found the following quote in Megger's "A Guide to Diagnostic Insulation Testing above 1KV"

Also if you have access to IEEE C57.152 Guide for Diagnostic Field Testing of Fluid-Filled Power Transformers, Regulators and Reactors the following is said:

"The polarization index method should not be used to assess insulation condition in new power transformers."

"The polarization index for insulation liquid is always close to 1. Therefore, the polarization index for transformers with low conductivity liquids may be low in spite of good insulation condition."

[LHammermeister]

That is another excellent point! But there still could be many other variables that won’t allow you to compare apples to apples.

Humidity could be different causing lower readings. I have yet to see any correction factors for humidity. And I think we all have to agree that humidity is a major factor.

Was the transformer turned off the night before you arrived? Or the morning of? We’re there any delays in getting started with your work or mine for that fact? Temperature of windings may not have been properly documented on my part or yours.

But I do like your response. For some reason I am looking for a technical answer. But you could be 100% right with your responses. Everybody seems to look at technicality for reasoning. If you give a technical answer that makes sense, it’s easier for someone to buy into it. Rather than for you to give a basic response and people blow you off.

I want there to be a technical answer, but am afraid there might not be one.

Thanks for contributing to discussion. Not many people offer much feedback, more like a 1 response and done. Hopefully we can change that

Hello,

As is standard practice with transformer testing you always complete your AC tests before DC as with DC you will magnetize the transformer core in such a way that the AC tests will have problems. While we all know the an insulation resistance tests will have a very low current however there is still current flow through the windings to ground or the oppossite winding. I believe this is why you would short them all to ensure each winding is at the same potential.

A recent example of a megger causing me some greif was when we were completing maintenance transformer testing work. I was running the Doble PF set and an electrican of mine without telling me started doing Polarization index tests on the transformers ahead of me. I did not have an issue with PF(on any of the xfmrs) or excitation on the first transformer as he had not tested this one yet. However, when I went to complete excitation testing on the following transformers my numbers were way out of whack. After an hour of troubleshooting the worker told me that he had started testing ahead of me and was not jumpering the bushings of each bushing together (all highs together, x's together). I did notice every time I ran my excitation test the results would get slightly better on the winding under test. I confirmed with Doble that the xfmr was most likely magnetized unevenly for lack of a better term although the megger being such a low current doesn't typically cause magnetization issues.

Not quite a technical reason but a piece of recent field experience.

[RabbleRabble]

Had an interesting discussion on the job today about shorting together transformer windings when doing a PI test. Every technical manual I've come across tells you to do it but I can't seem to find a clear answer why.

I've asked around and get different answers. Some guys argue it reduces stress on the winding and others have said you do it to reduce induction, which opens up a whole new topic of discussion.

Personally, I do it because that's the way I was always taught, but I do see how you can argue that its unnecessary because transformer windings are normally connected together anyway. I also know that using jumpers can reduce your actual reading because I've seen it happen.

What do you normally do in the field? Looking for opinions on whether or not you use jumpers and why.

I would think this is more of an equipment and safety issue. Charging up a big inductor will create an impulse as the EMF collapses. Using jumpers to short the winding instead of making the tool eat that impulse. That’s just a guess, I never jumper The windings for a PI test.