Permanent Magnet Open Frame Motor

Yes.  But but neither the windings on the armature or the strength of the magnet have been optimized for the design, so it would probably be very inefficient. 

How about using neodymium...smaller magnet required to generate field and could maybe be more easily optimized due to space availability...or perhaps stronger field using magnet of same dimension as field coil but with neodymium material generating the field...

Short answer is YES it could be done.  And you could indeed control motor direction by reversing polarity to the brushes.  However the 3-pole armature would still produce big changes in back-EMF as it rotates, which would likely frustrate attempts at closed-loop speed control.  TrainAmerica experimented a LOT with this, and the 3-pole armature was a deal-breaker.

Marklin locos are built a lot like postwar Lionel.  In the late '90s Marklin offered an upgrade kit for their vintage AC motored, spur-geared locomotives.  It actually replaced the wound field assembly with a new one containing a rare earth magnet.  More importantly, the kit included a FIVE-POLE armature to complement the new field.  (Google "Marklin Five-star high efficiency propulsion" for details.)  I would love it if Lionel offered something like this to upgrade a postwar 2046, for example.  But I've been waiting 20 yrs, and unless Neil Young makes it a personal project and gift to the faithful, it probably won't happen. 

Another approach is to replace the field AND armature with a "pancake"-style can motor.  See this thread for an example:  https://ogrforum.com/...-gear-2036-127-pitch

There are also a couple of threads on the forum about replacing Berkshire or Hudson-style longitudinal Pullmor motors with a cylindrical can motor.  Frank Timko offers a service like this.  But the can motor has to be pretty small to match the form factor of the Pullmor's offset field winding.  And IMO a tiny can motor isn't well-suited to the tall gearing of a postwar loco.  Good question, worth discussing!

So basically...this could probably be done, given:

Willingness to invest r/d money to create 5 or 7 pole armature...make brush holder to complement armature ...manufacture permanent magnet assembly...and apply to Lionel open frame architecture...

Sounds like a great go-fund-me maybe...

The pullmor motors are universal motors and already can run on DC "out of the box". As such, I think the appeal, while understandable, is very narrow given the many other accepted alternatives such as installing a pancake motor, command control boards and so on...

American Flyer did this in 1950, in their DC series of engines. They used a permanent magnet field with a standard AC armature.

Hmmm...another reason to win the lottery...

@bmoran4 there really are no other alternatives.  I agree with Gibson Man about the Kickstarter / GoFundMe campaign (although a benefactor with connections like Neil Young would have a huge headstart on the learning curve.)

Some of us don't want "scale" trains, or just don't have the space for them.  I've racked my brain for years, and there is currently no upgrade path to make beloved postwar steam locos run consistently at realistic speeds, nor even achieve a modest improvement in performance.  So  they end up piled on tables at meets like so much scrap zinc.  An improved drivetrain might have been developed in the '90s.  But then RailKing surfaced, and LionChief Plus a few years ago.  Unfortunately, to my eye, many of these modern "toy trains" just aren't artfully rendered.  Where Postwar steam can resemble the prototype compressed through a telephoto lens, some of the newer ones are more like an S-scale body slapped onto an O-gauge chassis.

If you squint hard, the LionChief Pacific resembles a 2055.  Sad, but I'm not sure I'll see closer in my lifetime.

How would this be better than just doing a Timko motor replacement?

Now I'm wondering if one could install a small can motor into the space occupied by the armature... 

Mitch 

So the idea, for me, is a drop-in, DC motor, with the form of the open frame motor, that can be speed-controlled...such that you unbolt the old pullmor...drop in this new DC motor...connect it to an err/3rd rail motor drive/back-emf speed control board...and for all intents the engine retains a vintage appearance but runs like a modern engine.

I have no doubt it would not be a financial failure...more of a money pit based on the niche it would appeal to...but there are a bunch of engines out there that could potentially be modernized to run in a more true to scale manner than just running like little speed demons...

If I won the lottery, this is indeed a project I would take on...consider it a loss and a tax write-off...and a gift to myself and my fellow hobbyists...

