So an idea I have had for a while is to try and make my own HID ballasts, primarily HX-NPF types. Since ballast cores are made of several thin sheets of steel that are welded together, it seems like I might be able to do something like that with sheet metal and a CNC cutter. The plastic part that goes around the ballast core that separates and protects the windings could be 3D printed, and I could make a tool for winding the ballast pretty easily. The main things I still don't know how to do would be finding what thickness of wire is needed in each winding, how many turns there needs to be, and what kinds of shunts would be needed. I'm most interested in making ballasts for 50w mercury lamps, and while I do have a few already and I could tear one apart to figure out what exactly I would need to replicate, I'd prefer not to do that for obvious reasons.

Does this sound like it's actually an option, or am I too crazy?

It depends how good your maths is. I have tried but didn't work out. The winding bit isn't that hard you just need the right size of wire that takes the current, low resistance and will fit into the core. Hardest part would be figuring out the windings, that's the part I'm struggling on because If I managed to fail year 10 maths then complicated magnetics formulas will never make sense. If you are really dedicated to making them like myself they you should be able to. Just ask away and surely you will have an answer.

You might have to do some digging but I believe the transformer E-I laminated cores as well as the bobbins are available as an off the shelf part. The wire for the windings is known as magnet wire and I've used this to refinish ballast cores before. It does stink quite a bit so I would do it outside or in the garage with the doors open.

There are kits available online to wind your own power transformers, perhaps you could modify the core on one of those to add the required magnetic shunts to do the current limiting.

Yeah if I ever were to make my own HID ballast, 50w MV would probably be the first one I'd make. The correct ballasts are too rare to find a reliable supply of.

What you can do is message someone like Ash or Medved as they know everything

Yeah if I ever were to make my own HID ballast, 50w MV would probably be the first one I'd make. The correct ballasts are too rare to find a reliable supply of.

I may be able to hook you up with one. Its a 240V 50hz one but if you run in on 277V it should work as I'm told.

Those 240 volt chokes will work ok on 60hz no problem

  Iv been Using 58 watt fluorescent  chokes as  it a good match using 240 volt.   
   I am starting to learn how to rewind Fluorescent ballast
   These old HO rapid start ballast.  These fluorescent ballast is not that far off running hid lamps.   

  Reason i am doing this with in 20 years all these ballast will become very hard to find.   I go to scrap.  yard and get few ballast At a time to learn to rewind them.   100 watt is easy one to make.   

Those 240 volt chokes will work ok on 60hz no problem   

With reactive chokes you can't go up or down in frequency or the reactance unless you change the input voltage so make the choke limit the current perfectly

So an idea I have had for a while is to try and make my own HID ballasts, primarily HX-NPF types. Since ballast cores are made of several thin sheets of steel that are welded together, it seems like I might be able to do something like that with sheet metal and a CNC cutter. The plastic part that goes around the ballast core that separates and protects the windings could be 3D printed, and I could make a tool for winding the ballast pretty easily. The main things I still don't know how to do would be finding what thickness of wire is needed in each winding, how many turns there needs to be, and what kinds of shunts would be needed. I'm most interested in making ballasts for 50w mercury lamps, and while I do have a few already and I could tear one apart to figure out what exactly I would need to replicate, I'd prefer not to do that for obvious reasons.

Does this sound like it's actually an option, or am I too crazy?

First you would need the steel for this. It is a special (sometimes called "transformer", "dynamo" or "electrical") steel, designed specifically for magnetic cores with high permeability, high saturation flux density and high electrical resistance, so with high Si doping (around 4%, if I remember right) and few other "nuances" in the composition.
Regular construction steel wont work at all - you would have HUGE losses in the core... It happens to be the same steel as used for large power resistors (current limiting resistors in electrical locomotives before the PWM thyratrons, then thyristor and recently IGBT based electronic regulators took over that function; or used as "dynamic brake load" resistors in many locomotives even today - to dump the power generated by the traction motors during dynamic braking).
This steel isn't usable for anything structural, because it is mechanically rather weak and brittle, so it is made solely for the magnetic cores of generators, motors, transformers and so on, and for the high power (I mean 100kW and above) resistors.

