Switching transformer is hot

Thread Starter

MacIntoshCZ

Joined Sep 4, 2019
22
Hello everyone,
I created switching transformer 39 turns primary , 16 secondary, effective are 211mm2. I calculate it for 360V to be at 250mT when having 50% duty cycle and 50khz frequency. Primary voltage is actually 300V so i should be at aprox 165mT.
I used 0,5mm wire. One for primary, two for secondary. Peak power is 200W. Secondary voltage is 120V
200W/300V ->0,67A
200W/120V -> 1,67A
0,5mm should handle 1,17A, If i take 6A for 1mm2
If i remember correctly primary lenght was 1,5m and secondary 0,75m. Aprox...
I already created thread at electrotechonline, but more people more know...
Here is link to created thread
https://www.electro-tech-online.com/threads/switching-transformer.161273/

I think winding is getting hot. I took measurement with current sense transformer and values are aprox same as my calculations up.
Core losses at 50Khz 165mT should be around 3W.
I though selected wire would be ok.
Can someone refute it for me?
Also i using 2 switch forward topology.
Thanks for help
1617432957420.png
 

DickCappels

Joined Aug 21, 2008
7,157
Depending upon the physical layout, you are running at a high enough frequency to warrant using multiple parallel conductors for the windings. I had a transformer and some inductors that you could not tough when running at 64 kHz until I switched to many parallel strands of much smaller magnet wire.

Some call this skin effect losses but it is really eddy currents induced by the huge gradient in the changing magnetic field next to the core.

16 turns doesn't sound like much for 300V input. That could be related to your worry.
 

Thread Starter

MacIntoshCZ

Joined Sep 4, 2019
22
16 turns doesn't sound like much for 300V input. That could be related to your worry.
Hi Dick,
I used 39 turns on primary to aware core saturation and chose 16 turns becouse when i am at 50% duty cycle i get 60V average at output. Can you please tell me more about what you mean?
 

ronsimpson

Joined Oct 7, 2019
1,157
From your data the switch is on 70% of the time. (where does 0.7/100k come from)
You have the input voltage at 85 volts but I think you are running more like 300 volts.
I think the top half of the formula in #4 is (volts X time X10^8). It is volts and time on the core.
Is the core hot with almost no load but keep the duty cycle high? I think the flux is too high.
 

Thread Starter

MacIntoshCZ

Joined Sep 4, 2019
22
EQ post in #4 is ftom application note.
When you entry 300V 50% 50khz 2,11cm2 1700mT
V = 300V
ton = 1/50000
0,7 stands for duty cycle, in my case its 0,5
2 - at the moment dont know
Ae- effective are of core in cm -> in my case 2,1
B- Mr. Tesla -> in miliTesla
 

Thread Starter

MacIntoshCZ

Joined Sep 4, 2019
22
Try using many parallel strands of #30 or even smaller. Subjectively, I think 0.3 mm would still have a lot of eddy currents in it.
I dont understand calculator for skin effect then (that i posted above).
Its should add AC resistance to dc resistance.
Also i calculated skin depth for 50khz its 0,29mm for copper.
With diameter 0,3mm I'm halfway there

1617628950596.png
*sorry skin effect is not eddy current...
 
Last edited by a moderator:

DickCappels

Joined Aug 21, 2008
7,157
I am not talking about skin effect. I am talking about eddy currents induced in the wire by the large changing magnetic field from the core. The lines of force in this field cut oright across the cross section of the wire, inducing eddy currents. Everybody called the effect skin effect (which is a different effect -skin effect is increase in resistance but eddy currents are currents made by shorting a changing magnetic field) until the early 1980's when a paper published by an engineer at Tektronix explained the effect and I was lucky to have been listening because only a few months later I was working on a 64 kHz switching circuit with a very hot transformer.

Here is an excellent article by Lloyd Dixon who used to be a prolific writer and speaker on the subject of power supplies and magnetics. It explains the eddy current in windings problems.
https://www.ti.com/lit/ml/slup197/slup197.pdf

I am sure you also have some skin effect losses but in power supplies and similar circuits using ferrite cores the eddy currents are the biggest problem.
 

Thread Starter

MacIntoshCZ

Joined Sep 4, 2019
22
I am not talking about skin effect. I am talking about eddy currents induced in the wire by the large changing magnetic field from the core. The lines of force from this field cut right across the cross section of the wire, inducing eddy currents. Everybody called the effect skin effect (which is a different effect -skin effect is increase in resistance but eddy currents are currents made by shorting a changing magnetic field) until the early 1980's when a paper published by an engineer at Tektronix explained the effect and I was lucky to have been listening because only a few months later I was working on a 64 kHz switching circuit with a very hot transformer.

Here is an excellent article by Lloyd Dixon who used to be a prolific writer and speaker on the subject of power supplies and magnetics. It explains the eddy current in windings problems.
https://www.ti.com/lit/ml/slup197/slup197.pdf

I am sure you also have some skin effect losses but in power supplies and similar circuits using ferrite cores the eddy currents are the biggest problem.
Thanks for help. Where i live its hard to get litz wire. If i want lower losses i need to decrease flux and frequency.
I dont want to decrease frequency. If i add more turns in primary there will be more copper -> higher losses?
But if deacrese two times Bmax, losses will be 4 times lower.
 

