transformer heating up

[eigenvictor]

I needed a 120V to 12V  step down transformer, so I rewound an old transformer (both primary and secondary) and it was working fine but I needed to add some more insulation.  So I reassembled it and now I noticed that it is getting warm just after a few minutes of idling. So I measured the idle current and all of a sudden it increased sharply from 280mA to almost 500mA!  My first thought was a short in the winding but the secondary voltage hasn't changed and the resistance measures the same.  Since these are old laminations and I didn't re-varnish them, I would expect some increase in Eddy currents and a bit of extra heat, but they shouldn't contribute that much at the mains frequency.  So I don't understand a sudden current spike from 280mA to 500mA after basically no changes just disassembling it and putting everything back together.  What do you guy think?  It is a smallish E core transformer rated at about 100W so I used about 4 turns per volt which sounds about right based on the cross sectional area and Bmax of about 1.2T (just a guess as I have no specs).  Even 280mA sounds a bit too high and it was probably slightly into saturation.   Now it is obviously even more saturated but is working fine, yet I'd like to understand what is going on.

[Refrigerator]

Next time you wind your transformer you should do some proper calculations, here's a video that helped me quite a bit:

You should also never plug in the transformer blindly without any protection, in that case most people just hook up a light buld in series.

[tautech]

Here's a very good site Practical transformer winding  that was linked in another thread:
http://ludens.cl/Electron/trafos/trafos.html

[wraper]

Now it is obviously even more saturated but is working fine, yet I'd like to understand what is going on.

You winded less turns than in original primary winding.

[eigenvictor]

I know how to do transformer calculations: the Faraday's law and Ampere's law combined yield the universal transformer equation which needs to be solved for N. It is pretty straightforward.  And like I said, I got it working just fine.   

The problem is this.  I had to disassemble it and then re-assembled the laminations back in place and that caused the current increase which I don't understand.  I have not changed the number of turns on either primary or secondary. 

I just wanted to rule out a possibility of this happening due to the disturbed varnish coating on the laminations. 

[Refrigerator]

I know how to do transformer calculations: the Faraday's law and Ampere's law combined yield the universal transformer equation which needs to be solved for N. It is pretty straightforward.  And like I said, I got it working just fine.   

The problem is this.  I had to disassemble it and then re-assembled the laminations back in place and that caused the current increase which I don't understand.  I have not changed the number of turns on either primary or secondary. 

I just wanted to rule out a possibility of this happening due to the disturbed varnish coating on the laminations.

Did you wind your primary side tightly, making it all nice and putting insulation between layers ?
I've seen transformers where the primary was wound without insulating material between layers and it shorted yout and caused an arc over, yours couls be similar.

[eigenvictor]

I know how to do transformer calculations: the Faraday's law and Ampere's law combined yield the universal transformer equation which needs to be solved for N. It is pretty straightforward.  And like I said, I got it working just fine.   

The problem is this.  I had to disassemble it and then re-assembled the laminations back in place and that caused the current increase which I don't understand.  I have not changed the number of turns on either primary or secondary. 

I just wanted to rule out a possibility of this happening due to the disturbed varnish coating on the laminations.

Did you wind your primary side tightly, making it all nice and putting insulation between layers ?
I've seen transformers where the primary was wound without insulating material between layers and it shorted yout and caused an arc over, yours couls be similar.

The primary is wound very tight and neat, not "random" wound.  I'd say the fill factor is about .90. That's the best I can do since I wind by hand.   But no insulation between each layer, only between the primary and secondary.  Yes my first thought was that maybe I damaged wire while reassembling the laminations and it shorted out.  But that would have changed the turn ratio and as a consequence the secondary voltage, correct?  But the voltage on secondary is exactly the same as before.  I also measured the primary resistance and it read about the same but that is a less reliable test since my multimeter is probably not accurate enough.   

[The Electrician]

You say "But the voltage on secondary is exactly the same as before."  If it was already a 120V to 12V transformer, why did you rewind it?

It's too bad you didn't measure the exciting current, unstack the core, restack it and measure the current with no changes to the windings.  Then you would have known that any increase in exciting current was due to increased losses in the core alone.  If you have another similar transformer, you could try this.

But to get to your current situation, consider this:  a transformer in good condition connected to the grid with no load.  The core will not heat up very fast.  If your rewound transformer's core is heating up, but the windings are not, then it would be reasonable to believe the increased losses are due to loss of insulation between the laminations.  The windings in an unloaded transformer should not noticeably heat up at all except by heat conducted from the core, and that takes quite a bit of time.  If the windings are heating up quickly then you have a problem there.

Make sure the butt joints between each E lam and its mating I lam are tight.  I use a rubber mallet to pound the joints tight.

