The Changeout That Would Not Pull Below Two Thousand Microns

The stall

Day two of a 3-ton R-410A split system changeout: new condensing unit, new cased coil, the existing line set reused after a flush-free check showed it was the right size and in good shape. The crew had cut the old equipment out the afternoon before, brazed the new connections first thing that morning, run a standing nitrogen test, and put the vacuum pump on at 9:30. At 10:45 the micron gauge was sitting at 1,900 to 2,000 microns and had not moved in 40 minutes. The installer called the lead and asked whether he could "just open it up, it's close enough."

It was not close enough, and the reason it was stuck mattered more than the number. A vacuum that stalls is the system telling you something is boiling off faster than the pump can remove it, or something is leaking in faster than the pump can keep up, or the rig itself cannot get any lower. Those three call for completely different fixes, and opening the valves fixes none of them.

Answer one question at a time

The lead drove out and worked the three candidates in the order that costs least to rule out. See related: Just Pulled Vacuum But It Won't Hold: Equipment Vs Leak Decision Tree, which this case follows in its first two moves.

Is it the rig? The cheapest question first, because a pump with wet oil or a hose that leaks at its gasket produces exactly this stall on a perfectly dry, tight system. The lead valved off the core removal tools so the system was isolated, then put the micron gauge directly on the pump's inlet with the pump running. It pulled to 45 microns in under two minutes on fresh oil changed that morning. The hoses were short, large-diameter vacuum hoses with no core depressors in them, and the gauge sat on the core tool's side port, at the system rather than at the pump. A rig that can pull 45 microns blanked off is not what is holding a system at 2,000. Eliminated. See related: Vacuum Rig Layout for a Fast Clean Evacuation.

Is it a leak? The standing nitrogen test that morning had held with no drop on the gauge over its 30 minutes, with the line set temperature steady, and every new joint had soaped clean. That makes a leak unlikely but not impossible, because a nitrogen test proves the system holds pressure from inside, and a few leaks show more readily under vacuum. The way to tell is the shape of the rise after isolation, so the lead reopened the core tools, let the pump bring the system back down to its stall, then isolated the system from the pump at the core tools and watched the gauge.

The decay log

MICRON LOG   job 4402   gauge at suction core tool
pump isolated at 11:52, system valved off

time    microns   change in the last interval
11:52    1,900    isolated
11:53    2,700    +800
11:54    3,400    +700
11:55    3,900    +500
11:57    4,300    +400 in 2 min
12:00    4,500    +200 in 3 min
12:05    4,600    +100 in 5 min
12:10    4,600    0    leveled
line set in crawlspace, about 45 F

Reading it. The rise was fast at first and slowed every minute until it stopped near 4,600 microns. A leak does not do that. Air leaking into a vacuum through a fixed hole enters at a nearly constant rate, because the outside pressure barely changes as the inside goes from 2,000 microns to 5,000. So a leak curve keeps climbing in a roughly straight line, and it would not level off far short of atmospheric. This curve leveled.

A curve that rises and flattens is moisture. Liquid water in the system boils off under vacuum until the pressure in the system reaches the vapor pressure of that water, then it stops boiling and the gauge stops rising. The plateau tells you how warm the water is. Free water at 32 F has a vapor pressure of about 4,600 microns. The line set ran through a 45 F crawlspace, and water boiling under vacuum cools itself as it evaporates, so a pocket of water sitting near freezing in a cold crawlspace is exactly what levels a gauge at 4,600. That also explains the stall during pumping: the pump was pulling vapor off a pocket of water that kept boiling, cooling, and in places probably freezing, which slows the boiling further and drags the evacuation out for hours.

The leak was eliminated not by the nitrogen test alone but by the curve agreeing with it.

Where the water came from

The record. The crew had cut the old condensing unit and coil out at 3 pm the day before and left for the night. Heavy rain came through overnight. The line set ends at the condenser pad, outdoors, sat uncapped and untaped from the cut until the braze at 7:30 the next morning, and the suction line ran down the foundation wall into the crawlspace through a low dip before it rose to the indoor coil.

The confirmation. With the system back at atmospheric with dry nitrogen, the lead loosened the core tool on the suction line and blew nitrogen through from the liquid side at a low flow. Water spat out of the suction port in drops for several seconds before the stream ran dry. Rainwater had run down the open suction line into its low dip and been sealed in by the braze.

