SMPS's for thousands of Amps?

[Flyback]

For a single non-isolated SMPS, what is the highest current they tend to go up to?
How do they manage the ripple current and switching node for those converters?
Also, what are the likely switching frequencies for eg a 1000-15000 Amp , single stage SMPS?

Also, what type of inductor cores do they use?
And also, what kind of noise mitigation measures do they take?

For example, a 4000 Amp SMPS, done with a single stage (no paralleling)......what kind of resistance do they get in the inductor?
I can't imagine say 1 milliOhm of resistance in such a thing being achievable?.
(eg say if it was a PWM controlled resistive load, with filter inductor.)

There are no pictures of these kind of things on the web. Someone gave me a doc, but no real specifics. No detailed circuit diagram or BOM. No pictures.
No scope shots.
I guess this stuff is deep industrial secret?

The following is just called a "rectifier"...so it just rectifys 3 phase?...not an SMPS.

I am doubtful there are many high frequency SMPS's which do 1000Amp in a single stage?

 

[schmitt trigger]

Back in the early 1980s, I saw something similar built with inverter-grade SCRs made by Semikron. Those SCRs commutated well within the audible range, the inverter made a loud whine and was installed on a separate room.
Because SCRs can’t be turned off by removing the gate drive, commutation was achieved with an immense tank circuit.

 

[ZipZapOuch]

4000 Amp SMPS, done with a single stage (no paralleling)......what kind of resistance do they get in the inductor?

It's called engineering for a reason. All the physics equations are available. Make the calculations and start designing. There is no deep industrial secret. It's just math, physics, and the design effort to work around the parts that are available.

The biggest problem will be DC resistance of the coil. 0.1 ohms of resistance with 1000 amps will be a 100v drop and 1000w of heat. So, you'll have to select fat wire for the inductor and you'll need a big core to get the required number of turns.

And since you won't have room (or available DC resistance) for a lot of turns, you'll need to operate at a fairly high frequency, so you'll need a core material with very low hysteresis.

But, high switching frequencies and high currents mean switching losses are a concern and you'll need the right switching methods for the frequency that works.

Speeking of high current DC supplies, I've been to the old Dupont Sodium metal plant in Niagara Falls. They were using 1" x 12" bus bars (solid copper) to feed each wall of reactors. If I remember correctly, it was about 7 volts* at 20,000 amps. The plant had three wings and each had a similar setup with its own rectifier.

The reaction only needs about 4 volts but some resistance forms as the reaction continues so the over voltage is needed.

 

[ronsimpson]

I am using 'power bricks'. I cannot show you exactly what we are using but this is close. I am using 1mOhm shunt to watch the current.

After rereading your post you want a low voltage part. I will think about that.

 

[Flyback]

Thanks,
Thanks, And on a related note does any one know where to source a 700uH inductor rated 1500A, and will suffer 120Apkpk ripple?
And must withstand voltage across it of some +150V then -150V at 1kHz.
Will only be conducting for 100ms every half hour.
DC Resistance preferably no more than 5millioms.
I assume we are talking an iron cored part here.

 

[ZipZapOuch]

You'll need to worry about EMI with those conditions.
And I doubt you'll be able to charge the inductor in that short of time so it will need to be in a standby mode (energized but only supplying a dummy load) and then switching over to the actual load.

 

[ronsimpson]

Some people in my group build high power "things" and we use modules. We parallel up bricks to get to the power level you want.

 

[Flyback]

Thanks, i agree that's a good way, but in this case the document specs out just a few (4) paralleled converters for a 6kA power supply. I wonder if anyone has ever done eg a 1500A current source PSU, by way of say 75 20A current source's in parallel

 

[ZipZapOuch]

Thanks, i agree that's a good way, but in this case the document specs out just a few (4) paralleled converters for a 6kA power supply. I wonder if anyone has ever done eg a 1500A current source PSU, by way of say 75 20A current source's in parallel

It could be a mess unless you have current limiting on all of them.

 

[Diver300]

The biggest problem will be DC resistance of the coil. 0.1 ohms of resistance with 1000 amps will be a 100v drop and 1000w of heat.

It's far worse than that. 1000 A and 100 V is 100 kW of heat.

Many years ago, on a 500 kW variable speed DC drive, a connection was glowing red hot. I worked out what resistance that would need and it turned out that 1000 A and 1 mOhm gives 1 kW of heating.

 

[Diver300]

I am using 'power bricks'. I cannot show you exactly what we are using but this is close. I am using 1mOhm shunt to watch the current.

Have you looked at the Hall sensor current meters? If you are measuring 1000 A, that 1 kW of heating is going to be costing somewhere around £1000 a year if it runs all the time. Your payback on a Hall sensor meter will be quite short.

 

[ronsimpson]

75 20A current source's in parallel

75 is too many. I can do 200A so 8 in parallel.

 

[Tony Stewart]

Wide sheets of copper foil are used in the secondaries of MVA transformers using Silicate coated CRGOS laminates each several uF between layers. That would still yield a relatively low SRF but much higher than the grid.

It seems to me you should consider just a dual array of Li Ion cells in parallel for operation and series/parallel for charging.

 

[ronsimpson]

Flyback: made in Hong Kong

I found a machine in China that makes these. I bid on making some large coils. Glad I did not get it.

I found flat/rectangle Litz wire which my winding machines can handle. (no pictures)

 

[AnalogKid]

I guess this stuff is deep industrial secret?

No, just a rare requirement. If you work at CERN or at a hypersonic rail gun test facility, it's just a normal day.

Occam's razor - Wikipedia

en.wikipedia.org

k.i.s.s.

ak

 

[Flyback]

75 is too many. I can do 200A so 8 in parallel.

Thanks, what would you say to thread #10 of this..

Biggest inductor you can realistically get? - Page 1

Biggest inductor you can realistically get? - Page 1

www.eevblog.com

...which says 40A is maximum for a single stage SMPS?