My comment was not clear, what I meant was that they did not "convert" the traction motors, they had to replace them with AC motors. I was not addressing the generating system. Sorry for the confusion. The starting question was why can't reconvert these toy train motors from DC to AC, and I was trying to say, you have to replace not convert.
Locomotives went from universal DC traction motors to AC induction motors (using an AC-DC-AC transmission).
Lionel trains went from the AC universal "Pullmor" motor to a DC can motor that is used in today's hobby products.
On the surface the prototype change over seems analogous to the hobby change over but they really are not.
If I interpret your question correctly, you are asking why we can't or don't go back to the universal motor of the old days. There are several reasons as to why:
1. The motor is too large to use in scale proportioned products. The Pullmor motor is a scale 7 feet across which requires hoods to be a scale 8 feet to accommodate the motor. Hoods on most locomotives are 6 feet. This is a big strike against the Pullmor motor in modern products.
2. As G,J, wrote earlier, the Pullmor motor is inefficient. The Pullmor motor is a design that is essentially from the 1930's. It has tinny motor brushes in proportion to the size of the motor. It also has a flat faced commutator which leads to uneven wear of the motor brushes. The tiny size of the brushes in conjunction with the flat faced commutator leads to poor contact, excessive sparking, and ultimately a large heat generation (and ozone production ). In the end, the Pullmor motor is roughly 35% efficient whereas the DC can (RS385PH) motors are roughly 60% efficient.
3. As G.J. mentioned, the armature of the Pullmor motor has only three poles. 3 pole motors are generally not as smooth in operation as armatures with more poles. The field is important as well. Lionel made different designs of the universal motor and, not getting into much detail, the field contours greatly impact motor performance.
4. Expense of manufacture. The Pullmor motor only has utility in Lionel trains and as such had limited production runs. Limited production run translates to more expensive. DC can motors are used for many other applications and are adapted to use in model trains. They have much larger production runs and benefit from the economy of scale. Can motors are much less expensive.
5. Speed control technologies could not be adopted for use with the Pullmor motor.
All these are solvable with some design work. The flat faced commutator can easily be replaced with a drum commutator for better electrical continuity. In truth, nobody even uses flat faced commutators anymore (and probably the majority of industrial motors don't even use commutators anymore). With modern winding and production technologies, the number of poles can be easily be increased on both the stator and rotor. Favorable gear ratios can be used. The size of the Pullmor motor can easily be reduced in size and fit into the width of a scale hood. Inherently, universal motors are more efficient than can motors when modern designs are used. Most applications that utilize universal motors also utilize speed control to maintain speed under different loads so speed control can work with universal motors. However, I do not believe that Lionel, MTH, Atlas, 3rd Rail and the others employ engineers that are knowledgeable in motor design. If they did, something like Lionel's F40PH fiasco (the motor was too tall and cut into wires which blew out the whole thing) would not have happened.
So in short, yes it is technologically feasible to go back to universal motors and have an improved product.
In the end, it is just easier for the manufactures to buy an off the shelf DC can motor and plop it into their product.
As far as AC-DC-AC system like in the videos to control synchronous or asynchronous motors like what exists in locomotives, off the shelf products exist for that too that would easily fit inside O gauge/scale (even HO scale) trains.