The oil makes it worse. The old R-410A system had left a film of POE oil coating the inside of the reused lines. POE is hygroscopic: it takes up water from wet air and from liquid it touches, and it holds onto it far harder than mineral oil does. A vacuum pulls free water out readily once it is warm enough to boil; water dissolved in POE comes out slowly and keeps a system wet long after the free water is gone. And POE that holds water hydrolyzes with heat into acids, which attack motor insulation and form sludge.

That is the cost of "close enough". Had the installer opened the service valves at 2,000 microns, the condenser's factory charge would have mixed with the remaining water and the wet oil film, and the new compressor's own POE would have started taking up that water on the first run. The drier would have saturated in days, and the system would have carried moisture and acid into a compressor that was new that morning. See related: Compressor Oils: POE, PVE and Mineral on the Service Call.

The fix

Break vacuum with dry nitrogen, never with air. The system went back to a slight positive pressure with nitrogen through the regulator. Pulling harder on a cold pocket of water only freezes it, and ice under vacuum takes far longer to leave than liquid water does.

Nitrogen sweeps to move the liquid out. With the suction core removed at the outdoor service port and the outdoor service valves still closed so nothing reached the compressor, the lead flowed nitrogen through the line set from the liquid side in short bursts at a higher flow than a brazing purge, letting each burst push water ahead of it toward the open port, and repeated until three bursts in a row came out dry. The discharge was pointed away from everyone, and both techs wore eye protection, because a sweep throws water, oil and debris out of an open line at speed. The line ends were held by hand at the discharge so they could not whip, and the regulator, never the cylinder valve, set the flow.

Replace the drier. The liquid line drier brazed in that morning had spent its first hours in a wet system under vacuum and had taken up water the sweeps could not reach. It was cut out, not heated off, and the new drier was brazed in under a nitrogen purge, following How to Replace a Liquid Line Filter Drier. The joint was soaped under nitrogen pressure before evacuation started again.

Evacuate again, with nitrogen breaks. Evacuated to about 1,500 microns, broken back to a few psig with dry nitrogen and allowed to sit for 10 minutes so the dry gas could absorb moisture from the oil film, then evacuated again. The second pull reached 900 microns in less than half the time the first stall had taken. One more break and pull brought the gauge to 420 microns.

The decay that counts. System isolated at the core tools at 420 microns. Over 10 minutes the gauge rose to 470 and leveled, staying under the 500 micron decay limit this maker and this shop use. That passes the decay rule in Evacuation and Vacuum Decay Verification: a small rise that flattens well under the limit is the last trace of moisture in the oil film, not a wet system and not a leak.

Open and charge. Core tools replaced cores, service valves opened, and the charge set by subcooling after the line-length adder was weighed in, logged per the shop's refrigerant record SOP.

The comparison that settled the argument

The installer's position, that 2,000 microns was nearly there, rested on reading the vacuum as a distance: 2,000 is a lot closer to 500 than atmospheric is. The decay log is what changed his mind. At 2,000 microns the isolated system rose to 4,600 and stopped. After the sweeps, drier and nitrogen breaks, the isolated system started at 420 and stopped at 470. The first system had a pocket of liquid water in it; the second did not. No distance between the two numbers on the gauge says that, but the shape of each curve does.

What changed on the crew's checklist

Cap or tape every line end the moment it is cut, and keep the ends pointed down until they are, so a reused line set never sits open to weather. One night with the ends open cost most of a second day on site.

A stall gets an isolated decay before any decision. The checklist used to say "evacuate to 500 microns". It now says what to do when the gauge stops: isolate, log the rise every minute for 10 minutes, and call the lead with the log. A leveling curve goes to moisture removal; a straight climb goes to leak search; a fast pull blanked off rules out the rig.

Nobody opens the valves on a reading above the decay limit, whatever the time of day. Opening on a wet system turns a two-hour delay into a compressor claim.

References

• See related: Just Pulled Vacuum But It Won't Hold: Equipment Vs Leak Decision Tree
• See related: Evacuation and Vacuum Decay Verification
• See related: How to Replace a Liquid Line Filter Drier
• See related: Vacuum Rig Layout for a Fast Clean Evacuation
• Equipment manufacturer's installation instructions for evacuation target and decay limit