sub

I've tested as far as I can, had it looked at again by the same shop and the diagnosis was either a bad connection to ECM, or ECM itself. Connections were reset but problem persists. I think at this point I might get a second opinion from the dealership before going down the ECM path. If I'm able to resolve this I'll update.

Ok, good idea. I've already tried swapping relays, no change. I've checked voltage at the relay terminals (85 and 86 as labeled on a relay) Here's what I see: * When tested against a ground screw, I get 14V from one terminal. * When using the other terminal as ground, I get nothing. This is tested with the AC switch on and car running. With the AC on, shouldn't I get 14V when testing across the relay terminals?

Oh shoot, I just noticed that I fat fingered the title of this - it's actually a 2006! Sorry for that. I wonder if I can edit the title....

No, it’s still not working properly. No change with the recharge to 15oz, same behavior as with 9oz.

Thanks for that, I'll see if I can gather anything useful from that. I'm not really convinced it's a wiring issue, just based on behavior, but I don't know the AC systems well enough to say whether things were tested properly by the shop or not. I'll keep poking at it.

EDIT: This is a 2006 Legacy, not 2005. Sorry. Looking for someone to confirm the logic and validity of this diagnosis. My AC will sometimes work, and sometimes not. Sometimes for just a bit, sometimes longer than others. No real rhyme or reason to it that I can tell. Lately it seems to work less and less. For instance, I just went out and started it this morning, and AC worked for about 1 minute blowing cold air, then the air turned warmer, to about ambient outside temp. Had it looked at and they say: * Found A/C clutch engaged once but now it will not come on unless relay is bypassed. * Checked charge in system found it was low * A full charge is 15oz and it only 9oz. Recharged system. * Suspected high pressure switch but after testing it found ground signal is not getting to the compressor clutch relay even if pressure switch is bypassed * Appears to have a wiring short or other issue in the harness, need continued diagnostic time. The suggestion at this point is that the wiring needs to be traced, looking for a short or bad connection. Does the testing methodology that was used seem complete and valid to determine a wiring short? Anything else that should be tested before starting to dig into wiring harnesses? Thanks