+1 on Bob's opinion that this may be the regulator board.  It's an easy to repair, fairly modern design.

I can't be certain but the two tubular devices look like (aluminum body with once intact plastic skin) electrolytic capacitors; they are something that is very cheap to replace, and something that does fail with time.  Not having a schematic for this lovely, easy to reverse engineer single sided, through hole component board makes it a bit tougher. 

Worst case, you can get a Chinesium brushed generator regulator to replace it with, but it's well worth trying to fix if you can.

A meter which can measure capacitance is pretty cheap these days, if the labels aren't readable.   A hot air gun or hair dryer might melt the wax to let you read components and resistor color bands to reverse engineer it.  If you can get high quality photos of each side, and get the part numbers off the components I may be able to help.  It may be possible to figure out a suitable value once the circuit diagram is completed.

That said, I also agree that the mechanical brush/slip ring issues should first be addressed.  Checking the excitation coil resistance via ohmeter, with and without brushes can be a useful diagnostic. 

A brushed alternator is as simple as they get, and is a good way to learn.  Once you get it sorted out, you'll be able to keep her running in the future without breaking a sweat.

Best Wishes,
Bruce

Seems you've done some good research and have a good plan, John.  Let me know if you need a hand.

One issue I've run into with ST generator heads is that the waveform distortions are bad enough to confuse most inexpensive (edit) non-RMS voltage/frequency monitors, particularly on AC frequency.  I solved that by switching to a Siemens RMS voltage monitoring relay, which has been problem free.  I recall that Siemens also makes a voltage monitor that runs on 24VDC, but can monitor 230VAC.  They are spendy but are available in a wide range of models.  So that's a possibility to look into if you have difficulty with the one you found.

Very helpful info, John.  I'm surprised that the minimal power of a voltage and power monitoring circuit would keep the SOM running once the larger load is removed.  Yes, sensing on the 24V side might be a good solution to avoid messing with the SOM start circuit.  There are DC voltage controlled relays which could then do the job. 

If you can connect on the DC generator side so that the relay is only powered when the generator is running that would avoid a load on your batteries.  (Upstream of a diode between DC generator and battery would do this.)  I'm not familiar with the SOM 24V charger circuitry so if you have a schematic/wiring diagram I can look at, that would be helpful.  Ideally we'd like 24VDC to power your overspeed sensing that is only present when the engine is running.

You'll have to test and monitor via VOM to see if DC sensing will reliably detect engine overspeed, once you've repaired your charge current limiting resistor.

If a rise in DC volts with overspeed doesn't look viable in testing, then we might look for an AC over frequency sensing relay powered by 24VDC to avoid the SOM start circuit troubles.  Frequency sensing relays are quite pricey, so I'd look at voltage sensing first. 

AC frequency sensing via Arduino is readily available, and the Arduino 5V could be powered from the 24V side, but hopefully we can avoid that approach.

I'm in central eastern Arizona, USA.

Best Wishes,
Bruce

I have rpm monitoring on my CS, for both over and under speed, with closing of the injection pump rack and lifting of the exhaust valve.  Rpm is sensed by a Cherry magnet/hall effect sensor near the spokes, rack and lifter via air cylinders, control via Picaxe 40X2 and hand wired board with logic level MOSFETs for air solenoid valve control.  Over temperature, oil high/low, and excessive vibration also cause an emergency shut down.

The same can be done today with almost no custom hardware via Arduino plus available interface boards.

The commercial engine rpm limit detectors I can find aren't suited for the CS due to it's low speed.  A Picaxe 08M IC (8 pin dip) could do it easily with a few lines of Basic.  Let me know if you need help.

Best Wishes,
Bruce

I took out the TXV that can't be adjusted without evacuating the system, and put in a needle valve plus the cap tube.
That let me restrict flow as temperatures climb to keep a constant 4.6 amps.

I tried wood then a metal cover for the condenser coil to restrict airflow so I could simulate a hotter day. The leaky stack of lumber gave better results; the metal cover seemed to cause greatly elevated current for the same temperature. Something about too much of the coil not being cooled at all.  The metal cover is shown in the picture.

As liquid line temperature got above about 95F, the BTU perfomance starts to fall off rapidly.  I'm able to get 13.6K BTU's at 90F,  13.0K at 94F, 10K BTUs at 102F.   Then it falls on it's face, and there's no chance of it not being adjusted properly, since I was doing it manually while watching compressor current, every so slowly.

So evaporative precooling is a must; my former LG window unit doesn't have the advertised 12000 BTUs, EER of 11.3 at 95F ambient at my elevation, at least.  (95F would be 110F liquid line temp at 15F approach).  I'm not getting anywhere near this performance at 95F equivalent liquid line temps.  More like 4000 BTUs.

I can likely live with the needle valve/cap tube combo, but could get the best cooling performance with an electronic expansion valve and custom microcontroller programmed to adjust for max rated amps.  They are cheap, but no one but me could maintain it or appreciate it. 

Sporlan is the highest priced TXV maker in the world, as far as I can tell.  Too rich for my low budget experiment with  higher suction temperature,  BPHE water cooler, I think. 

With a bigger budget I could nail down this floor cooling thing quite nicely; a bigger condenser coil with much better approach temp, slightly bigger compressor, increase the in floor water flow rate to 4 GPM via Lang D5 Strong pump to allow higher indoor humidity and add an evaporative pre-cooler plus air intake about 6 foot above ground level to reduce air inlet temps another 10F in the afternoon sun.   

I'm having no luck finding an appropriate valve.  Lots of the 1.5 Ton, but not for R410A, 1 Ton.  I expect since all the smaller units have gone to inverter/electronically controlled expansion valve.  I'm visually exhausted, will try more poking about later. 

No packing, the adjustment is deep inside the open cavity via 6mm allen key. Adjustable means they can figure out the turns for different models the hard way;  not ideal for an experimenter.

I wasn't expecting this kind of "adjustable".