Looking for some A/C troubleshooting tips

[dr bob]

DUFO is a universal mnemonic for "doesn't work anymore". From some industrial times when coarser language was common, it would leave folks trying to fit the F into a phrase.

That connector Tom pictures is the same one that's used in the V8 car, and is subject to all kinds of interesting stress and flex in service. On the glove side of the connector in Tom's picture, for instance, you can see where the insulation has been pulled back, and some now-brown copper wire is exposed. This is really common stuff in our well-used cars. I'm all for redoing connections like that, getting back to not-so-oxidized copper and using a crimp-and-shrink terminal with an added layer of adhesive-lined shrink sleeve going back just far enough to be able to grab and pull on it without yanking on the wire insulation. On my car I went through the trouble of rotating the connector sleeve so the wires pass back around it, then a simple zip tie holds the wires and the connector to the harness. Full support, no strain, no flex at the connectors.

The point though is that you are looking for poor connections, a corroded wire, etc., adding resistance in the circuit. So more Ohms when you read the coil, and less current flowing (amps) so less ability to get the clutch engaged. Every element (pressure switch, temperature anti-freeze switch, power supply, control head, relay contacts) and every wire and connector is a possible cause of low current flow available. You get to bypass all those things when you connect a battery-plus wire to the clutch coil side of that connector.

Most electrical troubleshooting is an exercise in dividing the possible causes. For this exercise, the test at the coil terminal immediately determines if it's a supply problem or a coil problem. If it's a coil problem, it gets replaced. If it's a supply problem, we reconnect the coil, and start measuring voltage at each component on the way back towards the battery. At some point the low voltage you'd read at the clutch connection increases to battery voltage. And you've identified the resistance or interruption in current flow. I jump first to the middle of that string of circuit elements to cut the possibilities in half, and than go towards the battery or the clutch depending on what I find. Preference is given to measuring at easy places in the process. But you get the idea. Divide-divide-divide until you find the flow blockage.

Resistance is futile. But we are stuck with it.

[Tom]

I actually came up with several amusing possibilities using that f word, but was trying to model good, family-friendly, carpokes-style behavior.

Interesting to hear the 928 uses that same spade-connector anti-disassembly device. Any tips for separating the spade without yanking on the wire? If I ever really need to separate it, I'll probably ditch that plastic cover altogether (and just take my lumps with the Pebble Beach judges) in favor of a more serviceable weatherproof connector like a Weather Pack or even Deutsch connector.

In the meantime, horrified by your comment, I went and looked at the connector wiring (which is relatively new from when I changed the compressor this decade) and am happy to report that picture was misleading. It actually has multiple lawyers of protection -- an outer sleeve the fits loosely over wires, and inner nylon braided sheath (which looks like an exposed copper wire but isn't) and a traditional wire with white insulation under the nylon braiding.

close-up-ac.jpg (117.82 KiB) Viewed 28 times

[dr bob]

Glad to hear you are are a step ahead on the wire protection. To disassemble, I usually reach in with a small hemostat and pull on the exposed copper spade connector where it wraps around the wiring. The only time it comes apart on the 928 is for a timing belt change, so not that often.

The troubleshooting methods are still the same. I can poke a meter probe into that plastic connector to test for voltage there, to see if full battery voltage is making it all the way to the clutch coil when it's supposed to. If it is, and the coil still isn't closing the clutch. I'm looking at the ground side of the coil, and testing resistance through the coil from there to ground. If voltage isn't there, I start with easy stuff like a fuse, then go back through the circuit to find what's blocking current flow.

----

The clutch coil is an "inductive"" load, so stores some electricity in the generated magnetic field. Sort of like how an ignition coil works. When the supply circuit opens for any reason, the stored field collapses and sends a pulse of current back up the wire towards whatever interrupted the flow. It can arc at that opening contact. For a switch or relay coil, the contact arcing risks building a tiny blob of weld metal, and eventually it's enough to resist the flow of current to the coil. Hence the voltage testing at each element in the circuit, looking perhaps for the one that's suffering from the contact blobs.

----

Shortly after purchase almost 30 years ago now, my then SoCal "needs AC all year" 928 started suffering from intermittent clutch operation, thanks to 'the blob' in a control head relay. The relay contacts were spec'd for AC current, and under-spec'd on amps regardless. The circuit design was a compromise from the drawing board, and suffered in service apparently. I was able to tuck a much larger relay into the console, glued to the outside of the control head, and hide the symptoms. The design cure moves the relay to be the last element on the way to the compressor, plus adds new suppression diodes where the ones from the factory had failed. Somewhere I have the full documentation for the modification I made, with some troubleshooting instructions included. It is very late-928-specific though. The whole HVAC package was designed by Behr, and they should have known better.

[Tom]

Glad to hear you are are a step ahead on the wire protection. To disassemble, I usually reach in with a small hemostat and pull on the exposed copper spade connector where it wraps around the wiring. The only time it comes apart on the 928 is for a timing belt change, so not that often.

The troubleshooting methods are still the same. I can poke a meter probe into that plastic connector to test for voltage there, to see if full battery voltage is making it all the way to the clutch coil when it's supposed to. If it is, and the coil still isn't closing the clutch. I'm looking at the ground side of the coil, and testing resistance through the coil from there to ground. If voltage isn't there, I start with easy stuff like a fuse, then go back through the circuit to find what's blocking current flow.

