Why is ferrite core hot?

Thread Starter

the kid

Joined Jan 4, 2015
80
I am creating an AC-DC converter circuit that has the following requirements:
Input voltage: 220V / 50 Hz
Output voltage: 200 VDC
Output power: 500W
I use the principle diagram as shown below.
halfbridge.png
https://ibb.co/q0Bt34K
My transformer uses EE42 core
When the output current is 2.5A, I see the ferrite core is heating up.
I don't understand why this happened, and how to fix it.
My circuit is working at 33 KHz pulse frequency
Please help me
 

crutschow

Joined Mar 14, 2008
23,516
Two primary things can cause the core to heat.
  1. Resistive losses in the transformer winding wire due to the primary and secondary currents.
  2. Ferrite core loss to to such factors as eddy currents, and magnetic hysteresis.
Depending upon the primary source of the heat, you either need a better core, or a larger core with larger wire in the windings.

Also Litz wire will reduce skin effect resistance in the transformer.
33kHz pulses have a lot of high frequency harmonics that can cause significant skin effect losses.
For example at 100kHz the effective current depth is only about 0.2mm.
This means the any wire larger than about 26AWG will have its effective resistance increased at 100kHz and above.
 

xox

Joined Sep 8, 2017
372
I am creating an AC-DC converter circuit that has the following requirements:
Input voltage: 220V / 50 Hz
Output voltage: 200 VDC
Output power: 500W
I use the principle diagram as shown below.
View attachment 176241
https://ibb.co/q0Bt34K
My transformer uses EE42 core
When the output current is 2.5A, I see the ferrite core is heating up.
I don't understand why this happened, and how to fix it.
My circuit is working at 33 KHz pulse frequency
Please help me
Well the magnetic flux within the core always generates some heat. Depends on the frequency too. Probably just need to use a beefier transformer.
 

MisterBill2

Joined Jan 23, 2018
4,050
If the heating of the core is not due to the wire in the windings becoming heated by the current, then you need to look at the specification of the core as far as power handling capability. Cores do have limits in that area. You will probably need to check the manufacturers website since a distributor may not have that information available.
 

MrAl

Joined Jun 17, 2014
6,626
I am creating an AC-DC converter circuit that has the following requirements:
Input voltage: 220V / 50 Hz
Output voltage: 200 VDC
Output power: 500W
I use the principle diagram as shown below.
View attachment 176241
https://ibb.co/q0Bt34K
My transformer uses EE42 core
When the output current is 2.5A, I see the ferrite core is heating up.
I don't understand why this happened, and how to fix it.
My circuit is working at 33 KHz pulse frequency
Please help me
Hi,

There are a lot of possibilities.

FIrst, i assume you are talking about the output transformer.
33kHz isnt that high for a converter using MOSFETs but you still have to make sure you are using the right core material.
Assuming that, you have to make sure you are using the right wire size and number of strands for the expected current.
You also have to check for core saturation which can cause lots of problems.

A design point that i dont see being observed here is associated with the output filter capacitor input filter. That's not usual for higher power converters as there is usually some external inductance or at least some leakage inductance in the output transformer to reduce peak rectifier current which also traces all the way back to the power MOSFETs. There is no external inductor so the question then becomes do you have any leakage inductance built into the output transformer? Since we dont see any winding specs (except for turn count) we cant know by looking at the schematic. In conjunction with that, for safety reasons are the two windings on the output transformer separated or is the secondary wound right on top of the primary? There needs to be a separation and that also provides some leakage inductance. You could check the peak current to see if it is enough.

Finally, there is the primary DC offset current issue. With converters that dont mitigate that directly there could be enough to cause problems. That has to be checked.

Another interesting point though is that i dont see any voltage feedback. Modern converters always have some form of feedback to stabilize the output voltage during input changes or load changes.
 
Last edited:

mvas

Joined Jun 19, 2017
538
My transformer uses EE42 core
When the output current is 2.5A, I see the ferrite core is heating up.
I don't understand why this happened, and how to fix it.
You said, "... heating up ..."
That tells us nothing, because "heating up" is expected.
What is the ambient temperature?
What is the operating temperature?
Now compute ... your actual temperature rise = 10°C, 20°C, 30°C, 40°C or more ?

When you did the calculations to select the EE42 core, the core cross-sectional area, the core volume and the material type,
and given the # turns, 500 watts and 33 Khz frequency, then what was your expected temperature rise calculation?

You need to compare your "expected" temperature rise vs the "actual" temperature rise, to determine if your transformer is operating properly or not.
 

MrAl

Joined Jun 17, 2014
6,626
You said, "... heating up ..."
That tells us nothing, because "heating up" is expected.
What is the ambient temperature?
What is the operating temperature?
Now compute ... your actual temperature rise = 10°C, 20°C, 30°C, 40°C or more ?

When you did the calculations to select the EE42 core, the core cross-sectional area, the core volume and the material type,
and given the # turns, 500 watts and 33 Khz frequency, then what was your expected temperature rise calculation?

You need to compare your "expected" temperature rise vs the "actual" temperature rise, to determine if your transformer is operating properly or not.

Hi,

Yes good point. I assumed that because the problem was presented it really was a problem which implies it was overheating not simply heating up.
 

