Problem with half bridge

yoosefheidar

New member
HI
I use this schematic for a 700 watt power supply.output voltage is 36v and current is 20amps.the core is etd44/freq is 50KHZ/primary is 30 turns litz(6*0.55)/secondary is 9 turns litz(22*0.55)
circuit worked but after 1 minute secondary winding become very very hot and i can't use power supply more than 2 minutes.
what is the problem and cause of this hot secondary winding?
photo_2018-05-18_01-22-19.jpg
 

Silvio

Well-known member
The wire for secondary with 22 wires 0.55mm can surely carry the 20 amps needed. I am assuming that the winding get hot during a load of 20 amps. I do not know if you want this smps to work 100% duty.

I also want to know if your litz wire is made of only one bundle of 22 strands or divided in 3 bundles of say 7 strands each. If it is a single bundle i guess the diameter of the bundle may be too large and the effect of litz will be lost. The current may be travelling only on the circumference of the bundle and the wire strands that happen to be caught in the middle will be useless. In such a case only the wire on the perimeter is being loaded thus it is not thick enough to carry the current and getting very hot.

I would suggest that you either put the wires side by side or group them up in 6 bundles of 4 wires each. If the rating of the smps is 100% then the current carrying capacity of the wire should be calculated at 3.4A/mm². Without cooling cooling fan,

You will have a better result if you had to use an insulated copper sheet of 0.2mm thickness with the width of the bobbin. This will surely give you a better coupling and also better current handling.
 

yoosefheidar

New member
The wire for secondary with 22 wires 0.55mm can surely carry the 20 amps needed. I am assuming that the winding get hot during a load of 20 amps. I do not know if you want this smps to work 100% duty.

I also want to know if your litz wire is made of only one bundle of 22 strands or divided in 3 bundles of say 7 strands each. If it is a single bundle i guess the diameter of the bundle may be too large and the effect of litz will be lost. The current may be travelling only on the circumference of the bundle and the wire strands that happen to be caught in the middle will be useless. In such a case only the wire on the perimeter is being loaded thus it is not thick enough to carry the current and getting very hot.

I would suggest that you either put the wires side by side or group them up in 6 bundles of 4 wires each. If the rating of the smps is 100% then the current carrying capacity of the wire should be calculated at 3.4A/mm². Without cooling cooling fan,

You will have a better result if you had to use an insulated copper sheet of 0.2mm thickness with the width of the bobbin. This will surely give you a better coupling and also better current handling.

thank you very much Silvio
wire is made of one bundle of 22 strands.but they are not parallel and they wrapped together so for example in one strand current sometimes is in center of bundle and some times in circumference so i think if strands were parallel you were right and center of bundle was useless but in this wrapped litz wire i think effect of litz is present.am i right?
i think another reason could be proximity effect at 20 amps and 50Khz.may it be cause of this extreme hot?
 

Silvio

Well-known member
thank you very much Silvio
wire is made of one bundle of 22 strands.but they are not parallel and they wrapped together so for example in one strand current sometimes is in center of bundle and sometimes in circumference so i think if strands were parallel you were right and center of bundle was useless but in this wrapped litz wire i think effect of litz is present.am i right?
i think another reason could be proximity effect at 20 amps and 50Khz.may it be cause of this extreme hot?

At the current of 20A these effects will surely come out and that is the reason of the extreme heat. I suggest you wrap 4 wires at a time and make 6 bundles. If you have a copper sheet it will be far better to wind the necessary turns with it. I do not know if you ever came across a winding made of copper sheet. The current handling in a wire all ends up to the cross sectional area of the wire or wires used

If the smps is going to be loaded at full power continuous then my advise regarding the current density must be observed otherwise you will still have a heating problem. One other thing that you did not consider when wrapping 22 strands together is that due to the wire is thick it will be rather awkward to work with and also the amount of coupling within the windings of the transformer will be poor. Using a copper sheet to make the secondary winding will surely get you better performance where it comes to coupling and current handling. Your next choice after the copper sheet is with multiple bundles. I am not sure how they will fit in the transformer and I guess it might be the case to opt for a larger core like an ETD49 so that it can accomodate the windings better.

I suggest you see the blog post so that you learn something about making litz wire. Download the pdf file for full explanation.

