New guy here(first SMPS build)

[twenglish1]

Hello everyone, i am starting a project to build a high power switchmode supply that will act as the power source for a homemade plasma cutter, i don't want to hear start with something easier and lower power since i am new to this, i have made the decision that the best way to learn is to just jump right in and just do it, i also understand all the necessary safety precautions. This power supply i plan to build will have 330v open circuit voltage and need to be limited at 40 amps short circuit current. I am still in the research and design phase. I have decided to use a half bridge inverter design, and i will be driving it with an IR2153. So far i have purchased a lot of EE80 ferrite cores, the mains rectifiers, large filter capacitors, the IR2153 chips, and a lot of large 1200v, 150amp IGBT half bridge modules, i will post my schematic shortly for all to see. i dont have any specific questions as of now, just posting to see if anyone has any input or pointers that could help me along with this project

 

[MicrosiM]

Hello everyone, i am starting a project to build a high power switchmode supply that will act as the power source for a homemade plasma cutter, i don't want to hear start with something easier and lower power since i am new to this, i have made the decision that the best way to learn is to just jump right in and just do it, i also understand all the necessary safety precautions. This power supply i plan to build will have 330v open circuit voltage and need to be limited at 40 amps short circuit current. I am still in the research and design phase. I have decided to use a half bridge inverter design, and i will be driving it with an IR2153. So far i have purchased a lot of EE80 ferrite cores, the mains rectifiers, large filter capacitors, the IR2153 chips, and a lot of large 1200v, 150amp IGBT half bridge modules, i will post my schematic shortly for all to see. i dont have any specific questions as of now, just posting to see if anyone has any input or pointers that could help me along with this project

I like the EE80 for this project, but I recommend to replace the IR2153 with SG3525 as at least you have SS. witch IR2153 dont have.

Also a full bridge SMPS would be better option for 330Vx40 AMPS SMPS 13200W

I am not sure if you take into account harmonics will be injected and EMI

Hope that helps.

 

[twenglish1]

How much power can the EE80 cores handle at say 40khz? If one isn't enough I could parallel two cores, right? And what is the advantange of the sg3525 over the ir2153? One more quick question is what is the best way to determine the value of the capacitors in the half bridge? Not talking about the main filter capacitors, I'm talking about the half bridge output capacitors that are connected between the positive and negative rails and the point between them goes to one side of the primary winding, not sure what they are called

 

[wally7856]

The SG3525 has “soft start”, it is a real SMPS controller chip with a lot of design examples and good ap notes.

The EE80 core can deliver a theoretical power of 8KW at 40 khz, if you can find room to wind that much copper.

You can stack cores. Each core is a set of E cores. One set is 8KW at 40Khz. Two sets would give you 16KW. Realistically though, over 3 or 4 KW you usually go to multiple transformers and series or parallel the outputs to get the power you want.

You gave the short circuit current of 40 amps. What is the maximum working current and voltage you will see.

As a general rule a half bridge is good for 1KW. But it can be pushed to 2KW or so if you really want to do it. Also If you want to do KW’s of power then a half bridge is a poor choice. As Microsim said go full bridge.

 

[KX36]

i don't want to hear start with something easier and lower power since i am new to this, i have made the decision that the best way to learn is to just jump right in and just do it

Much like the multi-kW TIG welder project elsewhere on this forum I admire your guts and naive determination. You know people with some experience in the field will tell you it's an unthinkably difficult project nigh-on-impossible for an amateur especially a beginner, a couple of orders of magnitude more power than might be reasonable, but you continue anyway. Cudos.

I can't really give you much advice except to say be very careful, don't kill yourself or burn down your house etc. Don't be too disappointed if the project turns out to be more expensive and poorer performing than the commerically available alternatives or you never get it working at all as you know going into in that these are all distinct possibilities. And finally, good luck!

 

[twenglish1]

I will probably start at less than 40 amps for a start, maybe 10 - 15 amps Say i design my transformer to draw 15 amps at the rectified line voltage which is around 330v, that is just under 5000 watts, less turns on the primary means a higher current draw correct, but also a higher saturation level, i calculated around 11 primary turns will be a good starting point for one EE80 core, 11 turns puts the saturation level at about 4500 gauss, which is getting close to the max saturation of the core, so for a start should i go with more than 11 turns?

Question about doing a full bridge, will i be able to drive a full bridge with one of the SG3525 drivers? i know the chip only has two outputs but could i connect the gates together of the two opposite transistors and just use the two drive signals? or will i need 4 separate drive signals?

 

[wally7856]

“Will i be able to drive a full bridge with one of the SG3525 drivers?”

With a full bridge you usually use a gate drive transformer with separate output windings for all 4 gates. The gate drive transformer is driven by auxiliary transistors and the gates are assisted with more circuity.

Here is a good example showing the whole setup, but not necessarily the only way to do it.

EDIT, i do not know why this forum is making the image so small but i will try to upload again in next post.

 

[wally7856]

2nd try, converted PNG to JPEG.

