Making litz wire at home

Silvio

Well-known member
Here you will find my experiences when making litz wire at home based on information that I gathered through the net. I also examined and experimented myself with several types and these are my findings.

Cannot upload zipped file of 1.16MB If anyone wish to see this file I can post it through his mail address
 
Last edited:

Silvio

Well-known member
Thanks very much MicrosiM I will upload a few threads of general knowledge about smps in SMPS information. I guess this will be the right place to post them.
 

Silvio

Well-known member
Here you will find my experiences when making Litz wire at home based on information that I gathered through the net. I also examined and experimented myself with several types and these are my findings.
 

Attachments

  • Making litz wire at home.pdf
    1.2 MB · Views: 115
Excellent article Silvio, I discovered this table a long time ago but I am not sure if I apply it correctly. In the case that I use my smps at 50kHz ...... I could use 23AWG wire, but if I only have 18AWG (it supports 17kHz) I must do 3x18AWG (3x17kHz) ????. Or in that 3x18AWG arrangement it will only handle more current but will still work at a maximum frequency of 17kHz? Thanks and regards!
 

Silvio

Well-known member
18AWG is too thick to use in smps. The thickest wire to be used is 0.8mm as this goes down to 20Khz. Making a switching power supply working lower than this frequency is not practical as the core will not be utilized efficiently.
Due to skin effect at high frequency AC tend to travel around the circumference of the copper wire. Using thicker wire will be useless as the skin depth will be only to a certain amount according to the frequency used. The higher the frequency the thinner the wire should be. We then use multiple thin wires to overcome this problem, either side by side or twisted together.
 
18AWG is too thick to use in smps. The thickest wire to be used is 0.8mm as this goes down to 20Khz. Making a switching power supply working lower than this frequency is not practical as the core will not be utilized efficiently.
Due to skin effect at high frequency AC tend to travel around the circumference of the copper wire. Using thicker wire will be useless as the skin depth will be only to a certain amount according to the frequency used. The higher the frequency the thinner the wire should be. We then use multiple thin wires to overcome this problem, either side by side or twisted together.
I mean that (to see if I explain myself better):
If I have 25AWG I can use that wire up to a maximum working frequency of 85kHz (preferably not to reach the maximum) ...... assuming that my smps works at 75kHz ........ I could use 25AWG, but if not had that wire? If I had 23AWG? (which supports 53kHz) what would I do? Is there any possible combination of wires that can be made? The only thing I know is that as I put more wires in parallel, the more current it will support. But how often would it be done?
Thanks again! Each answer that he gives me is enriching for me and surely for those who read this post.
 

Silvio

Well-known member
As discussed already, if the copper wire is thicker then the skin depth for that particular frequency it will not be utilized fully and the center core will be invisible to the current. Using thicker wire than needed will occupy more space on the bobbin. 50 Khz is also a good switching frequency and anything between 20 and 80Khz is good. Ferrite cores tend to give higher power for a given core size as the frequency goes higher. Be careful but try not to exceed 100Khz as core losses tend to get higher. When at higher frequency the flux density needs to be lower due to losses encountered. I found that a frequency between 40 and 80Khz is the best range for Half and full bridge topologies.

Some rules of thumb for you regarding current density.
For an smps working at full current and at 100% duty cycle then the current density will be 3.2 amps per mm^2
For 80% duty then the current density will be 4A per mm^2
For audio purpose from 6 to 8A per mm^2 possibly with a cooling fan blowing at the transformer.

Example. If I am designing a transformer to work in an audio amplifier and my output current is to be 6A. The frequency of oscillation is 60Khz.
So for a frequency of 60Khz according to the table not more than 0.5mm copper wire is to be used.
So now we calculate the cross sectional area of a single wire Area of a circle = Pie x Radius squared so 3.142 x 0.25 x0.25 = 0.196 mm^2
This diameter will give us a current of 0.196 x 6A = 1.17A for each 0.5mm wire used
Now we divide 6A / 1.17 =5.12 wires so we cannot make a partial wire and we have to round off either for 5 or 6 wires. If there is enough space in the transformer bobbin then use 6 wires either side by side or twisted together depending on the fit. We try to cover the length of the bobbin as much as possible leaving enough margin space at the ends. You can also opt for 2 bundles of 3 wires twisted together which will also make 6 wires. You have to see what fits best according to the number of turns needed.
With six wires the current density lowers to about 5A per mm^2

I hope my explanation is clear enough
 
Last edited:
Hello Silvio! How are you? I am making a DC-DC converter from 12Vdc to +-35Vdc for car audio.
It turns out that I have tried to use the ATX EE33 transformer without rewinding, they give me 90Vdc at no load and then with a load of 36 to 40 ohms the voltage drops to 46Vdc consuming 1.3A so I want to try rewinding the transformer.

I have 18, 21 and 25AWG enameled wire. I was thinking of using the 21AWG (0.7mm) one since I don't often use it and working at 30khz it will be fine.