----

The clutch coil is an "inductive"" load, so stores some electricity in the generated magnetic field. Sort of like how an ignition coil works. When the supply circuit opens for any reason, the stored field collapses and sends a pulse of current back up the wire towards whatever interrupted the flow. It can arc at that opening contact. For a switch or relay coil, the contact arcing risks building a tiny blob of weld metal, and eventually it's enough to resist the flow of current to the coil. Hence the voltage testing at each element in the circuit, looking perhaps for the one that's suffering from the contact blobs.

----

Shortly after purchase almost 30 years ago now, my then SoCal "needs AC all year" 928 started suffering from intermittent clutch operation, thanks to 'the blob' in a control head relay. The relay contacts were spec'd for AC current, and under-spec'd on amps regardless. The circuit design was a compromise from the drawing board, and suffered in service apparently. I was able to tuck a much larger relay into the console, glued to the outside of the control head, and hide the symptoms. The design cure moves the relay to be the last element on the way to the compressor, plus adds new suppression diodes where the ones from the factory had failed. Somewhere I have the full documentation for the modification I made, with some troubleshooting instructions included. It is very late-928-specific though. The whole HVAC package was designed by Behr, and they should have known better.

I'm elevendy-two percent sure the 944 climate control box has a flyback diode in place for exactly that reason, but I'd need to look again to make sure I'm not thinking of something else (like the onboard defrost relay).

As for poking into the connected spade, I did confirm on my car the clutch coil still shows at 2.7 ohms even when everything is connected. My guess is that is all the coil and that the upstream signal is disconnected or high impedance, but I'd need to pull apart the un-pull-apartable spade to really confirm that.

[notny41]

really good stuff here. still waiting to rearrange vehicles
in the garage so I can get this one up on the lift, so in the meantime I checked a few things which I can't remember if I had checked a while ago...

Checked fuse #17. It's a 7.5A fuse and it was fine. Also pulled relay G17 out and tested ot. there was continuity across 87a and 30 until I energized 85 and 86. then it was open. from the schematic on the relay, that's how it is supposed to operate.

also put a jumper across where 87a and 30 would plug into the fusebox for that relay with the car on and did not hear anything going on with the a/c clutch.

will check resistance on the coil when i can get it up on the lift and can get to the wire/connector.

[dr bob]

<...>
I'm elevendy-two percent sure the 944 climate control box has a flyback diode in place for exactly that reason, but I'd need to look again to make sure I'm not thinking of something else (like the onboard defrost relay).

As for poking into the connected spade, I did confirm on my car the clutch coil still shows at 2.7 ohms even when everything is connected. My guess is that is all the coil and that the upstream signal is disconnected or high impedance, but I'd need to pull apart the un-pull-apartable spade to really confirm that.

Might be getting a little deeper than a casual troubleshooting effort, but...

The 928 includes a few of those diodes in a plug-in "suppressor" module that lives in a socket in the CE panel. The modules are NLA and were insufficient from the first day with 1A 100PIV diodes installed. There's a similarly under-spec'd suppressor cap in there too. At least they connected it in almost the right place, at the load end of the anti-freeze switch contacts. The pressure safety switch is in between that and the clutch coil, but since that's not a cycling contact I can forgive the designers.

----

Back to the troubleshooting stuff --

Following Tom's example a little, getting a good reading on coil current at that spade connector means the next logical step would be to turn the clutch circuit on, and measure the voltage at that same connector. If you read close to battery voltage there and the clutch is not engaging, the circuit is good and clutch is bad.

If you see significantly less than battery voltage, move to the pressure switch contacts and test there with wires still connected. Voltage on one side but not the other suggests the switch is open. Voltage on both sides offers a clue that there's a fault in the wire between the pressure switch and the clutch. Low or no voltage on both side tells us the problem is further upstream in the circuit. The three main elements are typically the relay, the pressure switch, and the anti-freeze switch on the way to the clutch coil. But you want the current flow diagram to confirm.

----

A predominant fraction of 928 clinic cars came in with low system pressure, meaning that refrigerant has leaked out. Static pressure in the system is related to ambient temperature, and a lot of the refrigerant has to escape before you see a large drop in a gauge reading. Testing the pressure switch static with an ohm meter may not tell you the pressure is too low, based on the physical nature of the switch internals. Testing it for voltage with full coil current load present is a bit more useful. The definitive test has the gauges connected. The pressure safety switch is in the high-pressure plumbing, so is closed until the system is almost completely empty of refrigerant. "Testing" by depressing a service port valve isn't as useful as we might think; more test for totally-empty vs. not totally-empty.

----

For converted cars -- Later versions that were delivered with R-134a from the factory use a dual pressure safety switch. It opens on too-high as well as too-low pressure in the high side plumbing. R-134a has an interesting spike in its temperature/pressure curve that corresponds to about >110ºF ambient conditions and a high load on the system. It can also happen if the system is even marginally over-charged, if there's a cooling fan failure, etc.. If the condenser isn't cooled, the spike will risk the compressor. The receiver drier has overpressure blow-out plugs intended to protect the compressor, but Fig Newton's Fifth Law of Rankine Reciprocity suggests there's a much greater chance of the compressor fragging first, protecting those drier plugs. The later switch screws onto the same port, but takes a small wiring connector modification to support the extra contact terminals. A very worthwhile modification for converted cars. There may be other similar switches that use just two terminals and have the contacts in series internally. Worth shopping for the better system protection.