 

[wally7856]

"I will probably start at less than 40 amps for a start, maybe 10 - 15 amps Say i design my transformer to draw 15 amps at the rectified line voltage which is around 330v, that is just under 5000 watts, less turns on the primary means a higher current draw correct, but also a higher saturation level, i calculated around 11 primary turns will be a good starting point for one EE80 core, 11 turns puts the saturation level at about 4500 gauss, which is getting close to the max saturation of the core, so for a start should i go with more than 11 turns?"


My semi educated input would be as follows. Pressman’s switching power supply book says not to use more than 1600 Gauss unless you really know what you are doing. He also says large cores have a problem dissipating heat. His book does not cover big cores like the EE80 so i would go less than 1600 Gauss, more like 1200 to 1400 maximum at 40khz.

Also using your 5000W as a guide, i would say the EE80 might be good for 3KW continuous duty. So use two of these cores with the primaries in parallel and put the secondaries in series, one winding for each core. You can then wire up the secondary’s with a full wave center tap configuration.

Now each primary only has to see 1/2 of the current, instead of 2x the current for a half bridge. Also with a full wave center tap you wind each output winding for the full voltage you want out but each winding supplies half the current.

To sum things up, you would have on each core one primary winding and one secondary winding.

 

[wally7856]

Thinking some more, it would be better to use a split primary. This is how the best transformers are wound.

½ Primary, Secondary, ½ Primary

This will give you the best coupling and give you more room for your primary winding.

340vdc Primary voltage
40khz
1200 Gauss
Ae 3.925 cm^2
= 45 turns

So you have a choice of 22 or 23 turns for each half winding.

 

[wally7856]

What ferrite material is your core made of.

 

[twenglish1]

I worked on an inverter tig welder and it used the same EE80 core, running at 40khz, using 11 primary turns, that would put the flux density about 4600 gauss, how are they able to do this? It also used a half bridge igbt module. That schematic of the full bridge driver will help out a lot, I figured a gate drive transformer would be the way to do it, I also like the idea of using two separate power transformers to spread out the load.

 

[wally7856]

BTW, I picked 1200 Gauss for my guess at 100% duty cycle with convection cooling. If you go with less duty cycle or use a fan you can go higher. How much higher? i do not really know. Best guess 1600 to 2400 Gauss max, if it is a modern ferrite. If not then 2200 Gauss max.

 

[twenglish1]

Does a higher gauss level mean more power transfer? I never could find that information

 

[wally7856]

The higher the flux level the fewer number of turns and the thicker the wire can be or more strands. So sort of, but the higher Gauss itself is not giving you more power.

 

[wally7856]

"I worked on an inverter tig welder and it used the same EE80 core, running at 40khz, using 11 primary turns, that would put the flux density about 4600 gauss, how are they able to do this?"

Did you take any pictures of how the transformer was constructed? Maybe it was really a split primary and there were 22 turns. Or how did you measure the 40khz, maybe you measured in the wrong spot and it was really 80khz?

 

[wally7856]

What did that tig welder use for windings. Wire, litz wire, or copper foil.

 

[KX36]

I would expect to use copper foil on high current windings. The core material makes a big difference to the transformer. At lower frequencies ferrite might not be the best and you'd have to assess that as at this power level small things can make a big difference to how much heat you have to deal with. other materials may have higher saturation flux densities. The power ferries I've used saturate around 0.3T and I don't specify any peak to peak flux density of more than 0.2T. (be careful as sometimes things are quotes in Bpk, sometimes Bp_p.) Big cores dissipate heat less efficiently as their surface area to volume ratio is less, so the power they can handle doesn't increase linearly with size. you will likely have to interleaved multiple transformers or multiple power supplies to get there.

Are plasma cutters normally constant current or constant voltage? Will you have regulation on the output or have it free running on a fixed duty cycle? Will you have a PFC (probably need to)? how will you limit current (a boost PFC will likely need its own complex current limiter to avoid inductor saturation)? What is your power source?

 

[wally7856]

"I worked on an inverter tig welder and it used the same EE80 core, running at 40khz, using 11 primary turns, that would put the flux density about 4600 gauss, how are they able to do this?"

Sorry, first day on the job. With a full bridge the primary see’s 340 vdc and with a half bridge the primary only see’s 170 vdc. That is why the full bridge is recommended for high power applications because for half bridge you must have double the current for the same output power as the full bridge.

This also means that when calculating turns or flux level you use 170 vdc for half bridge.

B = 170,000,000,000 / 4, 40,000 hz, 11 N pri, 3.925 cm^2 = 2,460 Gauss

And luckily it coincides with my 2400 Gauss maximum recommendation i gave before so i do not look to stupid.

With that said, this Gauss level was for a TIG welder. Probably a 20% duty cycle, with a good fan on it.

But for more duty cycle and still probably needing a fan look at KX36’s post of using a maximum of 2000 Gauss (.2T).

 

[twenglish1]

The tig welder used 11 turns of copper foil on the primary, I measured the 40khz at one of the gate/emitter pairs, is this correct? And this tig didn't use a voltage divider, one side of the primary went to the output of the half bridge module, and the other side of the primary went through two large film caps then to the negative rail