OK, so it looks (fingers crossed!) they found the culprit. This a/c has an overvoltage sensor. Which in turn apparently (didn’t see it myself) was
a) too sensitive
b) mounted in a position where it could be jostled by applying the hand brake and by turbulence in flight
It as now been moved and better protected. I’ll fly the a/c later in the week, so can hopefully report that all’s well and good.
Again, thanks all for your input, was much appreciated by the maintenance guys!
Thanks all for the input, great to have such a source of knowledge ! I’ll pass your ideas on.
@ Jan: the problem does occur on the ground (I’ve seen it myself on Saturday during the ground run as described in my post), but the mechanics are unable to trigger it. It simply comes and disappears again. I’m sure the a/c is in the shop as I write and am equally sure (not having spoken to anyone at the airport yet), that the dastardly thing behaves perfectly! As you say – the moment someone with a measuring instrument approaches, all the gremlins go into hiding…..
IF the problem can be reproduced on the ground, I am all with you! But reading the opening post, I see that, in spite of serious efforts, this has not been managed in four months’ time. And I am sure you know as well as myself that intermittent problems will cunningly hide as soon as serious analysis equipment comes near them, only to pop up again as soon as they see safe 
NB what’s wrong with the datalogger/datarecorder idea? Few people will have one around, but perhaps they are available for rent?
Jesse, if an amateur may criticise the words of a professional:
Sure you can Jan, we all learn here don’t we? I do at least.
Maybe my description wasn’t clear. I did mean to do a runup on the ground, as it seems that the problems occurs more often than before, and if might be easy to troubleshoot at the ground. The measurement setup I suggest requires also some long wires to connect to the alternator field. Using multiple meters will prevent you to do false measurments, as you will see the problem very easy.
The alternator output should be very stable when in the regulating range. The voltage shouldn’t change with RPM when it is in the regulating range, and their is very little noise. The noise has quite a high frequency as well, so you won’t be able to measure that. The voltage changes which go as fast as the amp indicator would wobble, so that is slow enough to be visible on an analog meter.
Basically it could be anything, from poor contacts about anyware at the battery and the alternator, trough a failed regulator (common), alternator switch (common), alternator field CB (common), alternator CB (main current, also common).
I recorder wouldn’t be much of use for something like this in my opinion. Just doing a run-up with two people and some meters.
This was the failure mode on a Mooney I used to share. The ammeter would start wobbling, and you could reach behind the panel and give the spade terminal a gentle push and the problem would go away.
I have seen this particular fault on Cessna’s before and it has turned out to be a bad alternator on/off switch that is one half of the battery master switch.
Jesse, if an amateur may criticise the words of a professional:
Your point of view sounds very nice and correct, but perhaps not very practical for everybody. Who can and will fly a plane AND keep a critical eye on 3 analog voltmeters? Must the plane only be flown with a second person on board, preferrably one with a bit of insight in electricity/electronics? And while I agree that a digital multimeter is close to worthless in this diagnosis process, it will not be easy to distinguish between vibration from the mysterious phenomenon vs. vibration from a change in engine regime. &c &c.
The one way to work out your idea reliably would be to install a data logger and analyse its logs when the phenomenon occurs. But who has access to such a device for an extensive period of time?
Admitting the example is extreme, one could even imagine that, by hooking up the voltmeters or the datalogger, one pushes the faulty contact just enough to make the phenomenon disappear – how many megabytes are you going to log before concluding the aeroplane is ok? And what will happen when you remove the test lead? Back to trouble, who knows? Mr. Murphy will do all he can to add to the confusion!
So I see no alternative but to stepwise replace components. Cleaning up/renewing cabling and contacts, as suggested, is an excellent first step, though, and the effort and cost will never be wasted.
I do not agree with the statements previously given.
It will be a seperate regulator, with possible a seperate over voltage protection.
Just start replacing components at random is the most unsenseable option in my opinion. I would use three volt meters (analog prefered), and measure at the SAME time the battery voltage at the bus, the battery voltage at the regulator input and the voltage at the field terminal of the alternator.
A single runup with the setup when it misbehaves will tell you a lot more and points you into the correct direction. As it can be anything, from poor brush contact, bad field wiring, bad regulator, bad over voltage protection, bad alternator switch, bad battery contact.
Troubleshooting will save you money in the long run, especially with intermittend complaints.
You need analog meters as digital meters are slow, and give incorrect readings at fluctating measurements. This is a point a simple analog meter outperforms any DVM. You will need to measure the three voltages at the same time as it seems to be intermittend. Measuring on point at a time could give you false conclusions.
Most problems with alternators are wiring issues, you only have to consider the environment forward of the firewall and you will see that it is very hostile to all the electrical cables.
Having has issues very like the ones described with a C152 all the electrical cables forward of the firewall were replaced on both my C152’s the result has been the end to electrical issues and also given better starting.
If the regulator is separate from the alternator (cheese us, that sounds really prehistoric!) then it is indeed a prime candidate for tentative replacement.
PS any pointer to (the relevant part of) the wiring diagram?