Commercial laminations are sometimes coated with powdered magnesium silicate to provide insulation between laminations. A lower cost method they use is to form an oxide coating on the laminations by heating them with steam.  You might try coating your laminations with something to provide additional insulation; about the only thing I can think of that everyone would have around for this purpose would be flour or corn starch.  This might be the best thing for a permanent fix!  But it would be an experiment to find out if your problem is indeed increased losses due to insufficient insulation between laminations.

[eigenvictor]

You say "But the voltage on secondary is exactly the same as before."  If it was already a 120V to 12V transformer, why did you rewind it?

I salvaged an old transformer and discarded the old wire and kept the core.    It was a 220V to 110V or something like that, it didn't really matter to me.  I just needed the laminations.  And then I wound an entirely new transformer with my own specs.

It's too bad you didn't measure the exciting current, unstack the core, restack it and measure the current with no changes to the windings.

That's exactly what I did.  When I first put everything together, I measured the magnetizing current to be 280mA which was what I expected.  I then re-stacked it (no changes in windings) and the current went up to almost 500mA.  And that is the reason I posted my question.

 Then you would have known that any increase in exciting current was due to increased losses in the core alone.

That's what I am not sure about. At first I thought it was due to a short caused by a damaged  insulation or something. But like I said, that would have changed the turn ratio and the secondary voltage.  So another possibility could be the core losses, more specifically Eddy current losses due to the disturbed insulation between the laminations. But that also sounds strange, as it didn't happen the first time around.  On the other hand, every time they are disassembled and re-stacked, some more insulation is lost, so who knows.

 If you have another similar transformer, you could try this.

But to get to your current situation, consider this:  a transformer in good condition connected to the grid with no load.  The core will not heat up very fast.  If your rewound transformer's core is heating up, but the windings are not, then it would be reasonable to believe the increased losses are due to loss of insulation between the laminations.

Yeah, when the heat is caused by the copper, i.e. when the wire ampacity is exceeded, then obviously the windings will get hot first and then the core will start heating up as a result, acting as a heat sink.  But I think in my case it is the opposite, windings are "cold" but the core is getting warm.

 The windings in an unloaded transformer should not noticeably heat up at all except by heat conducted from the core, and that takes quite a bit of time.  If the windings are heating up quickly then you have a problem there.

No, windings are not heating up.  It is the core that is heating up.  The primary is a 22 AWG wire, so it should be able to handle 500mA with no issues.

Make sure the butt joints between each E lam and its mating I lam are tight.  I use a rubber mallet to pound the joints tight.

A rubber mallet sounds like good idea.  One thing I noticed is the bobbin flanges are  bowed outwards slightly (wire pushing on them on the edges) so the oppositely facing E's  are not quite touching each other so there is a tiny little air gap between them.  But in theory that would just decrease the flux slightly without causing the heat related losses...  I know that if I don't tighten the screws well, it will hum, and that's vibration, and vibration is heat. But the way it is currently assembled, it is very quiet.

Commercial laminations are sometimes coated with powdered magnesium silicate to provide insulation between laminations. A lower cost method they use is to form an oxide coating on the laminations by heating them with steam.  You might try coating your laminations with something to provide additional insulation; about the only thing I can think of that everyone would have around for this purpose would be flour or corn starch.  This might be the best thing for a permanent fix!  But it would be an experiment to find out if your problem is indeed increased losses due to insufficient insulation between laminations.

Thank you for your input.  I will definitely try to re-coat them. After all, it is either that, or a short somewhere.

[Seekonk]

I was in Japan working on some equipment that was supposed to 50Hz but was over heating.   With no load connected to the transformer I raised the voltage going to the transformer with a variac and monitored the current.  At a certain voltage it was clear a knee was reached where the current suddenly ramped up indicating the core was saturating.  You  should connect a buck transformer to drop the voltage 20V and see if it is still overheating.

[The Electrician]

I dug out a 100VA transformer from my pile and measured its no load performance. I get:

Exciting current = 95 mA (120 VAC applied to primary)
Power factor = .484
Core loss = 5.5 watts

This power factor of about .5 is typical of several small transformers I checked.  If your transformer's PF were about .5, then with 500 mA exciting current, you have a transformer loss of about 30 watts!  No wonder it's getting hot fast!

But even with an exciting current of 280 mA, that's a loss of 16.8 watts, which is too much for a decent 100 VA transformer.

Maybe you were already having the problem the first time you re-stacked the core, and subsequent re-stackings just made it worse.

Too bad you didn't measure the exciting current while the original factory windings were in place and you hadn't re-stacked it even once.

Anyway, I'm getting an exciting current of 95 mA for a 100 VA transformer which is more reasonable than what you're getting.

