H-Bridge transformer design help

[FlapJack]

Thanks E-Design, this is starting to make sense. I will wait for next results and post any final questions then.

 

[usama14]

Thanks E-Design, this is starting to make sense. I will wait for next results and post any final questions then.

FlapJack. Waiting for your reply about the method I proposed.

 

[FlapJack]

E-Design, this is important you should use the same winding width as the OP has.

ferroxcube = 41.2mm total winding width of bobbin.

Also if you have time can you run two designs, one with foil primary and one with primary made up of magnet wire.

- - - Updated - - -

usama14, i am worried about skin depth with your method. I have time to think about it now.

 

[Warpspeed]

Skin depth at 100 Khz is only 0.2mm.
https://chemandy.com/calculators/skin-effect-calculator.htm

Thickest wire or foil should only be 0.4mm max.
Anything thicker just takes up space and does nothing.

 

[usama14]

Skin depth at 100 Khz is only 0.2mm.
https://chemandy.com/calculators/skin-effect-calculator.htm

Thickest wire or foil should only be 0.4mm max.
Anything thicker just takes up space and does nothing.

This makes me use SWG 28 for both primary and the secondary according to this I mean I will have to use 7 parallel wires for each turn of primary and 2 parallel for each turn of secondary.?

- - - Updated - - -

E-Design, this is important you should use the same winding width as the OP has.

ferroxcube = 41.2mm total winding width of bobbin.

Also if you have time can you run two designs, one with foil primary and one with primary made up of magnet wire.

- - - Updated - - -

usama14, i am worried about skin depth with your method. I have time to think about it now.

Sure..I am gonna wind it today though. Just to check it. Will share the results with you people.

 

[FlapJack]

Yesterday we did not take in account efficiency. Just picking a number of 80% we get.

1kw/.8 = 1250W
1250w / 39vdc = 32A

Now the skin effect will be even worse, we need even thicker wire. The rule of thumb is start with 500 circular mils per amp.

500 Cmils/A x 32A = 16,000 Circular mils.

Looking at the chart i gave you in post 31 that is 8 awg wire. This project is going from bad to worse.

I think warpspeed nailed it when he said use copper foil, but even the .008" foil is a little light, .010" may be needed.

 

[Warpspeed]

Ten thou foil should be fine.

 

[usama14]

Yesterday we did not take in account efficiency. Just picking a number of 80% we get.

1kw/.8 = 1250W
1250w / 39vdc = 32A

Now the skin effect will be even worse, we need even thicker wire. The rule of thumb is start with 500 circular mils per amp.

500 Cmils/A x 32A = 16,000 Circular mils.

Looking at the chart i gave you in post 31 that is 8 awg wire. This project is going from bad to worse.

I think warpspeed nailed it when he said use copper foil, but even the .008" foil is a little light, .010" may be needed.

Actually I wont be giving the transformer a 1000W input. I will be using it for just a system of 600W. 1000W is just to give extra wattage capability. So I dont think that much current will be flowing through the system at 600W.

- - - Updated - - -

Ten thou foil should be fine.

Warpspeed I dont have that wires right now. Only SWG gauges are available. Thats why I asked you about my method.

 

[E-design]

When there is a final decision on the input parameters, I will run it through the program. By not using copper foil for the primary will increase the copper losses substantially.

 

[Warpspeed]

The "bundle of wires" primary winding method does work, but it is very inefficient of space on the bobbin. It produces an ugly lumpy winding which is not very nice to place another winding on top of.

The trick to winding (designing) these transformers is to use the very limited winding space available on the bobbin as efficiently as possible with the maximum amount of copper, bearing in mind insulation and skin effect requirements.

I might suggest you go ahead as intended, get it up and running, and do some testing and measuring, design a suitable choke, and test the feedback loop for stability and response. There is still plenty of work ahead.

At that point, you may decide to slightly adjust the number of secondary turns.
That would also offer a good opportunity to rewind the primary with foil.

 

[usama14]

I made the transformer yesterday according to this mechanism:
a) 18 turns of 2 parallel SWG28 wires for the secondary
b) Then on the top of it I gave 4 turns with the 7 parallel wires of SWG28 for the primary.
c) Finally I gave the 18 turns of the 2 parallel wires for the secondary.
I got the following results:
Lp=Primary inductance=68uH
Ls=Secondary inductance=2.6mH
Now I am falling short of the required inductance. Now kindly help me to know whether such values of the inductances will work or not? :roll:
The theoretical values were:
Lp=Primary inductance=95uH
Ls=Secondary inductance=7.7mH

 

[FvM]

Either different number of windings or incorrect measurement.

 

[usama14]

Either different number of windings or incorrect measurement.

No. I am positive that I gave correct number of turns :/

- - - Updated - - -

The "bundle of wires" primary winding method does work, but it is very inefficient of space on the bobbin. It produces an ugly lumpy winding which is not very nice to place another winding on top of.

The trick to winding (designing) these transformers is to use the very limited winding space available on the bobbin as efficiently as possible with the maximum amount of copper, bearing in mind insulation and skin effect requirements.

I might suggest you go ahead as intended, get it up and running, and do some testing and measuring, design a suitable choke, and test the feedback loop for stability and response. There is still plenty of work ahead.

At that point, you may decide to slightly adjust the number of secondary turns.
That would also offer a good opportunity to rewind the primary with foil.

I did the experiment..but getting pretty low inductance values

 

[FvM]

No. I am positive that I gave correct number of turns

O.K., how did you measure the inductance?

 

[usama14]

O.K., how did you measure the inductance?

With the inductance meter.

 

[D.A.(Tony)Stewart]

If your E cores are damaged or contaminated surfaces, then your inductance will be low.

YOu can try burnish lapping on glass in an elliptical pattern with very fine compound or very very fine wet-dry sand paper with something to hold it co-planar at all times.

Surface roughness in microinches.....

Then clean and clamp properly.

Did you measure near 100kHz? and SRF?

 

[Warpspeed]

What material did you end up using, 3C90, 3C94, or 3C95?

 

[usama14]

What material did you end up using, 3C90, 3C94, or 3C95?

3C90 and it DID actually break from two legs when it was shipped from USA. I had to fix it with the help of a strong binding liquid. Maybe that's the cause of low values?

- - - Updated - - -

If your E cores are damaged or contaminated surfaces, then your inductance will be low.

YOu can try burnish lapping on glass in an elliptical pattern with very fine compound or very very fine wet-dry sand paper with something to hold it co-planar at all times.

Surface roughness in microinches.....

Then clean and clamp properly.

Did you measure near 100kHz? and SRF?

Yes Sir. It did actually break from the legs when it was delivered from USA. I used a strong binding liquid to fix it. Is that the reason for such error?
Also Enlighten me about "measure near 100KHz and SRF" kindly?

 

[Warpspeed]

I would have to say yes.
The small extra introduced air gap will make a significant difference to the inductance.

But in this application it will not matter.
Go ahead with testing your prototype.

 

[usama14]

I would have to say yes.
The small extra introduced air gap will make a significant difference to the inductance.

But in this application it will not matter.
Go ahead with testing your prototype.

Sure I will give it a try and share results.