So I was reading about winding DIY transformers using steel bar (here's the link: https://www.allaboutcircuits.com/textbook/experiments/chpt-4/build-a-transformer/) and now I want to know if it would be possible to do something like this to power a low wattage MV lamp (like 50w or 75w). I don't know the specs for a 50w MV ballast (no way I am tearing one apart), but has anyone ever tried something like this?

First the article you mentioned shows only an education toy, but the performance is unusable for any real task.
But it is still possible to make a decent transformer, but you have to stick to the standard transformer parts, like the core plates (forget to make anything like that as DIY even from scrap, it is highly speciallized chemistry/metallurgy; even cutting that material without proper equipment means you will ruin the material).

So calculating the transformer design is OK (OK means you are very likely to succeed if not making some error).
Making bobbin for the winding is OK too (hard paper sheet working, gluing,...; although it may offer less winding room than a facory made one, need counting on that in your design). But could be salvaged from some other old transformer.
Winding the wire is tedious, but perfectly OK.
Assembling the core stack from plates is OK.
Making the core plates - forget it. You need to get some factory made, e.g. salvaged from an existing old transformer of the desired size.


Then for the ballast: A commercial 120V MV ballast integrates at least two functions: A step up transformer (120V->230V) and a series inductor (either a linear one in case of a HX or a saturable in case of a CWA). To integrate these two functions, you need a magnetic shunt of exact parameters between primary and secondary. Very economical for a mass production, but practically unfeasible for DIY (needs special shape of the core plates, used only for such ballast, so possible to salvage only from another ballast).

For a DIY design you will need separate components for each function (separate 120/230V autotransformer, plus a separate ballasting inductor; you have to be consent with larger assembly size and a bit higher losses). These individual components are then feasible to make using standard generic transformer parts (the inductor core assembly uses standard plates, only assembled to allow the inductance setting air gap).
Although possible, I wouldn't start with a CWA, because there the inductor has to have not only the inductance correct, but the saturation current as well. Aligning two critical parameters in one coil I see as nearly impossible, mainly when you wont have any detailed specs of the used core material (assume salvaged from some transformer). Even in a proffessional environment designing that involves a lot of "trial and errors", but 10's wasted ballast attempts is a no problem there, when rhe resulting design is then reproduced in many 1000's. DIY has to be made to work already the first piece...

Read through : https://ludens.cl/Electron/trafos/trafos.html

Note, that they insert the E plates from alternating directions. That is, as in most transformers the aim is to make the leak inductance as high as possible (i.e. the current drawn by it as small as possible) - the alternating directions help reduce the effective gap between the E and I parts

In a choke ballast (that is what i'd recommend starting with), the aim is to make the inductance (the same one, but not considered as a "leak" in this case) not as high as possible, but instead of a specific value (i.e. to get the correct current for the lamp). So, there is a desired gap size. All E's are inserted from the same side, and there is a gap between them and the I's

The gap can be made as some layer of non magnetic material inserted inbetween (then it's effect is doubled, as the flux have to cross it twice), or using special E's with shorter center leg, or special I's with indentation in them. The latter was and still is one way used in actual ballast production

The more modern approach used by ballast manufacturers like Vossloh Schwabe is, to use an E with short center leg, and shorter I that fits into the E completely, which then is pressed in until the gap is of the exact wanted size (it is pressed while the ballast is connected to a measurement circuit, which determines when to stop). I dont think that the "short" laminations are generally available for purchase, and they would probably be too damaged to reuse if taking them out of a ballast