You might have a look at the waveform of the exciting current with a scope, exercising due care because you would be measuring current from a grid connection; I use a clamp on current sensor for safety.

[pikachu]

I needed a 120V to 12V  step down transformer, so I rewound an old transformer (both primary and secondary) and it was working fine but I needed to add some more insulation.  So I reassembled it and now I noticed that it is getting warm just after a few minutes of idling. So I measured the idle current and all of a sudden it increased sharply from 280mA to almost 500mA!  My first thought was a short in the winding but the secondary voltage hasn't changed and the resistance measures the same.  Since these are old laminations and I didn't re-varnish them, I would expect some increase in Eddy currents and a bit of extra heat, but they shouldn't contribute that much at the mains frequency.  So I don't understand a sudden current spike from 280mA to 500mA after basically no changes just disassembling it and putting everything back together.  What do you guy think?  It is a smallish E core transformer rated at about 100W so I used about 4 turns per volt which sounds about right based on the cross sectional area and Bmax of about 1.2T (just a guess as I have no specs).  Even 280mA sounds a bit too high and it was probably slightly into saturation.   Now it is obviously even more saturated but is working fine, yet I'd like to understand what is going on.

Do you know/can guesstimate enough parameters of the core to calculate the magnetizing current/primary inductance?

Does the core have an airgap? Did you take the airgap away?

Measure primary current and primary voltage, integrate the voltage and X-Y-plot to see what the B/H-curve looks like.

[eigenvictor]

I dug out a 100VA transformer from my pile and measured its no load performance. I get:

Exciting current = 95 mA (120 VAC applied to primary)
Power factor = .484
Core loss = 5.5 watts

This power factor of about .5 is typical of several small transformers I checked.  If your transformer's PF were about .5, then with 500 mA exciting current, you have a transformer loss of about 30 watts!  No wonder it's getting hot fast!

Wait, are you talking about a no load power factor of .5 in a typical transformer? That sounds high. I thought that since the core/iron losses are typically very small, the no load current is mostly purely reactive.  But since my core may be somewhat "defective", all bets are off.  Could be .5, could be higher.  Realizing that doesn't get me any closer to understanding WHY this is happening, but, yes, I understand that high core/iron loss will generate heat. But there is no way it is dissipating 30W! That is a lot of heat. It is getting "lukewarm" after 20-30 minutes, maybe 35-40C (105F) if I have to guess. Just curious, how do you measure the PF?  With a watt meter, a v meter and an amp meter or something like that? Or are you using one of those fancy digital scopes?  In your experience, what is the prevailing factor of the in phase current component of a properly functioning (E-core) transformer at no load?  The hysteresis loss? Eddy currents should be negligibly small in a laminated/properly insulated core at 60Hz, correct? 

But even with an exciting current of 280 mA, that's a loss of 16.8 watts, which is too much for a decent 100 VA transformer.

Maybe you were already having the problem the first time you re-stacked the core, and subsequent re-stackings just made it worse.

Too bad you didn't measure the exciting current while the original factory windings were in place and you hadn't re-stacked it even once.

Anyway, I'm getting an exciting current of 95 mA for a 100 VA transformer which is more reasonable than what you're getting.

You might have a look at the waveform of the exciting current with a scope, exercising due care because you would be measuring current from a grid connection; I use a clamp on current sensor for safety.

Yes, 280mA is still too high and the waveform was quite peaky to begin with, now it is even more so, but that is OK.  That is a compromise I had to make.  Even though this core is rated at 100VA, I think it is overly optimistic since there is just not enough room on the bobbin/winding window to accommodate the required number of turns of a reasonable thickness so it could deliver a continuous duty 100VA, say,  10A at 10V into a resistive load.  I think the winding area utilization factor turned out to be quite high and so is the fill factor.  But still the thickest wire I could get away with was 16 AWG on the secondary. That's about 200-300 mils per amp which is kind of low.  But it is OK for my needs.  After all, lots of smallish commercially made transformers are designed to work well into saturation but they seem quite efficient in dissipating heat by convection due to their size (high surface area relative to their volume).

[eigenvictor]

I needed a 120V to 12V  step down transformer, so I rewound an old transformer (both primary and secondary) and it was working fine but I needed to add some more insulation.  So I reassembled it and now I noticed that it is getting warm just after a few minutes of idling. So I measured the idle current and all of a sudden it increased sharply from 280mA to almost 500mA!  My first thought was a short in the winding but the secondary voltage hasn't changed and the resistance measures the same.  Since these are old laminations and I didn't re-varnish them, I would expect some increase in Eddy currents and a bit of extra heat, but they shouldn't contribute that much at the mains frequency.  So I don't understand a sudden current spike from 280mA to 500mA after basically no changes just disassembling it and putting everything back together.  What do you guy think?  It is a smallish E core transformer rated at about 100W so I used about 4 turns per volt which sounds about right based on the cross sectional area and Bmax of about 1.2T (just a guess as I have no specs).  Even 280mA sounds a bit too high and it was probably slightly into saturation.   Now it is obviously even more saturated but is working fine, yet I'd like to understand what is going on.