Is it possible? Yep, as others have said (the Pullmor is a universal motor from the get go). In reality, I don't see any reason to do this, it would be a lot easier and cheaper if you like an engine to drop in a replacement motor that can be controlled via DC or command control easier. Pullmor motors are kind of cool, I love the sound of them and so forth and am familiar with them, but I wouldn't modify them like that, not worth it, would just replace it with a modern motor. 

@C W Burfle Timko only offers replacements for steam locos with longitudinal Pullmor motors (773-type Hudson, Berkshire, 18006 Reading T-1, etc.)  The motor he uses is tiny; I doubt it makes as much torque as the original, but it is a 5-pole DC motor suitable for use with ERR.  Despite its small size, it sticks obviously out of the cab.  There is no upgrade path for the great bulk of pre- and postwar steam having the "parallel plate" type motor.

@M. Mitchell Marmel this is *exactly* what Lionel did in 1979 with the 2-4-2 Columbia type train set loco, and again in 1986 with the 2-6-4.  A small Mabuchi motor was installed in place of the armature and brush plate.  This change was made for cost reasons and the motors they used were nothing special.  (Only a very small cylindrical can motor would fit crosswise within the boiler shell.)  The 2-6-4 received an additional stage of reduction gearing to make up for the lack of torque.  I had one of these.  When it ran, it sounded like a movie projector, and it suffered a lot from "cogging" at low speeds until it was thoroughly broken in.  Frankly I would rather have a stock 2037. 

@gibson man neither one of the motors / designs mentioned above works well with ERR.  I've tried and so has @gunrunnerjohn, I believe.  Again they are 3-pole non-skewed motors.  The gear ratio and lack of a flywheel creates challenges, changes in load and back-EMF that the electronics can't overcome.  My post above contains a link to a thread in which someone installed an unknown "pancake" type motor in a 2037.  I'm not sure what motor was used, but if it's good quality, that might make the modified loco a candidate for ERR.  My understanding is that most motors with a "pancake" form factor are designed to run at a constant speed for electric clocks, turntables, etc., so probably not ideal for a variable-speed application like electric trains.

What I have in mind is a high-quality upgrade or a replacement chassis.  K-Line did a much better job upgrading a classic postwar design when they reissued the Marx 333 in 1992.  I think the TCA's Al Ruocchio had input on this design.  It was a whole new chassis with a high-ratio worm gear drive, that gives good control and realistic speeds.  Unfortunately it lacks a flywheel, and there is absolutely no good way to add one, so the loco is not "kid-friendly."  Careless use of the direction button results in a bone-jarring stop, and in extreme cases the brass worm could cut a burr into the fiber worm wheel.

Lionel tried a couple more times... around 1995 they released a 4-6-2 Pacific using a simplfied version of the old 2-6-4 superstructure.  These locos had a plastic(!) chassis.  An HO-sized can motor equipped with a tiny flywheel drove the last axle.  I think this light-duty drivetrain survives in the later models of Thomas and Percy.  The mounting points inside the boiler shell were changed to accommodate the new chassis, so it can't be retrofitted to a 2037, etc., even if you wanted to.

A more promising attempt was the 1998 Torpedo.  Instead of re-using the 1668 mechanicals, the chassis was based on the 18000 scale switcher.  I think there were QC bugs with the electronics, maybe even a few cases of zinc pest? and the switch to Korean production the next year killed any further applications of this mechanism.  Still waiting.  

There is no upgrade path for the great bulk of pre- and postwar steam having the "parallel plate" type motor.

What would be the upgrade path for a parallel plate motor? It seems to me that any upgrade would require the separation of the two parallel plates to get to the field. It's no easy task to separate and rejoin those plates. 
I think your mention of an entire replacement chassis would be the way to go.

IMHO, some of the old modern era locomotives that used that postwar style parallel plate motor mechanisms would benefit from a replacement chassis. In my experience, they don't run as well as the postwar ones.