Link https://www.diysmps.com/forums/entry.php?145-Making-litz-wire-at-home

Regards Silvio
 

yoosefheidar

New member
At the current of 20A these effects will surely come out and that is the reason of the extreme heat. I suggest you wrap 4 wires at a time and make 6 bundles. If you have a copper sheet it will be far better to wind the necessary turns with it. I do not know if you ever came across a winding made of copper sheet. The current handling in a wire all ends up to the cross sectional area of the wire or wires used

If the smps is going to be loaded at full power continuous then my advise regarding the current density must be observed otherwise you will still have a heating problem. One other thing that you did not consider when wrapping 22 strands together is that due to the wire is thick it will be rather awkward to work with and also the amount of coupling within the windings of the transformer will be poor. Using a copper sheet to make the secondary winding will surely get you better performance where it comes to coupling and current handling. Your next choice after the copper sheet is with multiple bundles. I am not sure how they will fit in the transformer and I guess it might be the case to opt for a larger core like an ETD49 so that it can accomodate the windings better.

I suggest you see the blog post so that you learn something about making litz wire. Download the pdf file for full explanation.

Link https://www.diysmps.com/forums/entry.php?145-Making-litz-wire-at-home

Regards Silvio
thank you again
i going to use copper sheet instead of round wire for test
i think about another way:increase freqquency to 100 KHz and reduce primary turn to 12 and then reduce secondary turns to 4;is that good idea?

there is a little problem at primary side.
in schematic at primary side there is 47ohm resistor in series with a 470pf cap.at no load there is no problem but when smps is under load 47 ohm resistor become exteremly hot.i change it with a 5 watt resistor and still was hot.again change it with a 120ohm resistor and still same problem.
do you have any idea what's problem?
 
Last edited:

Silvio

Well-known member
In schematic at primary side there is 47ohm resistor in series with a 470pf cap.at no load there is no problem but when smps is under load 47 ohm resistor become exteremly hot

I think you are referring for the snubber across the primary winding. This is there to suppress any spikes generated during switch off of every cycle. Through my experience When a transformer does not have tight coupling between the windings it is most likely to generate spikes due to the leakage inductance within the transformer. This holds true for hard switched smps.

I suggest that you do not alter anything for the time being regarding the snubber components. It may work differently if you are going to rewind the transformer, it could be the case that coupling in the new transformer winding will be better and less heat is generated in the snubber resistor.

Regarding the frequency change and less turns. Well if you are using mosfets they will take the higher frequency better than IGBTs but still you will be increasing switching losses. The trafo will give you a little more power at the higher frequency. The primary wire has to be of a smaller gauge due to the higher frequency 0.4mm will be fine I guess. You have to make multiple strands for this gauge to carry the maximum current experienced during load. Do not forget to add another 20% with the current calculated as smps will be around 80% efficient and the extra current has to pass through the primary winding.
For the winding turns see that you wind appropriate turns according to the frequency of operation. As for the secondary turns keep the same ratio as the previous winding for the lower frequency.

Good luck Silvio
 
Last edited:

yoosefheidar

New member
I think you are referring for the snubber across the primary winding. This is there to suppress any spikes generated during switch off of every cycle. Through my experience When a transformer does not have tight coupling between the windings it is most likely to generate spikes due to the leakage inductance within the transformer. This holds true for hard switched smps.

I suggest that you do not alter anything for the time being regarding the snubber components. It may work differently if you are going to rewind the transformer, it could be the case that coupling in the new transformer winding will be better and less heat is generated in the snubber resistor.

Regarding the frequency change and less turns. Well if you are using mosfets they will take the higher frequency better than IGBTs but still you will be increasing switching losses. The trafo will give you a little more power at the higher frequency. The primary wire has to be of a smaller gauge due to the higher frequency 0.4mm will be fine I guess. You have to make multiple strands for this gauge to carry the maximum current experienced during load. Do not forget to add another 20% with the current calculated as smps will be around 80% efficient and the extra current has to pass through the primary winding.
For the winding turns see that you wind appropriate turns according to the frequency of operation. As for the secondary turns keep the same ratio as the previous winding for the lower frequency.

Good luck Silvio

it seems at first step i should replace secondary winding with copper sheet to see what happend.hope that solves snubber problem!
in case of increase freq you right switching losses goes up but still didn't produce much heat in heatsink.
 
Top