Do you know/can guesstimate enough parameters of the core to calculate the magnetizing current/primary inductance?

Does the core have an airgap? Did you take the airgap away?

Measure primary current and primary voltage, integrate the voltage and X-Y-plot to see what the B/H-curve looks like.

No, I don't the have the specs for this core, but it looks very similar to other cores I used before with a pretty typical magnetization curve approaching a "knee point" where the permeability starts to drop sharpy crossing the magnetization curve at around 200 amp turns / m^2 which corresponds to about 1.2T and the permeability of about 4000-5000.  I don't have this transformer on hand, but I can guesstimate its flux path length to be about  .2m.  The primary has 470 turns.  So that corresponds to the magnetizing current of roughly 100mA and the inductance of maybe 3H.  But the total no load current will depend on the core/iron loss which I have no idea about. But the total of 280mA  sounds way too high  in this case, so my calculations may be off or the losses are really high.  If I draw a quick phasor diagram, the active component works out to be about 250mA with a power factor of a whopping .89 and the active power loss of 25W! Something isn't right.    Well, since I really have two unknowns (could be a short in the winding or a loss due to lack of core insulation), I will first try re-coating the lams. 

[The Electrician]

Wait, are you talking about a no load power factor of .5 in a typical transformer? That sounds high. I thought that since the core/iron losses are typically very small, the no load current is mostly purely reactive.  But since my core may be somewhat "defective", all bets are off.  Could be .5, could be higher.  Realizing that doesn't get me any closer to understanding WHY this is happening, but, yes, I understand that high core/iron loss will generate heat. But there is no way it is dissipating 30W! That is a lot of heat. It is getting "lukewarm" after 20-30 minutes, maybe 35-40C (105F) if I have to guess. Just curious, how do you measure the PF?  With a watt meter, a v meter and an amp meter or something like that? Or are you using one of those fancy digital scopes?  In your experience, what is the prevailing factor of the in phase current component of a properly functioning (E-core) transformer at no load?  The hysteresis loss? Eddy currents should be negligibly small in a laminated/properly insulated core at 60Hz, correct? 

I do have a wattmeter similar to this: http://www.ebay.com/itm/NEW-SIMPSON-PANEL-WATTMETER-MODEL-79-/231497230518?pt=LH_DefaultDomain_0&hash=item35e64f68b6
which I use for measuring core loss.  But for the measurements I gave above, I used some power analysis software with my scope.  I pulled out three 100 VA transformers from the pile and measured each, with the following results (green trace is applied voltage, purple is exciting current, red is the instantaneous product of voltage and current):







It's possible to get an idea of whether the core losses are predominately hysteresis or eddy current losses.  Eddy current losses are essentially resistive (linear) losses, and when the exciting voltage is reduced, the current waveform begins to approach a sine wave.  The hysteresis losses tend to be more non-linear, especially near saturation.  Reducing the exciting voltage moves us away from saturation, so if loss currents become more in phase with the applied voltage that represents losses that are resistive in character.  Here's the capture for the third transformer above with reduced excitation voltage:



All of these 100 VA transformers have a PF near .5; this has been my experience with larger transformers.  Of course, a manufacturer could do better with better iron and more conservative excitation, but this is typical of the ordinary commercial products.

Here's what I see for a small 15 VA trahsformer; the PF is significantly lower:

[The Electrician]

Here's an example of a small (12 VA) transformer with very good iron.  The hysteresis loop is very narrow and similar to high nickel square loop core material:



With reduced excitation voltage, the core never reaches saturation, because the onset of saturation is sudden when the loop is nearly a square loop characteristic.  Most of the losses with rated excitation voltage are due to saturation, so with reduced excitation the losses are very low and almost purely resistive, probably mainly eddy current--look at the high PF:

[eigenvictor]

Here's an example of a small (12 VA) transformer with very good iron.  The hysteresis loop is very narrow and similar to high nickel square loop core material:



With reduced excitation voltage, the core never reaches saturation, because the onset of saturation is sudden when the loop is nearly a square loop characteristic.  Most of the losses with rated excitation voltage are due to saturation, so with reduced excitation the losses are very low and almost purely resistive, probably mainly eddy current--look at the high PF:

Wow, awesome screenshots, thank you for posting this.