A couple of computer fans would move enough air to cool a 6/1 with a modest radiator...

Since rod knock is the result of slop in the bearings being taken up when the rod changes directions, it can usually still be heard when you interrupt fuel or decompress. The fact that you said the sound goes away when you remove fuel/decompress leads me to believe it is IDI diesel knock.  Now if your timing was off it could be more pronounced, but as mentioned would run poorly at low speed and be difficult to start...

Starfire, that was a great description of IDI:)

I don't use a tank at all on the primary loop.  I have my 6/1 thermosiphon thru a small(5" X 12") 10 plate brazed flat plate heat exchanger with 1" ports.   Because of the now low mass in the cooling side of the thermosiphon loop I did have to mount the heat exchanger about 30" above the 6/1 cylinder head but it functions well at all loads.

The primary loop has a 195F thermostat at the cylinder head outlet so the engine rises to and maintains op temp quickly.  The primary loop is unpressurized with a small expansion tank at the top of the system near the heatex. The primary loop has propolene AF(non toxic) and the secondary loop is dom water circulated by a pump to a domestic water tank. 

I put a 120F outboard motor thermostat on the outlet of the heatex secondary to modulate the flow of domestic water so the heatex outputs 120F water. The heatex is rated to 400 PSI so I can run domestic water pressure on the secondary. The water circulates into the tank from the top down and the return to the heatex is from the bottom of the tank.  Once the tank is hot, and hot water starts flowing out of the bottom port, I dissipate the heat from the tank return to the heatex thru a fan-coil unit.  Because of the modular nature of the assembly, I can place the fan-coil unit inside the house near the tank if I want the heat there, or out in the generator shed to dissipate the heat outside...     

The ammount of cooling you need has a direct corrolation to the ammount of fuel you burn.  At idle, not much fuel, not much cooling.  Any load and both these numbers go up.  Basically you need to dissipate about 1/3 the ammount of the BTU content of the fuel burnt. 

An open tank cooler heat dissipation is completely dominated by evaporation.  As an example in 60F still air with 210F coolant the surface area of the water will shed around 3240 BTU per SQ/FT while the tank side will only loose around 360 BTU per SQ/FT.  In 80F still air the evap number drops to around 2700-2800 BTU per SQ/FT.  You could dissipate all the heat you need from a large surface area low volume pan.  The screen cooler is a wonderfull example of this.  The drawback is that it might be a little tricky to get a low volume pan to thermosiphon, so you may need a pump.

My 6/1 at full load(a little over 3KW on the generator) needs to dissipate around 18,000 BTU/HR.  In 80F air, a 6.5SQ/FT pan should be able to handle it.  1/2 load 3.2SQ/FT, 1/3 load 2 SQ/FT ECT...       

Try a different freq counter.  I have had good luck with a Kill-a-watt reading pretty consistently. 

If the belt were slipping the frequency would droop.  SInce the belt is only going so fast, I do not see it ever going above it's fraction of engine RPM so it does indeed look like your frequency meter is reacting to something else in the waveshape.

What kind of alternator is it, What kind of regulation does it use.  If it has a pulse type regulator, as the voltage droops artifacts from the increasing regulater input pulse can show up in the output AC waveform.  Under load, distortions can also be more evident on harmonically excited generators also.

With the mechanical governor on the roid, you should really be setting the no-load frequency to around 62 HZ.  ALL mechanical governors have RPM droop across their load range.  Since the mechanical flyweights are in constant opposition with a spring, They MUST have a decrease in RPM to provide the mechanical movement which increases the throttle to help counter the drop in RPM.  My 6/1 set at 62HZ no-load droops to 58HZ at max load.  This is actually kind of handy, as with a kill-a-watt plugged into any outlet in the house I can, at a glance, tell the load on the generator...  Starting out at 62HZ or so might help with the inverter tripping out for low freq when it places a load on the generator...   

Especially in a pumped system, too much bypass and low load can be a problem.  I maintain a 100KW Generac that wouldn't get up to op temp at 20KW load.  Generac added a bypass line and a little manifold to connect additional temp sensors.  At that load in our cool PNW climate with it's massive radiator, the bypass provided all the coolant flow necessary and the thermostat never would open.  I made up blocking panels to glock off about 3/4 of the radiator to more closely match it's capacity to the generator's load. Now it runs properly at design thermostat temp when powering the site load. 

Well if it is a harmonically excited ST3 the brushes and the diodes are the only "active" components in the system.  The original chinese diode bridge, at least on mine was pretty cheesey.  The generic one he linked to is probably way better quality(i replaced mine with one).  A bridge is also pretty generic and although it may look different will perform the same.  There are two AC inputs and a pos and neg DC output, pretty straight forward.

Excellent point!