It is time for an update on the ammeter problem.  For the record, the ammeter should not be jumping around excessively and Alan is right about the problem being resistance between the battery and the voltage regulator input.  While I don't agree with Alan's specific understanding of how the resistance causes the problem it is generally correct.  

I would like to say I found what was causing the problem and have been able to correct it, but I can not.  However, the jumping is significantly reduced, I just wish I knew why!

Since the plane was apart for the annual in April, I used that opportunity to undertake some troubleshooting.  I was looking to determine the resistance from the battery to the input of the voltage regulator with the intent of finding a connection or component that was abnormal, whatever that means!  You can measure resistance with a good ohmmeter, I have one of those, but it is difficult at best because the resistance of the measuring system can be significant when trying to measure resistance less than 1 ohm.  A better way is to measure voltage drop with a known current flowing through the system.  Since I was looking for an abnormal resistance, the exact value of the current flow wasn't important.  It only had to be repeatable.  That is easily obtained by simply turning on the maser switch.  If you don't leave the switch on to long then the battery voltage doesn't change much and you have a repeatable current.

The first step was to see what the voltage drop was between the positive terminal of the battery and the input to the voltage regulator in the tail.  That would give us, a co-owner and myself, a general idea of the situation. 

A note here.  These kind of measurements are not made without a lot of care and good knowledge of what you are doing.  Our aircraft batteries are relatively low voltage, 12 volts, but are capable of many amps of current.  An inadvertent short to ground can easily damage an expensive and hard to replace component and will make you jump and probably hurt yourself on some air frame part!  Our approach was to fix the positive connection of the voltmeter to the battery positive terminal and then with the master switch off to secure the negative lead of the voltmeter to the desired measurement point.  With everything secure and checked so there were no inadvertent shorts to ground we turned on the master switch and made the measurement after a couple of seconds to let transients settle out.  After the measurement was made the master was turned off.

A nominal voltage drop of 0.7 volts was measured between the positive terminal of the battery and the output of the over voltage relay in the tail.  That was as close as we could readily get to the input of the voltage regulator. The ammeter showed a nominal current of 4 amps so that would be an approximate resistance of 0.7 divided by 4 or 0.175 ohms.  Not a lot!

We next measured the voltage drop across the master relay.  It was less than 0.001 volt.  The master relay was certainly not the source of the voltage drop.

Next in line after the master relay is the aircraft ammeter itself.  Unfortunately, we could only get to one side of the ammeter and that showed an insignificant drop compared to 0.7 volts.  The ammeter could still have been the source because we couldn't figure out if we were measuring on the battery or the bus side.

The bus side of the ammeter goes to the fuse panel by the pilot's knees.  We could get to one side of the master switch without any further disassembly.  That would be the next component coming from the voltage regulator side.  The drop there was 0.7 volts.  We couldn't tell if it was the side leading to the voltage regulator or the side connected to the alternator breaker switch followed by the 5 amp field circuit breaker. I didn't realize it at the time, but the 85 amp circuit breaker in the alternator output incorporates a switch that is placed in line with the alternator field circuit.  If the 85 amp alternator breaker is tripped, the alternator field circuit is opened as well.

Having measured every circuit point that we could easily get to it was time to take something apart so to be able to isolate the 0.7 volt drop.  We knew it wasn't the wiring from the battery to the ammeter and we knew that it wasn't the wiring from the voltage regulator to the master switch.  It could be the master switch itself, the ammeter, the 85 amp circuit breaker switch, the 5 amp alternator field circuit breaker or the wiring between any of those components.

The main fuse panel isn't particularly difficult to open up if you don't mind working upside down in cramped quarters in a dark spot. Mine is held in place by only three screws, it is supposed to be four screws but there is no way to get to the fourth screw with the upholstery side panel in place.  That fourth screw hasn't been there since the first time the fuse panel had to be opened up!  Even with the three screws removed the panel doesn't drop down easily.  The outside mounting bracket has to be pushed aside slightly to allow the fuse panel to rotate downward so that it can slide clear of the upholstery side panel.  Then there is a plastic cover that shields the connection ends of the breakers from the aircraft wiring above the panel that has to be removed to get to the actual breaker connections.  With that all done we were ready to proceed.

First step was to measure the voltage drop to the other side of the master switch.  It measured 0.4 volts!  I though we had found a significant portion of the drop, the master switch.  That lasted until we measured the other side of the master switch which was also 0.4 volts.  Somehow, lowering the fuse panel had eliminated 0.3 volts of the voltage drop, nearly half of it.  We measured the rest of the circuit behind the fuse panel with 0.4 volts being the nominal value.  I guess the drop across the ammeter is 0.4 volts at a nominal 4 amps of current.  Sort of makes sense.

We check all the connections and a couple of solder joints.  All were tight and certainly looked good.  I suppose we could have taken all of the relevant connections apart and put them back together but we hadn't really moved anything other than the fuse panel.  The wires to it are tightly bundled so it seemed unlikely that we moved or changed the stress on any of the fuse connections.  But something changed and we were unable to get it to change back.

We were already at this project for several hours with no clear way forward.  So, we put it back together.  The drop was still 0.4 volts when we were done.

It was a few days later that the aircraft engine was able to be started after the annual inspection.  At first glance it seemed the ammeter was more steady but hard to say for certain.  Now after about 25 hours of flight time I will say the ammeter reading is generally more steady than it was especially with the engine at idle.  It is very steady while in flight.  I believe that is expected because we affected a significant reduction in the voltage drop, i.e. resistance, between the voltage regulator input and the battery.  If we only knew specifically what caused it I would feel the effort was a total success.

