Hi,

Has anyone got any experience with designing a resonant inductor for 800V input phase shifted full bridge operating at 120KHz at 1kW output.
I need 50uH and the volt-sec is around 250Volt-usec. I have tried EFD 25x13x9 with 20 turns but the core is running at 85 Deg. C with no load condition.

 

This paper should help... it gives a detailed theoretical approach and a practical step-by-step process, starting with selecting the right core for the job.

https://ieeexplore.ieee.org/document/9648941

 

What are you using for your main transformer? A lot of the resonant supplies utilise the leakage inductance of the main transformer to produce the resonant inductor, by winding the main transformer on a two-section bobbin.

By the way, @Irving ,I doubt that very many of us on this forum have access to IEEE documents!

 

This paper should help... it gives a detailed theoretical approach and a practical step-by-step process, starting with selecting the right core for the job.

https://ieeexplore.ieee.org/document/9648941

Thanks. This is very useful. I have followed the same procedure for calculation. However, I think I am making a mistake in estimating flux density when I am considering peak current.

I have current moving from +2A to -2A. Should I be using 4A as peak current for calculating flux density (B).

 

What are you using for your main transformer? A lot of the resonant supplies utilise the leakage inductance of the main transformer to produce the resonant inductor, by winding the main transformer on a two-section bobbin.

By the way, @Irving ,I doubt that very many of us on this forum have access to IEEE documents!

Thanks Ian. To achieve ZVS at such high input voltage, I don't have enough leakage in the transformer. Also, I need to be able to consistent with resonant inductance so I have to use a 50uH

 

By the way, @Irving ,I doubt that very many of us on this forum have access to IEEE documents!

Hi @Ian0 true, but you can often find them elsewhere, not behind a paywall. I prefer to give the source initially. From his posts he sounds like a professional and may well have access. (it seems he does!) Worst case, someone could email him a copy.

Thanks. This is very useful. I have followed the same procedure for calculation. However, I think I am making a mistake in estimating flux density when I am considering peak current.

I have current moving from +2A to -2A. Should I be using 4A as peak current for calculating flux density (B).

You're welcome. My gut feel is no, since the flux is proportional to current and flux density is flux/core area. All the sign says is which way it flows in the core. However, if you show your working we may be able to advise better..

 

Hi @Ian0 true, but you can often find them elsewhere, not behind a paywall. I prefer to give the source initially. From his posts he sounds like a professional and may well have access. (it seems he does!) Worst case, someone could email him a copy.


You're welcome. My gut feel is no, since the flux is proportional to current and flux density is flux/core area. All the sign says is which way it flows in the core. However, if you show your working we may be able to advise better..

I will try to send over the current measurement through the inductor to give you a better idea. However, I have done the volt second measurement which is 258V-usec.

 

I will try to send over the current measurement through the inductor to give you a better idea. However, I have done the volt second measurement which is 258V-usec.

I have number of turns at 18V and gap of 0.6mm. core is EFD 25X13X9 N87.

 

I have number of turns at 18V and gap of 0.6mm. core is EFD 25X13X9 N87.

Sorry I meant number of turns 18.

 

Have you tried running the numbers on your core through their procedure? From your original post it sounds like your core losses are very high, therefore your core is too small???

 

I tried the procedure but the flux density is well below 0.15Twhen I use 3A peak current. However, in reality looking at the temperature it seems it's well over 0.2T

 

The paper does talk about optimization using FEM which I haven't done yet. It seems the analytical method provides a good starting point but I will have to tweak the turns and gap to control the core losses.

 

Hi @Ian0 true, but you can often find them elsewhere, not behind a paywall. I prefer to give the source initially. From his posts he sounds like a professional and may well have access. (it seems he does!) Worst case, someone could email him a copy.


You're welcome. My gut feel is no, since the flux is proportional to current and flux density is flux/core area. All the sign says is which way it flows in the core. However, if you show your working we may be able to advise better..

Core loss is proportional to flux excursion, so 4A is the figure to use. ±2A will give about 5.5 times the core loss as from 0 to 2A.
Perhaps redesign for a lower flux excursion,

 

Perhaps redesign for a lower flux excursion,

With out doing math. If the core is hot with no current, then probably the VT is too much for the core.
What frequency?
When I started, we ran at 20khz, we hit saturation long before ac core loss was a problem. By the time I was running 200khz I never worried much about saturation because the ac losses were so high. I am saying the core could not tolerate going from 0 to saturation at 200khz. Now at mhz, the ac ripple in the core is small compared to saturation point.

At this point, where you have ac losses and no dc loss, the core should be at 1/2 of the max temp you will be running. Then the load current will add that much more heat.

What is the core inside temperature? Or outside temp at the hottest point. 50C is not much temp.

 

VT is 250V-usec and frequency is 120KHz.

 

Temperature is 85 Dec C on the core when there is +2A to -2A current through the inductor. It's a resonant inductor so it would always see AC current only.

 

Look at the datasheet for the core, and see how it defines core loss. There usually is a graph. It is really important to find out whether it is defined a peak-to-peak flux excursion, or ΔB being the variation of flux either side of the steady state. It makes a factor of 5.5 difference if yiu get the wrong one.

 

I think N87 isn’t suitable as an inductor. It’s transformer core material, not able to accumulate lot of energy. It’s most probably saturated.

 

I think N87 isn’t suitable as an inductor. It’s transformer core material, not able to accumulate lot of energy. It’s most probably saturated.

Hi..I have tried N95 but kind of similar results.

 

Use the N87. Any transformer material has a high saturation flux and low loss. It will need the appropriate gap.
Definitely, don‘t try iron powder, although it will stand a very high flux without saturating, its core Losses with an alternating flux are too high. It’s very good for Dc output inductors but little else. An exception might by Micrometals #2 material, but its permeance is rather too low