Ferrite Core & Power capability 1

[zacky]

Can a ferrite Core deal with a large power such as 20KVA
power transformer ?
If not, what is the maximum power rating can be acheived with using ferrite core?

 

[itsmoked]

I think they top out about 7kW...

 

[zacky]

itsmoked,Can you please specify some cores which can handle high power as high as 7 KVA

 

[itsmoked]

Sorry, but I can not.

You need to talk to the suppliers!
Call them and ask.

Epcos
FairRite
Ferroxcube
MMG
Magnetics
radiuspower
TDK

http://www.adamsmagnetic.com

is a rep for many of these companies.

These companies are all happy to help and are usually very helpful.

 

[CarlPugh]

For large custom ferrite cores try
Ceramic Magnetics, Inc.

http://www.cmi-ferrite.com

 

[ScottyUK]

Yes, absolutely a ferrite transformer can handle 20KW. You may have to build up a core using a number of 'U' sections to get adequate core cross-section to handle the required flux. I built a liquid-cooled transformer handling over 40KVA at 1kHz as part of my degree project, but kinda over-estimated the cooling requirements. Based on winding temperature rise it could have taken considerably more - I forget the figures now, but doubling the original design throughput would have been quite feasible. Air cooled it was good for over 10kVA.

The primary winding was square section enamelled wire; the secondary was two parallel 8mm copper microbore central heating pipes jacketed in heatshrink with an oil-air forced-draft radiator rejecting heat to atmosphere. Each 'U' section was approx 1" square section, 3C8 grade ferrite running at 300mT. The ferrites were from Philips Components, who are now Ferroxcube. The core structure was a shell type using eight cores held together in a frame constructed from aluminium angle and brass studding.

The easiest way to push more power is to raise the frequency. 3C8 (or maybe 3C85) wqs the most common power transformer ferrite and is good for 100kHz or so before the core losses become too significant. Ferroxcube list their U cores in pretty big sizes and in a 3C94 grade which is a better (less lossy) core than those available to me. Try their U93/76/30-3C94 which looks to be the modern equivalent of the cores I used.

You have stirred some memories of a project which was damned good fun to do, and memories of a lot of friends too - thanks!


----------------------------------

One day my ship will come in.
But with my luck, I'll be at the airport!

 

[itsmoked]

That's cheating... He said "a" ferrite core.

That's cool you can stack them "parallel them". Never thought of that.

What did you need a 40kVA high freq for? Sounds like a nice design BTW.

 

[ScottyUK]

Ha-ha, it was a 'core' made up of lots of bits!

It was an induction heating demonstrator rig using a CSI with a resonant load - I also made a cute little controller for the inverter to keep the switching frequency just above resonance to maximise power transfer into the load. The resonant frequency changes as the ferrous metal load reached the Curie point where its magnetic properties change. If the frequency gets too close to resonance the commutation fails because there is not enough time for the off-going thyristors to recover their blocking state before the on-going ones switch in.

The load coil had enough stray flux to disturb CRT monitors in the adjacent lab - I was asked to shut it down a few times so that the scheduled class could take place! Looking back it had a few safety issues: molten metal, live parts everywhere, a huge DC link inductor. Of course when you are that age, you believe you are immortal and safety takes a back seat to meeting the deadline for completion!


----------------------------------

One day my ship will come in.
But with my luck, I'll be at the airport!

 

[zacky]

For 'U' Core , is it better to have both the primary and secondary windings on one limb or to use one limb for the primary and the other limb for the secondary ?

 

[CarlPugh]

Both windings on one limb is usually better.
Primary and secondary windings on separate limbs are sometimes used when high leakage reactance is desired. With primary and secondary on different limbs, there will be high leakage flux. (There shouldn't be a magnetic path from the top of the core to the bottom of the core)

 

[zacky]

Could you please explain why sometimes high leakage reactance is desired ?
As far as I know , leakage inductace causes the primary to take more current to produce the same flux which is normaly not desired.

 

[itsmoked]

ScottyUK; Thanks for the description sounds like loads of fun! I remember the strange experiment/systems running in school lab basements at several schools sigh..

 

[Warpspeed]

Golly this stirs ancient memories for me as well. I also used the large Philips U cores (identical sizes are also available from TDK).

You can place four U cores together to make a shell type of transformer that looks (in concept) something like a transformer constructed of E and I laminations. Then you can stack these in multiples of four to easily get a core large enough to handle tens of kilowatts. The core part is easy.

The design and layout of the windings is usually far more of a challenge. Skin effect,leakage inductance, and capacitance, are going to be your biggest problem. Using thin foil windings at a hundred milliamps in a miniature transformer at high frequency is one thing. But having a hundred amps or more to cope with in a monster transformer at a similar frequency can be problematic.

One or more really large toroids may be worth considering as well at very high power levels. The advantage of having a very long winding length and fewer layers can simplify some things.

