SMPS Transformer Size Selection help

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi SMPS masters,

I am building a discontinuous mode resonant type half-bridge supply , directly connected to mains to power up an amplifier[2.5KW]. The power supply capacity is to be somewhere around 3KW continuous and Switching frequency is 100kHz[soft switching] using IGBTs drived by IR2110 gate driver -> SG3525 @ 48% modulation...No regulation at all....
I have currently Access to ETD and EE Ferrite cores. My problem is that for a peak power of 4-5KW[3KW continuous] which core size would be best to handle this load very comfortably.....also allowing the windings to be done whithout hazzels....

Secondly, there are no winding foils are available for winding, thus i am totally relying on copper wire winding...so according to me it decreases the leakage inductance of transformer which necessiates another inductor to be inserted in series with transformer and series capacitor in order to maintain resonance even at lighter loads..is it so.

Correct me if I am Wrong...

Thanks in advance,
K a n w a r
 
Some folks are aware that I have been playing around with a similar ZVS type circuit. The one thing that threw me in your description is that copper foil windings would seem to reduce leakage inductance instead of increasing it. Leakage inductance helps the resonance.

I have been seeming to find, so far, that on my test circuits, it is not as big a deal as I thought to maintain complete ZVS operation at light loads so long as gate turn-on time is not too fast, I guess. So that eased requirement may help you some.

I am not up to the task of helping you with narrowing down your transformer size. Maybe I am just lazy because I use trial and error or a guess to select one of my surplus cores for a project.
 
It's quite hard to determine the required transformer size for a given power output, particularly when the power output is not constant at all. Essentially, power output is limited by winding heating, and this is a compromise between how much copper may be fitted in the winding window and the winding resistance increase due to skin effect, that increases with wire diameter.

I suggest building a prototype and measuring transformer temperature increase when driving the actual load. Start with something like E55 or ETD59. Be prepared to use two or three smaller transformers with primaries in series and secondaries in paralell (sometimes this is a very convenient approach that makes winding task and PCB layout easier). Also, be prepared to consider low operating frequencies in the 35Khz range, this requires slightly bigger magnetics but provides a lot of advantages (like halving losses in the switches and no longer caring about skin effect nor litz wire).
 
I pretty much agree with Eva on the starting core size, though an E42/20 might work as well. If you are confining yourself to using magnet wire instead of foil, you will need room for a lot of wire. Core size will also depend on how much air flow is available for cooling.
At the power level You're trying to achieve, it would be best to go with a separate transformer and resonating inductor, rather than try to incorporate everything into a single piece of magnetics. As has been pointed out, you will need to design in a fairly substantial amount of leakage in order to get a single piece transformer to work with the resonant half bridge topology. Depending on how you achieve it, the leakage can cause flux focusing that will result in substantial heating in the windings, even if litz wire is used. As an example, I tried building a resonant 1/2 bridge of about 150-200W rating running at 100kHz using a split bobbin EER28L transformer. Even though the total power chain efficiency was around 87% (including both PFC and DC-DC converter), the copper next to the bobbin split was getting intolerably hot, even when very fine strand litz wire was used. I would imagine that the effect would be even more severe at the 2.5kW power level. Using a separate transformer and inductor cooled things down quite a bit, and the inductor was very small in physical size (I used an EI22 core).
A separate transformer and resonating inductor can be optimized for low loss. The transformer should be designed for lowest possible leakage. The inductor can be a gapped ferrite core. Powdered iron won't be suitable due to the high losses.
You will also need to pay close attention to the ripple current rating of the resonating capacitor. At 100kHz operating frequency, it will be fairly small in value, and it is hard to find a small value capacitor (read also small physical size) that can pass a lot of current. Wima FKP1 series capacitors would be a good start - you may have to parallel several smaller capacitors to get the rating you need.
 
So far I thought that resonant mode topologies required special control circuitry, but from your posts it seems that it can be done with an standard PWM controller such as SG3525.
Please could anyone help us understand a bit more about that topology, its advantages weak points and how to design it?
Thanks!
 
Hi everyone....

Thanks for the precious suggestions..

I have currently access to EE55 and EE100 cores, If i use EE 100 will there be any thing wrong with oversizing cores...
I will now use a seperate inductor for invoking resonance and lot of parallel low value high voltage caps in series to invoke the resonance...

Thanks EVA, I would also try to achieve resonance with 35KHZ as switching frequency in my seperate project and will compare both worlds......

The 600V 72A , IGBT i am using are capable of Hard switching at 40kHz, but at soft switching 100KHZ couldnot be a problem ....


regards,
K a n w a r
 
If you are trying 35kHz first, go for the larger core. I'm thinking you'll have more problems with copper loss than core loss, and it's nice to have some room. You might also want to be thinking about putting some tape margins in the transformer for safety purposes, so the extra space will come in handy for that, too. If you are not familiar with adding margins to get safety creepage distance in a transformer, you might want to try going to www.powerint.com and downloading application note AN-18. It'll explain some of what I'm talking about...
 
I have recently been using a variety of large E cores although your power requirement may require some very large cores such as D100s. Try visitting the EPCOS website and down-loading their free ferro-magnetic design software. It may be useful to you and will certainly give you an idea of through power in single ended and push-pull configuration.

Cheers,

Steve
 
i have built small car smps of 300w continous power(small in comparison to 3kw discussed here).
torroid worked great here.i have seen huge torroids being fitted in huge DJ amps at local elect. market here in DELHI.
torroids r costlier than ETD.still amp makers using them in DJ amps,so there must be some very good and solid reason(appart from obvious small size) behind there selection for use in those monster amps.
 
Low voltage DC-DC converters are a different sort of beast. Toroids work very well in situations (like low voltage DC-DC converters) where safety isolation requirements ae fairly relaxed. Also, with low input voltages, the number of turns on a toroid will be fairly small, so that they are economical to wind using a hook winder or by hand winding. For a line input SMPS, it is much more difficult to use a toroidal transformer due to the extra requirements for primary-secondary insulation and creepage distance. The number of turns is also larger to support the extra voltage. Toroids are not commonly used in these applications.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.