I attach a screenshot of the calculation that the software produced.
DCDC EI33 PRUEBA BOBINADO.jpg
I want to understand how I should do my calculations to know how to accommodate the sets of wires and that they fit well.

According to the soft for the primary I require 11 wires of 0.71mm, I was thinking of braiding 2 groups of 6 wires leading them in parallel. As the internal height of my reel is 17mm, I think that half of the primary would fit in the first layer.
Is there any calculation to know what diameter a group of "x" number of 0.7mm cables will end up with? I had to do a bit and as a result it gives approx 2mm, that's why I know that only half of the primary would enter in 1 layer.

For the secondary it tells me 21+21 turns with double wire.

Thank you! I hope you can help me!
 

2pist

New member
Hello Silvio! How are you? I am making a DC-DC converter from 12Vdc to +-35Vdc for car audio.
It turns out that I have tried to use the ATX EE33 transformer without rewinding, they give me 90Vdc at no load and then with a load of 36 to 40 ohms the voltage drops to 46Vdc consuming 1.3A so I want to try rewinding the transformer.

I have 18, 21 and 25AWG enameled wire. I was thinking of using the 21AWG (0.7mm) one since I don't often use it and working at 30khz it will be fine.

I attach a screenshot of the calculation that the software produced.
View attachment 7417
I want to understand how I should do my calculations to know how to accommodate the sets of wires and that they fit well.

According to the soft for the primary I require 11 wires of 0.71mm, I was thinking of braiding 2 groups of 6 wires leading them in parallel. As the internal height of my reel is 17mm, I think that half of the primary would fit in the first layer.
Is there any calculation to know what diameter a group of "x" number of 0.7mm cables will end up with? I had to do a bit and as a result it gives approx 2mm, that's why I know that only half of the primary would enter in 1 layer.

For the secondary it tells me 21+21 turns with double wire.

Thank you! I hope you can help me!
Greetings! is the software program you are using available online? do you have a link you can share? thank you!
 

Silvio

Well-known member
Hello Silvio! How are you? I am making a DC-DC converter from 12Vdc to +-35Vdc for car audio.
It turns out that I have tried to use the ATX EE33 transformer without rewinding, they give me 90Vdc at no load and then with a load of 36 to 40 ohms the voltage drops to 46Vdc consuming 1.3A so I want to try rewinding the transformer.

I have 18, 21 and 25AWG enameled wire. I was thinking of using the 21AWG (0.7mm) one since I don't often use it and working at 30khz it will be fine.

I attach a screenshot of the calculation that the software produced.
View attachment 7417
I want to understand how I should do my calculations to know how to accommodate the sets of wires and that they fit well.

According to the soft for the primary I require 11 wires of 0.71mm, I was thinking of braiding 2 groups of 6 wires leading them in parallel. As the internal height of my reel is 17mm, I think that half of the primary would fit in the first layer.
Is there any calculation to know what diameter a group of "x" number of 0.7mm cables will end up with? I had to do a bit and as a result it gives approx 2mm, that's why I know that only half of the primary would enter in 1 layer.

For the secondary it tells me 21+21 turns with double wire.

Thank you! I hope you can help me!
 

Silvio

Well-known member
Well I also tried my software Excellent IT 7300 at this but as you can see the rated transformer power is far less the the desired output power. The trafo rated power is 175w while your desired output power is 280w!! The software is telling me that the winding fill factor is also larger then 0.3. It means the windings will not fit.
You need to select a larger core otherwise it will not work for that power.
I tried several different cores and EE42 or ETD 44 seems the best cores to choose for this power.
Considering different approach is to use copper sheet 0.2mm thick for the primary winding. This will greatly enhance the fitting in the bobbin and will surely carry the current needed in the primary if say it is 25mm wide. The cross sectional area for such a sheet 25mm wide x 0.2mm thick will have a current carrying capacity of around 30A at a current density of 6A/mm^2. This way you can use a smaller core which the windings will then fit. Also a higher frequency can be selected such as 50 to 80 Khz thus making the winding turns less.
Below some calculations for you with different cores.

I hope that helps
 

Attachments

  • ivanelectric.pdf
    464 KB · Views: 23
Thank you very much Sylvio! The IC I use is TL494, not IR2153. I have tried with an atx transformer without rewinding and 40khz with a nice square signal at the AC output. After rectifying I connected two amps with TDA7294 (one per channel). The only thing is that the output is not stabilized, it varies as the load increases.
I'll see how I keep playing
 

2pist

New member
18AWG is too thick to use in smps. The thickest wire to be used is 0.8mm as this goes down to 20Khz. Making a switching power supply working lower than this frequency is not practical as the core will not be utilized efficiently.
Due to skin effect at high frequency AC tend to travel around the circumference of the copper wire. Using thicker wire will be useless as the skin depth will be only to a certain amount according to the frequency used. The higher the frequency the thinner the wire should be. We then use multiple thin wires to overcome this problem, either side by side or twisted together.
OH MY!! I was just going to wind my first transformer. I was going to use 18 ga on the primary, 20 ga on the secondary and 25 ga for the 15 volt.. I read this just in time!
 
Top