Another sneaky way is to stack many smaller toroids and have a long thin cylindrical transformer. That will enable very few turns to be used which also is sometimes an advantage. This works especially well at very high operating frequencies.

So very much depends on operating voltage, current, required turns ratio, and frequency. Power by itself is not really an important parameter. Other problems concerned with the winding design will usually stop you long before reaching core overheating with a really large design.

 

[ScottyUK]

Hi Warpspeed,

I looked briefly at toroids but couldn't get enough ferrite cross section in any sort of reasonable shape. I'd toyed with tape-wound amorphous iron too, but couldn't get hold of a big core in small quantity. I wonder if the long tube shape built up from multiple toroids would work well? At the time I reckoned that it added a lot of winding resistance which could be avoided by a shell-type design as you describe.

The tubular secondary winding I used was an intentional way of combining cooling and minimising skin effect problems. Slightly heavier walled tube would have been almost perfect.

If you are able to make a number of closely matched smaller transformers I guess you could parallel them to a low inductance bus. For the bus I'd suggest a laminated structure, or for really large currents an interleaved laminated design +/-/+/-, to keep your leakage reactance down.

Zacky,

Leakage reactance causes poor regulation in the transformer. In the classic transformer model with all parameters referred to the primary side, leakage reactance and winding resistance are shown in series with the primary winding. Magnetising reactance and core loss 'resistance' are shown as being in parallel with the primary winding.

High leakage reactance can have benefits. A couple of examples:

The inverter that the transformer design I built formed part of needed a commutating reactance to be added to the circuit. A little more leakage reactance from the transformer and that additional component could have been avoided, but since this was a one-off design it was easier to build a low-leakage tranformer and add a reactor rather than get too much leakage and have to worry about getting rid of it. With hindsight I could probably have introduced plastic shims into the core structure of the low leakage design to create tiny well-defined air gaps to give the desired additional reactance.

In power transmission and distribution systems, transformers are often designed with a higher leakage reactance than might otherwise be achieved in order to keep prospective fault currents on the secondary side within limits which switchgear can handle. If the transformers were built with the minimum possible reactance then a fault would cause colossal currents to flow, requiring massively oversized switchgear, oversized conductors, additional mechanical bracing of windings, and so on. It is far more economical to allow for slightly poorer regulation and compensate for it than it is to design everything to withstand very high fault levels.


----------------------------------

One day my ship will come in.
But with my luck, I'll be at the airport!

 

[Warpspeed]

Scotty,
The multiple toroid "tube" idea is an easy way to get fairly high inductance with very few turns. The magnetic path length can be kept very short giving a high inductance per pass through each toroid. As you say, the individual turn length will be longer, but far fewer total turns are required, and the voltage per turn can be surprisingly high doing it this way.


Using stacked multiple smaller transformers also has several advantages just as you suggest, but balancing the winding currents can be rather interesting for parallel connection. Windings in series can sometimes work particularly well for higher voltages.

I once built a series trigger transformer for the xenon flash tubes used in a high powered laser. It used twelve high permeability ferrite toroids arranged in a tight rectangle with three per side. Each toroid had it's own one turn primary (twelve primaries) each driven from its own capacitor discharge system running off 1Kv. Each secondary turn looped through all twelve toroids developed close to 12 KV per turn, and there were four secondary turns generating something like a 45kV pulse.

The secondary winding was actually welding cable because it also had to carry the 900 Amp main flash tube current from the pulse forming network and dc supply.

Only a small transformer, but the four turn secondary had to both generate a 45kV trigger voltage, and also carry 900 Amps, an interesting transformer design problem.

Lots of different ways to skin a cat. Designing high power high frequency transformers is definitely an art form and a particularly fascinating design subject to get into.

 

[zacky]

ScottyUK, as you explained about the benefit of the leakage inductance, is that mean for Battery charger transformer ,the high leakage inductance can be used to limit the charging current so it does't exceed the requied limit ?

 

[itsmoked]

Warpspeed what kind of laser, neodymium:yttrium-aluminum?

 

[Warpspeed]

It was an niobium yttrium aluminium garnet Nd:yag (ruby) laser, peak flash tube power was 1MW. I cannot say exactly what it was for, It was part of a government project.

 

[itsmoked]

Thanks for that info. I walked though Nova at LLNL where they had lots of those and a huge warehouse full of caps to drive them all, (16trillion watts). Almost insane in principal.

 

[Warpspeed]

I have read in the non classified annual reports about the Shiva and Nova projects at Livermore, mind boggling stuff indeed. The raw electrical power to charge the pulse forming networks is almost beyond belief.

Getting back to the topic of this thread. I was using those large ferrite U cores in multiple parallel flyback supplies, each in the multi kilowatt range. These supplies were to charge the flash tube pulse forming network, and each was run slightly out of phase with respect to all the others, to minimise total input ripple current. That was all twenty years ago, but it still stirs some fairly vivid memories.

The whole power unit averaged out at roughly 12Kw, so yes multiple ferrite cores can be used to couple fairly serious power.