Thanks for the input so far everyone! The application is basically to try to create a low-field, low-cost MRI machine. I know an MRI expert who believes it will be possible to image with such a machine given a field of at least a few mT. Using superconductors for this is really impossible because I'd have to cool them with nitrogen and helium and that takes low-cost out of the equation.
The big transformer (e.g. modified MOT or from a welder) would give AC current with mains frequency.
A set of parallel batteries can give a high current for a relatively short time.
A PC supply can give a more or less well well regulated 3.3 V DC at some 50 A, maybe more with some models
In theory with a PC motherboard there is switched mode regulator for some 100-200 A at some 1-1.5 V. However it may not be easy to to make it work without the CPU.
So it kind of depends what is really needed: AC/DC, time, accuracy or stability of the current.
For the main coil, I'll need DC. It will ideally need to be on for half an hour to an hour at a time, with good current stability. I'll have to ask that MRI expert I mentioned what the exact stability needs to be, but the more stable, the better. I will also need AC at some point for an orthogonal magnetic field, but that's probably a question for another post or this one will get cluttered.
Application is to create a magnetic field (as large as reasonably possible) in a loop or several loops of copper conductor.
I don't see a Watt rating on most of the rectifiers I'm looking at.
We've made a few electro-magnets for Industry over the years - some in the many KW range, huge buggers, but we've used 3+3phases (STAR PLUS DELTA) and a sheetload of Semikron big-ass diode blocks. They make them in incredible high currents. Heatsinks were huge !! with several fans. It was for our now defunct car Industry.
We've also made "smaller" ones using Meanwell Power supplies - LRS-350-3.3 (= 3V3 at 60A) output or LRS-350-4.2 (4.2V at 60A), off-the-shelf units. I think these (or can be customized) to be SYNC'd, providing near unlimited current. They have other Voltages / models as well. Just another idea.
Oh, and for testing, we had a 3 Phase ~5KW variac :-)
That sounds like it was fun! I need something as cheap and simple as possible as I don't want to invest a ton.
You should move to superconductors. Extremely high | H | is simply not possible over an extended period of time unless you're using superconducting electromagnets. A coil made from thin copper wire will either melt or physically tear itself apart in minutes (if not seconds) at 50-100A...
Yes, but cost would become prohibitive... I have heavy wire that I was planning to use, which should be able to handle 50A. I can calculate how hot it will get, but it doesn't sound impossible right now.
That seems like a good option. As far as I understand, you can select the current and let it continuously supply the 100A through a length of copper? Is the stability relatively good?
Yes, you select the current (in amps) by turning a pot and whatever load you connect, it will keep the current constant. Of course only up to rated voltage, usually 80V. The TIG capable welders additionally have this remote input (where you can connect sth like arduino) and control current remotely. The inverter welders can work 100% of the time but usually the 35% current is placed on a sticker (check the datasheet). My welder does 140A at 35% and 100A at 100% and it will shut down to cool itself if you exceed the limit.
This is essentially a regulated DC current source powered from mains. Of course it is isolated from mains galvanically but be careful, 80V is not pleasant to touch and 100A DC arc can ruin not only your day.
As for the current stability - good question, but I am pretty sure you can find some current ripple screenshots on the internet. I do not have adequate shunt but I do have such welder and a scope.. Hmm, maybe 0.1R resistor in a glass of water?
I think the current stability is the main issue. I should have mentioned that requirement in my opening post, but yes, it needs to be reasonably stable. I'll have to dig around for spec sheets on those.
Application is to create a magnetic field (as large as reasonably possible) in a loop or several loops of copper conductor. I'm only seeing super-expensive forklift batteries, not sure where I'd get a deal on one...
I'm don't think you've given quite enough detail about the application. We know you want DC, but how smooth? A forklift battery or whatever will give you a smooth DC current, whereas a transformer + rectifier, welder etc. will give you a pulsating DC current (at 2 x mains frequency if you use a bridge/full wave rectifier (reservoir capacitors don't sound practical at these sort of currents). An SMPS (eg heavily loaded PC supply) [Edit or Inverter welder] would give you smooth DC with some high frequency ripple.
Does this matter for your application?
My apologies, I was trying to be concise but seems like it backfired. Since this is for an small MRI machine (think imaging berries, or at most your head), I think the current needs to be pretty stable. I'll get back with the exact smoothness requirement once I've talked to my expert. I was thinking of running the transformer, if I go that route, off a DC variable PSU capable of 30V and 10A, which should be more than enough power. Would anything be introducing additional instability at this point?
Why not use thinner wire with more turns (for the same total cross section) and a more common power supply rating, or an offline supply (basically just a rectifier, since the coil is its own filter)?
If it's about cooling (water), cooling tubes can be added regardless. Mind to use a reasonably conductive filling or potting. It is a bit more convenient to use tubing, true. As long as the total tubing length isn't so much that you need 10atm to force the water through it... (in which case you need bigger (and thicker walled, may be special order?) tubing, and even more current).
Or if this is really hot stuff, you'll have to use high pressure water and lots of it, and those perforated plates they use for the big boy magnets.
Tim
That would certainly be easier. I don't have a full grasp on all the MRI mechanics yet, there are some requirements like field gradient and I'm not certain if a longer solenoid will meet all of those, but I'll definitely look into it.