This isn't my first experience with intermittent electrical problems.  Most show themselves again at some future point.  When that happens, at least I know where to go first!

Bob:

That is a very thorough and clear description of a huge amount of effort.  I have lain on my back under that exact spot for a substantial amount of time .. you may have understated the amount of work involved.

Troubleshooting an erratic connection is a miserable business.  There is also the potential for the ammeter itself to be a bit worn and unsteady all on its own.  Certainly the voltage drop going from 0.7 to 0.4 after disturbing a breaker panel is a strong clue.

If it was a generator I might be inclined to suggest that the current actually was fluctuating as the method of control (shunting a field resistor in and out) lends itself to that.  But an alternator is a different animal.

I would be tempted to focus on the gauge for a bit.  Substitute with a known good one for a while and see what occurs - if that is even possible.

A digital voltmeter is relatively easy to install if you have a power outlet and it might provide some clues as well.  If the ammeter is fluctuating but the voltmeter is rock steady you might have more reason to suspect the gauge wiring.

Tracing the loose or erratic connection might be possible by installing a known load somewhere with a meter attached across it and then start "wiggling" various wire harnesses until you can track it down.  But that depends on being able to recreate the circumstances reliably in a manner that causes the meter to jump.

Very interested to hear how it all comes out.

William

William,

I appreciate your understanding of the difficulty in isolating just about anything that is intermittent!

I had previously verified that the bus voltage was jumping around a few tenths of a volt probably in synchronization with the ammeter's random movement.  That was with a Fluke digital voltmeter so it is not very quantitative.  I assume it is mostly steady now but I have not verified that.

I am waiting for it to return to its former unsteady state.  I am also thinking about it a lot while staring at the wiring diagram.  I have been assuming the issue was caused by excessive resistance in the path from the 14 volt bus to the voltage regulator input.  I am now wondering if a more likely scenario is simply a connection that is less than perfect and that vibrates randomly between a good state and a not so good state.

I think the next step is to check all of the physical connections in the wiring from the ammeter to the 14 volt bus and those connections on the bus to the 5 amp alternator field breaker.  If I recall correctly, the breakers in the panel are daisy chained together by short copper straps.  A less than perfect connection at any breaker from the point where the wire from the ammeter connects to the bus to the connection for the 5 amp alternator field breaker would cause the problem we were seeing.  When I had it apart, there was nothing obvious wrong.  I couldn't see everything, however, especially the connections to the ammeter.  I think a more careful investigation of all physical connections is needed.

It is hot in Phoenix right now.  Not a great time to be working on the backside on a Comanche instrument panel.  Maybe the issue will stay 'fixed' until the temperatures cool down a bit!

@bob

Very true.  Sometimes the issue can just be a "slightly less torqued down" bolted connection that looks tight, but over time has loosened up enough to allow some oxidization to get in between the metal to metal contact.  Changing temperature conditions can alter the tightness enough to allow vibration to induce random effects somewhere completely different.

Sometimes the only thing for it is to thoroughly dis-assemble, clean, dielectric grease, and reassemble to a specified torque the various suspect connections.

I am quite impressed with these old aircraft - how long they have managed to stay airborne with absolutely no maintenance of some of these electrical joints. 

But these connections do need to be maintained, say every 50 years?

I have to bring up the potential for old insulation to be cracking.  Some of that old white vinyl insulation can get very hard, crack, and allow potential for leakage, especially if it is in the engine compartment.

William

I think it is time to close this one out.  A couple of months ago I spent a couple of hours measuring voltage drops along the circuit path from the alternator back to the voltage regulator.  I had become convinced the most likely cause of the jumpy ammeter was an intermittently poor connection in that path.  My first attempt to find the problem corrected the issue without an understanding of what I did to fix it!  The only wires I moved were those attached to the fuse panel by the pilot's knees.  Without some positive knowledge of the problem and a fix, it will come back eventually.

As I indicated previously, I used the steady state condition of the master switch turned on and the engine not running to create a nominal 4 amp current in the circuit path from the voltage regulator to the alternator field winding.  With that I was able to measure the voltage drops across the connections and wires in that circuit path.  I was looking for any connection that seemed unusual and specifically for a connection whose voltage drop would change if I moved the connected wires.  

I did not find any connection whose voltage drop changed when the connected wires were moved, but I did find a wire that had a larger than expected voltage drop.  The wire from the field switch, part of the 90 Amp main alternator breaker, to the master switch had a 150 mVolt drop as I recall.  I can't find my notes for specific results.  That seemed like a rather large drop for a #20 wire that is approximately a foot long with a connection on either end.  The end connected to the switch is soldered and the joint looked good.  The other end is a crimp on terminal that is screw mounted to the master switch.  I did not have my soldering iron with me so I elected to replace the crimp terminal.  I probably should replace the entire wire!

The terminal replacement reduced the voltage drop by half, to 80 mVolt.  That is still more than I would expect across that wire connection but is a substantial improvement and most likely eliminated the source of changing resistance. 

The aircraft has flown another 25 hours since then with the ammeter being very steady and behaving as expected.  I believe this issue is behind me.  I will bring my soldering iron with me the next time I expect to open up that fuse panel so that I can replace that wire.  I don't recall ever finding a bad wire, i.e. high resistance, but I suspect it is possible.  Twelve inches of #20 copper wire should have a nominal voltage drop of 40 mVolt at 4 amps.