DickCappels

Joined Aug 21, 2008
7,157
I did not use Litz wire. I had spools of various diameter enameled wire. This might give you an idea:

I figured out what the total copper cross section of all of the parallel inductors needed to be then I took some fine magnet wire (I have gone down to #36 but the wire gets difficult to handle) and wrapped it around one pillar in our factory, then I walked to the next pillar and then around that next pillar and back to the first. I did this many times until I had the number of parallel strands I thought I needed. Then I took an electric drill and put a small wire hook in the chuck and turned it on to twist the conductors. A slight tug at the end of twisting assured that the wire would not un-twist itself.

This method works just as well if using an 8 year old boy to hold a screwdriver while stringing the parallel turns and twisting with the drill, only thing is that sometimes 8 year old boys are not always available.
 

Thread Starter

MacIntoshCZ

Joined Sep 4, 2019
22
I did not use Litz wire. I had spools of various diameter enameled wire. This might give you an idea:

I figured out what the total copper cross section of all of the parallel inductors needed to be then I took some fine magnet wire (I have gone down to #36 but the wire gets difficult to handle) and wrapped it around one pillar in our factory, then I walked to the next pillar and then around that next pillar and back to the first. I did this many times until I had the number of parallel strands I thought I needed. Then I took an electric drill and put a small wire hook in the chuck and turned it on to twist the conductors. A slight tug at the end of twisting assured that the wire would not un-twist itself.

This method works just as well if using an 8 year old boy to hold a screwdriver while stringing the parallel turns and twisting with the drill, only thing is that sometimes 8 year old boys are not always available.
I am using that part with drill. But part with pillars is also very interesting. Thanks for sharing!
 

Thread Starter

MacIntoshCZ

Joined Sep 4, 2019
22
I did not use Litz wire. I had spools of various diameter enameled wire. This might give you an idea:

I figured out what the total copper cross section of all of the parallel inductors needed to be then I took some fine magnet wire (I have gone down to #36 but the wire gets difficult to handle) and wrapped it around one pillar in our factory, then I walked to the next pillar and then around that next pillar and back to the first. I did this many times until I had the number of parallel strands I thought I needed. Then I took an electric drill and put a small wire hook in the chuck and turned it on to twist the conductors. A slight tug at the end of twisting assured that the wire would not un-twist itself.

This method works just as well if using an 8 year old boy to hold a screwdriver while stringing the parallel turns and twisting with the drill, only thing is that sometimes 8 year old boys are not always available.
=)
 

Janis59

Joined Aug 21, 2017
1,297
There are no such diagnoze like TRANSFORMER is hot. Should stand or the CORE becomes hot or WINDINGS become hot. First case correct the winding count (means Bmax) and frequency. Both may gives this effect. Second symptoms may be due too small cross section or too strong Focault effect. In first just take a thicker wire, in second take the thinner wire and make a Litz Wire of it. As the thumb-rule, at 20 kHz the wire should not exceed some 0.6 mm, but ar 200 kHz already 0.2 mm.
 

Janis59

Joined Aug 21, 2017
1,297
Receip for Litz fabrication. Take the Excel and count how long You need. Then find in the long corridor places where may beat two nails with distance about 20% larger than length. Wind between them so many wounds how much You need a total cross section be satisfied. Then put the bunch one side in the battery hand drill, fix the hook into it and swirl the wire bunch until some 1-2 turns to cm are gained. Then release the far end of litz off the nail. Few hundreds of meters long You may create in this manner
 

Thread Starter

MacIntoshCZ

Joined Sep 4, 2019
22
This guy must use a lot of "Litz" wire. I admire his enginuity.
Hi,
i winded 16x 0,118 diameter litz wire for primary and 32x 0,118 for secondary.
After 5 minutes i get around 45C and slowly rise up.
I dont know whats normal temperature. But i would like to be under 40C.
200W load
 

DickCappels

Joined Aug 21, 2008
7,157
That suggests that smaller diameter wire and more strands might help. Also a larger core.

Any particular reason you prefer a 40° limit? That doesn't leave room for much temperature rise, especially if operating in the tropics where the ambient can itself reach 40° (like it does here).
 

Thread Starter

MacIntoshCZ

Joined Sep 4, 2019
22
Any particular reason you prefer a 40° limit? That doesn't leave room for much temperature rise, especially if operating in the tropics where the ambient can itself reach 40° (like it does here).
I still dont have a capton tape :). So i am using regular / electrical tape.
 

DickCappels

Joined Aug 21, 2008
7,157
Before I had Kapton I just used simple mylar tape. Designed many switching power stages with that and had no problems. I imagine that getting some good tape that will not creep or melt at slightly elevated temperatures would be a priority.
 
Top