MisterBill2

Joined Jan 23, 2018
4,050
As I look at the circuit more I see that there is also a DC to DC converter, and I am wondering if it is actually a switching supply putting out 12 volts and powered by the 300 volts, or is it a supply running on 12 volts and putting out 300 volts. That part is not at all obvious.
And still the question about the rating of the core is not answered, but if the output power is indeed 500 watts then the core is handling 500 watts and it may not be sized to do that. Also, how hot? Even conservatively designed power supplies get warm because nothing is 100% efficient. So some temperature rise is totally reasonable.
 

noweare

Joined Jun 30, 2017
107
Normally if a core is running hot you have to much voltage per turn. Volts per turn will set the flux density (B). Manufacturers will normally give you a power loss vs flux density and at different frequencies. So for a certain frequency you need to stay below a certain flux density. This is true of all magnetic cores not just ferrites.
 

Thread Starter

the kid

Joined Jan 4, 2015
80
Normally if a core is running hot you have to much voltage per turn. Volts per turn will set the flux density (B). Manufacturers will normally give you a power loss vs flux density and at different frequencies. So for a certain frequency you need to stay below a certain flux density. This is true of all magnetic cores not just ferrites.
I understand what you mean, but the circuit works with maximum pulse width. So why is it that only when loaded, the ferrite core warms up significantly. When the circuit has no load, ferrite cores are only slightly warm.
 

kubeek

Joined Sep 20, 2005
5,625
That is correct, hysteresis heating should occur all the time, not just under load. Then what you see must be coming from the joule heating of the windings.
 

Thread Starter

the kid

Joined Jan 4, 2015
80
That is correct, hysteresis heating should occur all the time, not just under load. Then what you see must be coming from the joule heating of the windings.
Thank you, I will increase the area of the wire and then test again.
Hopefully next time people continue to care about this topic
 

MrAl

Joined Jun 17, 2014
6,626
I understand what you mean, but the circuit works with maximum pulse width. So why is it that only when loaded, the ferrite core warms up significantly. When the circuit has no load, ferrite cores are only slightly warm.
Hello,

It's not just the frequency when a square wave is applied, it's the pulse width. Wider pulse width works like higher voltage and with load the pulse width must get wider to support the voltage.
If the pulse width is constant then the extra current in the primary could be doing it along with any DC offset current.
See what happens with just one half of full load. That's often a test point too with converters.
 

noweare

Joined Jun 30, 2017
107
So the power loss of the core should not change as long as you are always driving the input of the transformer.
with the same input. If the core is heating up when the circuit is drawing significant current then it has to be the power losses of the
wire is creating heat that is also warming up the core material.
 

MrAl

Joined Jun 17, 2014
6,626
Hi again,

Oh i forgot to mention. The primary DC offset current changes with load because the driver transistors voltage drops change with load. Thus the imbalance can get worse or better with load. If it gets worse then the primary heats up.

If you indeed follow the recommendation of having the two windings separated for safety reasons, then you should be able to measure the temperature of each winding individually. That could tell you if just one winding is causing it. You have to be very careful if you feel the windings with a finger though as that could be dangerous with the voltages involved here.

You could also look at input current as you increase load gradually.
 

Thread Starter

the kid

Joined Jan 4, 2015
80
Hello,

It's not just the frequency when a square wave is applied, it's the pulse width. Wider pulse width works like higher voltage and with load the pulse width must get wider to support the voltage.
If the pulse width is constant then the extra current in the primary could be doing it along with any DC offset current.
See what happens with just one half of full load. That's often a test point too with converters.
I use a half-bridge diagram, so I think the DC offset current at the primary coil will be zero
 

BobaMosfet

Joined Jul 1, 2009
836
@the kid
A primary question I have is why you even have a DC DC converter? If you're converting from AC to DC a transformer and full-wave bridge rectifier with appropriate caps/filtering gets you where you need to be. You can control the voltage by adding a tap on the secondary side. As far as output current is concerned, that depends on how much power comes in from the input. Current can be whatever your load will draw, limited only by what your components can handle. If you want to limit current output, caps on the A/C side are the way to go, because you can limit current electromagnetically, rather than thermally, so there is no heat issue.
 

MisterBill2

Joined Jan 23, 2018
4,050
I understand what you mean, but the circuit works with maximum pulse width. So why is it that only when loaded, the ferrite core warms up significantly. When the circuit has no load, ferrite cores are only slightly warm.
When there is no load there is not much power being transferred by the core. And the effort to determine if it is the windings causing the heat is as simple as putting your finger lightly on the windings. If the core is warm and the winding is not, then it is the core creating the heat, while if the core is warm and the winding is HOT then it is indeed conductor resistance. OR it may be both.
 

Thread Starter

the kid

Joined Jan 4, 2015
80
When there is no load there is not much power being transferred by the core. And the effort to determine if it is the windings causing the heat is as simple as putting your finger lightly on the windings. If the core is warm and the winding is not, then it is the core creating the heat, while if the core is warm and the winding is HOT then it is indeed conductor resistance. OR it may be both.
I experimented and found that: when loaded, ferrite cores are hotter when there is no load. And I really don't understand this phenomenon
 

MisterBill2

Joined Jan 23, 2018
4,050
Now it becomes clear to me that the DC/DC converter is the +12 volt supply for the low voltage circuitry. So that part is clarified. It took a bit more thinking there.
But I find that your answer is not completely clear to me, but I believe that you are saying that when loaded, ferrite cores are hotter THAN when there is no load. That is what I would expect.
The core heating when the load is high is due to the fact that it is working much harder when it is transferring energy to the secondary winding. The core is what is transferring the magnetic flux, and so with a load much more flux is passing through the core. Unfortunately I am not able right now to explain the exact mechanism of why that is causing it to heat, but that is what is going on. In a classic drawing of the B/H curves, under load the opening inside the loop gets bigger and that means more work is being done, so thus more heat is created because of more losses.
 
Top