Belt tension is OK so I ordered a new alternator – ALU-8521LS (no pulley – we can move it across).
One thing I forgot to check is the field current drawn with the engine off. That should show up directly on the ammeter. But anyway the voltage reg does produce the right bus voltage so it should be driving the field flat out at lower voltages, so there isn’t much scope for a problem there.
I doubt it is the Lamar B-00368-5 voltage reg but it turns out very few people stock it, and the price is eye watering considering the $10 manufacturing cost 
Are you changing the alternator and regulator ?
No, just the alternator. I can’t even get the regulator – well not quickly.
Problem solved.
It was subtly mis-wired, with the net result that the voltage available across the field winding was about 1/3 of what it should be.
So I have a spare alternator. Actually it is a better one than my ~20 year old one – it has a hex key hole in the shaft, making it much easier to install and remove the pulley.
How does one subtly mis-wire an alternator?
wigglyamp wrote:
Try starting the engine with the alternator selected ON, just as you’d do in a PA28
You would? I always start a PA28 with the alternator OFF.
What is the rationale behind this sequence?
I would think it is to protect other aircraft loads from overvoltage and uneven bus voltage, during engine starting, and similarly during engine shutdown.
If you start the engine with the ALT CB in (i.e. the field current enabled) what will happen is that the voltage regulator will go flat-out on the field current as soon as it is able to, which on a 28V system will happen at something vaguely like 5-10V (depending on what loads appear and start drawing current) and any uneven alternator RPM will get directly translated into what comes out on the bus. With a bit of smoothing-out from the battery which is hopefully still connected and not dried-out
I am not sure one can trust the primitive voltage regulators to behave properly at low voltages.
The other reason for doing it this way is to be able to eyeball the voltmeter, which with the field current at zero should be reading 12V or 24V.
In planes which have an avionics master switch, one should not blow anything up anyway because that switch should not be ON until everything is running and the bus voltage has been checked. But loads of simpler planes don’t have one. I don’t think any of the ones I trained in had one, which makes it a lot easier to blow up the avionics.
Any other reasons for it?
BTW the overhauled alternator runs great. It starts to charge the battery at 1000rpm, whereas the previous one did so at 1100rpm and that was the whole time 2002-2017. I wonder what they did in the overhaul that made the alternator significantly better like this? The voltage regulator has not been touched.
To add: the wiring error was the F2 terminal going to AUX instead of going to ground. This is really weird. I can’t find an internal wiring diagram for this exact alternator. The classic 3 phase alternator is like this
where the regulator drives F1 (with a transistor between +28V and F1). F2 is grounded.
However most in GA also have an AUX terminal. I can’t find a decent circuit right now but there are two variations in automotive alternators. In some, AUX is the junction of the three coils (no-connect in the above diagram) while on others it is another output. Anyway, the issue manifested itself in the most bizzare way, although the alternator output was regulated OK during normal engine RPM.
