toroidal core for high current

ml

Surely you meant

l ?

The above page says "division by zero" and doesn't have active fields. = In fact, it requires POSTDATA to function.

Awfully large ratio. Why not forward converter?

50MHz?

About five inches of microstrip. Sounds reasonable.

Give this a try:

I prefer half bridge PP.

Tim

--=20 Deep Friar: a very philosophical monk. Website:

Either one of those topologies isn't the best choice. A flyback at

2.5kW would be ridiculous.

As Tim suggested you might want to look at a topology that uses a transformer. Half or full bridge. You also will probably need PFC on the front end if you plan on getting that power out of a European residential line.You will be needing capacitor's with an "S" for handling the large rms currents and large beefy semi's with an "S" and brick heat sinks fans etc.

You aren't going to be banging around FET 'S or IGBT's that you are going to be needing at 50 MHz and no controller I know of operates at

50MHz.

For such a large output current you may also want to look into interleaving or dual/triple phase controllers.

;-)

I would expect that at 100 amps you're going to start getting into issues not just of core materials, but of wire heating and possibly even mechanical vibration induced by magnetic effects.

I don't think so. An air core inductor would either have so much parasitic capacitance that your switching losses would go through the roof, or it would radiate enough to seriously impact your efficiency (and to interest the FCC).

But the numbers are absurd.

See if you can find the "Amidon Associates" website -- they used to peddle a book that went into that, and they had a core saturation calculator. It can be done, and the better ferrite manufacturers publish the data.

The core also heats directly, from both conduction losses (if it's one of the really low frequency ferrites) and from hysteresis losses.

Now you ask!!

Have you looked in the DigiKey catalog? I don't think there are inductors going up that high, but you can obtain semi-custom magnetics if you can find the names of the right transformer companies.

Suspect websites* and ridiculous numbers (50MHz?) aside, you've just asked a question akin to "I'm going to make potato salad for a party this weekend -- how big of a bowl do I need?".

The amount of energy that the core stores each cycle depends inversely on the frequency of operation of the circuit. The losses in the core are more complex, but also depend both directly and as the square of the inverse of the frequency.

In general, as the circuit frequencies get higher the potential core losses go down**, but the switching losses in the transistors go up. The art of switching supply design for as long as I've been around to watch has advanced by finding transistors and circuits that allow faster and faster switching speeds, allowing the magnetics (and thus the whole circuit) to get ever smaller.

But 50MHz is a bit high for a realistic circuit.

• _Any_ website that purports to design a switcher for you is suspect -- even the ones by the semiconductor companies.

** Depending on the core material.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

snipped-for-privacy@q22g2000yqm.googlegroups.com...

tor2...

-Calculator....

=A0In fact, it requires POSTDATA to function.

That should be 50khz and not 50Mhz. L =3D 450uH... still seems relatively small.

Guys, it should be obvious it was 50khz and not 50Mhz, I made a mistake when I was entering the data to paste it in.

Volts In 250 V Volts Out 25 V Load Current 100 A Freq. 50 KHz Vripple 0.25 V Duty Cycle 10 % Ipp Inductor 1 A Ipk Inductor 100.5 A Irms 99.500418759588 A L 450 uH C 800 uF

And look! The inductor value has changed!

Regardless, going from 250V to 25V means that the output stage will only be on for 10% of the time, which is putting severe demands on an already heavily used inductor. Why don't you want to use a transformer?

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

duh! Did I say they wouldn't? Just curious... have you ever made a mistake in your life? I made a mistake entering the data... get over it. It's not the end of the world.

I thought the whole point of smps was reduce the transformer size?

I wasn't criticizing; forgive me if it came across that way.

And I do make mistakes -- sometimes I think I've screwed up when I haven't :-).

When I and others say "why don't you use a transformer", we mean "why don't you use a switching regulator topology with a transformer". If you use a transformer with a turns ratio sufficiently less than 10:1 to insure 5V out when the input is at it's lowest ebb then you will minimize the amount of energy that must be stored in the magnetics, and hence their size.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

messagenews: snipped-for-privacy@q22g2000yqm.googlegroups.com...