The cut steel plates have to be specially treated after cutting:
- Deburring, to allow smooth assembly without damaging the isolation layer (the cutting tend to leave sharp edges or spikes)
- Oxide coating to make an insulation between plates
- Annealing (could go around the oxide coating) to releave the stresses along the cut lines, so to restore the magnetic properties there.

So it is far more than just cutting a random steel on a CNC cutter. But if you know what you are doing and have a source of the raw material, it does not need much special tools except the cutter and maybe a torch (I guess you have both).

And to the ballast design:
The autotransformer ballasts (both HX, as well as CWA) are a device integrating at least two magnetic functions into one device (voltage step up transformation and the series inductance), so need special core shape. Regular transformer EI plates are by far not sufficient as a building material. So that would go into the direction of cutting custom parts, if you master that process...
And of course, there is the whole design.
But the HX are the simplest to design: It is just a transformer, with a nonsaturated magnetic shunt between the primary and secondary coils. So pretty basic magnetic calculations.
The CWA is more complex, because there the shunt is supposed to saturate at an exact current and with exact current vs inductance shape (to get reasonable compromise between stabilization performance vs current crest factor). So a bit more complex and needing more "try and error" work. But I guess if you have the raw material, machine cutter and the time and attitude to do these experiments, it may work well...

I would go the route of breaking up the design to 2 individual blocks : Step up transformer (from 120V to desired Voc) + Simple choke ballast

This would simplify all calculations and building, as well as allow you to experiment with each part separately or substitute it for another part (including off the shelf parts like step up transformers and US or European FL and HID ballasts)

Both can be assembled using factory made components - You can buy electrical steel laminations pre-cut into the right shape, treated with all required treatments, and ready for use. For example from manufacturers like E. Grau (Germany). Check out also local companies in your area that make transformers, they might be cool about selling you a bunch of material for DIY. Same for Plastic bobbin to wind on, brackets to assemble it all etc

The transformer and ballast will require different types of laminations :

Transformer laminations are standard E/I laminations

Ballast laminations have the center leg of the E a little shorter (about 0.5..2.5mm), creating a gap when assembled. Those are not common but you gotta be able to find them as well

There is a possibility to use standard EI plates for cores with designed air gap (output transformers for class A audio amplifiers, or various chokes, include ballast ones):
You put all E's into one stack (and insert that into the winding bobbin), align it well, so you get an E-core block.
Put all I's into other stack, so make an I-core block from it.
Stack paper layers so you get thickness of 1/2 of the desired gap thickness. The 1/2 comes from the fact the magnetic flux has to pass it 2x - once in the center leg and once in the side legs.
Separate the E core part from the I part by that paper stack.
Make a clamping structure, mainraining a reasonable pressure fixing all the assembly together.
Measure the impedance at the desired bias (or working) current, adjust the thickness of the paper insert to get the desired impedance (in your case the required lamp current when supplied from about 90% of the OCV - you will need either a variac upstream of the boosting transformer, or some serkes resistor - e.g. an incandescent lamp of enough rating).

But there is one pitfall with this construction: The gap around the outer legs forms a significant field outside of the core.
So there should be no magnetic material nearby (so enough spacing to e.g. installation box steel wall, plus tye clamping structure should be nonmagnetic). Otherwise two thing may happen:
First it would be loud. The field will create magnetic forces towards the nearby magnetic part, causing it to vibrate.
Second the magnetic material will form an undesired path for the magnetic field, both altering the inductance, plus generating large losses in that material. I have seen such structures red hot on a home made transformer - quite a "smoke generator"...
Using brass for the clamping thing and keep steel 1cm or more away from the core is OK...