I had a pre-1950 2025 running on my layout today. Still runs very smoothly and quietly on very little power. 

@C W Burfle one upgrade path would be very similar to what the original poster suggested, and what Marklin did with its Five Star High Efficiency Propulson.  (Postwar fans, I strongly encourage you to google what Marklin did!)

No need to separate the plates, although I think it's a shame that the O gauge chassis can't be disassembled in a manner similar to the Bild-a-Loco motor.  One Forumite did manage to split the frame and put it back together with screws: see the thread linked in my first post.

To replace the field, I suppose you would drill out the spacers holding it in place and install the new one, energized by a rare earth magnet and secured in place by screws this time.  Replace the armature and brushplate with a skewed 5-pole version to complement the new field.  The locos would still be geared for warp speed, but I'm confident these changes would net a major improvement in operating consistency, and make them relevant again in the modern context.  Heck, just a 5-pole armature and complementary drum-style commutator would probably give a 30% improvement!

Lately I've been imagineering a conversion to a worm gearbox.  Motor directly above or a little behind the drivers, and a lower worm shaft driven by belt and pulleys similar to 3rd Rail steam and some 2-rail O scale diesels.  But this would require a completely new chassis, or at least some carefully machined and located bearings for the upper and lower shafts, a suitable gearbox, pulling and pressing to install the worm wheel on the axle, etc.  If this were to be a Kickstarter project it would have to be sold as a bolt-in replacement.  We all want the things we love to be perfect.  These upgrades will probably never happen, but it's fun to dream!!

A little-known fact about the design of the open-frame Lionel motors with the 3-pole armatures is this:

Enough winding turns are provided in the field that the iron circuit winds up saturated at a very low voltage,  [I measured saturation reached at 4-volts rms for the F3/GP7 motor (2028-100.)]  While this does not produce a constant speed motor (with speed proportional to applied voltage), it does produce a constant torque motor (with torque proportional to applied voltage).  The relationship is linear above 4-volts-- that is, within the operating range of the motor. an upward sloping straight line results from plotting torque for each voltage.

Plotting voltage upward on the left edge ot the graph worked best  for me--a parabola pulls the graph from 0,0 in an upward curve to the right, changing from a curve to a line at V=4 as the field iron saturates. [The straight, sloping line will appear to have begun at 2,0.]

Having set a given torque with a transformer at a suitable voltage, speed will remain constant on straight, level track with a fixed number of cars.  For an upgrade, a section may be inserted with a greater voltage, according to the steepness of the grade.  For a downgrade, likewise with a lessened voltage.  Actually, a ZW transformer is most convenient for this, as the B and C outputs can be used for the up- and downgrades.  [This was a suggestion from the booklet included with set boxes in the postwar era.]  An entirely acceptable hands-off operation results-- I built such a layout.  I had only a 1033 90w and an American Flyer 100w (75?) transformers; a 2023 diesel and a prewar 4-driver Torpedo engine (Torpedo and the AF being a gift from a friend); later a 2046.  So there was very little level but very long grades; train spacing was easily maintained by adjusting the 1033 on the upgrade train.

It isn't necessary to repeat my experiment, but to do so get a rolled paper stick from a candy sucker, attach it as a lever arm to the output worm of an unused motor by wrapping it on with a rubber band, setting it to 4 inches which is 10 centimeters.  Working on a kitchen island, park the end of the lever on a food scale marked in grams/oz so as to get gram-centimeters locked-rotor torque.  Work fast as hot windings will distort results.  For speed, pre-mark the faceplate of a 1033 with the volts output at the positions of the throttle handle, both for the 0-11 volt range and for the 11-16 volt points of the 5-16 volt range, using a soft black lead.  Three pieces of wire and four alligator-clips will suffice to clip the 1033 to the motor.  [If the windings get too hot from the rotor being locked (with each brush on it's own commutator segment), the motor will be damaged-- which is why I preferred to use the 10 centimeters as a quick multiplier   ]  [You may feel uncomfortable above 11 volts-- don't over do it.]

--Frank