Legg is right. Take a look at 100 amp DC "inverter" welding power unit. In fact, that might be a good source of parts. You would need to convert it from constant current to constant voltage. The Miller units have good technical support and documentation.

Tm

--- news://freenews.netfront.net/ - complaints: snipped-for-privacy@netfront.net ---

messagenews: snipped-for-privacy@q22g2000yqm.googlegroups.com...

It is four orders of magnitude greater than your previous effort, and your main source of error, here.

1% ripple in a single stage filter may be impractical. A 25V filter inductor will exhibit signifigant core loss and a 100A winding on same will not be simple to achieve.

If the solution is as 'small' as you think, you'll be a lucky guy.

RL

hehe, ok. np.

ly

dy

Ok, I don't have a huge problem with it but it seems that it may add significantly to the cost. I'd prefer the isolation provided by a transformer but I'd also like to reduce costs. A simple buck converter topology would work fine except I may have issues with discontinuous mode which I would like to get around. In fact I need a somewhat variable supply of ~1A to ~100A.

I'm not sure I could implement an efficient topology such as Full- Bridge Push Pull converter at those currents and voltages. The main issue I'm having is determining the transformer and inductors to use. The site you gave me shows some cores that should work but I'm not sure how well as they still seem awful small. I would expect a transformer to be at least the size of my fist and not one of those ones one typically sees in PC SMPS's. Of course I may be wrong if the size is somewhat proportional to the switching frequency.

Just a thought but I wonder if I could parallel several smaller buck converters, with or without isolation such as the flyback or fullwave, which would allow me to switch them in as I need more current. I could have one 50W, 25W, 10W and 1W. This would require current sensing but shouldn't be an issue as it doesn't have to be accurate. I guess the problem here would be current sharing?

Thanks

You can. In fact, this is what most multiphase controllers do = automatically, with the bonus of phase shift between outputs, which = reduces average current ripple.

If you're making one from scratch, remember to throttle them in = parallel, so none are hogging current.

Tim

--=20 Deep Friar: a very philosophical monk. Website:

A

It's not small as I 'think'. That was the whole point. I did the wrong calculation and it came out extremely small. Even with the fixed calculation it still seems relatively small. I was figuring it would be several hundred mH's and probably twice the size of my fist.

oups.com...

0A

Ok, I'll try to look at some of those circuits and see how different they are and maybe try to peek inside my welding machine and see what in there.

Thanks

snipped-for-privacy@w12g2000yqj.googlegroups.com...

cally, with the bonus of phase shift between outputs, which reduces average= current ripple.

so none are hogging current.

That maybe the way to go then. Maybe not as cost effective because of the added parts but probably a bit easier to do if I can get the control issues figured out. Although I don't know if there is a huge difference between doing the 50A and 100A versions. BTW, those should be amps instead of watts ;)

Possibly not, if you let the smaller ones go into constant current mode at their max output, providing you can setup the voltage feedback to allow bigger ones to shutdown when not required?

Some multiphase controllers shutdown some phases when not required to meet output current, needed for example with those modern CPUs go mA to many tens of Amps in microsecs, depending on spot demand.

Grant.

You're talking about a 500W supply -- that's not going to be cheap unless you can build them 10000 at a time like the PC manufacturers do.

The last switcher project that I worked on was a 300W unit that had two transformers, each a cube about 1.25 inches on a side.

Yes, transformer size goes down as switching frequencies go up, for the same reasons that inductor sizes do.

You could, but you'd have all the same issues with each buck converter as you would with one big one.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

Core area is the key; toroid, cup core, e-core, whatever will give the cross-section needed. But. I think you will find toroids cost a lot more.

If you are going to use multiple converters, why not look more closely on some multiphase system, running the separate converters from a single clock with different phase shifts.