Evacuation and Vacuum Decay Verification

Purpose

A micron reading is a snapshot and it can be produced by a system that still holds a leak, still holds water, or simply has a gauge sitting in the deepest part of the vacuum. The decay test is the acceptance, because it is the only part of this procedure that asks the system a question rather than the pump. What this prevents is the callback nobody traces: moisture left in a system reacts with the oil and the refrigerant, forms acid, and takes out a compressor a year later on a job whose paperwork says it was evacuated to 500 microns. It was. The number was true and the system was wet.

Scope

Covers evacuation and decay verification on residential and light commercial refrigerant systems after any opening: line set replacement, component changeout, leak repair, and new installation. Covers pump setup, gauge placement, the decay test and how to read its shape.

Does not cover the nitrogen pressure test that must precede it, which the brazing SOP owns, and does not cover recovery. Does not cover charging, which the charge verification SOP owns and which begins where this one ends.

Roles and responsibilities

Procedure

1. Confirm the system passed its nitrogen pressure test before the pump comes out. Evacuation is not a leak test and a vacuum pump will never pull a leak shut. Acceptance: a documented pressure test hold on this ticket, with start and end ambient temperature. What wrong looks like: going straight from brazing to evacuation because the joints looked good, which converts a small leak into a two-hour pull-down that never reaches target and a tech who blames the pump. Stop rule: no documented pressure test, no evacuation. Hazard in this step: confirm the system is at atmospheric before removing any valve core, because a core removed under residual nitrogen pressure leaves at speed and takes an eye with it.

2. Set the pump up so the system is the restriction, not your hoses. Fresh oil to the sight glass, the largest hose diameter the ports will take, hoses as short as practical, valve cores removed with a core removal tool, and an isolation valve between the pump and the system. Acceptance: pump blanked off at its own inlet pulls below your target within a few minutes; if it will not, change the oil and repeat before touching the system. What wrong looks like: quarter-inch charging hoses with cores still in, which can add hours and still stall. Hazard in this step: the pump is a line-cord appliance often set on wet ground beside a condenser, so it goes on a dry surface with a GFCI-protected supply.

3. Put the micron gauge at the point farthest from the pump. Mount it on the opposite service port from the pump connection, on the system side of any isolation valve, never on the manifold and never at the pump inlet. Acceptance: the gauge is reading the system, not the pump's own suction. What wrong looks like: a gauge on the manifold beside the pump showing a beautiful number while the far end of a 40 foot line set is still hundreds of microns higher. That reading is the single most common false pass in this trade. Stop rule: if the only available port is at the pump, say so on the ticket rather than reporting the reading as a system value.

4. Pull down and watch the rate, not just the number. Note the reading at intervals and pay attention to how it is falling. Acceptance: a steady, decelerating descent to your target, which for most manufacturers is 500 microns or lower, with the number in the installation manual governing because that is the figure attached to the warranty. What wrong looks like: a pull-down that stalls at a plateau and stops improving. That is information, not a delay: it means the pump is now removing something as fast as the system is producing it. Hazard in this step: never apply power to a hermetic compressor and never insulation-test its windings while the system is under deep vacuum, because the low pressure lowers the voltage at which the windings will arc.

5. Where moisture is suspected, break the vacuum with dry nitrogen and pull again. Introduce dry nitrogen to a slight positive pressure, let it sit long enough to mix with what is in the system, release it, and re-evacuate. Two breaks and three pull-downs is the common practice on a system that has been open to weather or has a burned-out compressor. Acceptance: each successive pull-down reaches target faster than the last. Hazard in this step: nitrogen only ever leaves a regulated cylinder, never a bare one, and it displaces oxygen in a closet or crawlspace, so ventilate the space. What wrong looks like: breaking the vacuum with shop air, which puts back the water you just spent an hour removing.

6. Reach target, then isolate at the valve nearest the system, not by shutting the pump off. Close the isolation valve and leave the pump running or shut it down after the valve is closed. Acceptance: the isolation happens between the system and the pump, so what you are about to watch is the system alone. What wrong looks like: switching the pump off with the hose open, which lets the pump's own volume and, on some pumps, its oil migrate back toward the system, and gives you a rise that tells you nothing about the equipment. Stop rule: if there is no isolation valve in the setup, install one before the decay test rather than running the test without it.

7. Run the decay test and read its shape, not only its endpoint. Watch the gauge for the hold period the manufacturer specifies, commonly on the order of 10 to 15 minutes for residential equipment. Acceptance: the reading stays below the manufacturer's stated criterion for the full hold. Read the shape: a rise that climbs steadily and never levels is air coming in, which is a leak. A rise that climbs and then holds at a plateau is vapor coming to equilibrium, which is moisture or outgassing, and free liquid water in a system near 70 F has a vapor pressure on the order of 20,000 microns, so a plateau up in the thousands is a moisture signature rather than a leak. Stop rule: a leak signature sends you back to pressure test, a moisture signature sends you back to step 5, and neither one gets charge released into it.

8. Release the charge without losing the vacuum, and record before you move on. Open the system to the refrigerant source or back-seat the service valves with the vacuum still held, then break the hoses. Acceptance: the vacuum was never broken to atmosphere between the passing decay test and the charge entering. Hazard in this step: liquid refrigerant on skin causes a cold burn and the vapor displaces oxygen in a closed mechanical room, so gloves and eye protection go on before any valve moves. What wrong looks like: pulling hoses first and reconnecting to charge, which admits room air with its moisture and quietly undoes the whole procedure.

The record this produces

One evacuation block per system: pump size and whether oil was changed; hose size and whether cores were removed; where the micron gauge was mounted; the pull-down curve as at least three timed readings; number of nitrogen breaks if any; the isolation method; and the decay test as four values, which are the reading at isolation, the reading at the end of the hold, the hold duration, and the manufacturer's criterion it was measured against. Where the test failed, the shape of the rise gets written down in words, because that is the diagnosis the next attempt starts from.

The reader who matters most is the one handling a compressor failure eighteen months later. A manufacturer assessing an acid-failed compressor will ask what the evacuation looked like, and "held below 500 microns for 15 minutes, isolated at the valve, gauge at the evaporator" is a defensible answer where "evacuated to 500" is not. The second reader is the next tech on the system, who can tell from a recorded plateau whether this system has a history of holding water.

Worked pass: line set replacement, evaporator coil and outdoor unit, R-410A

New line set brazed under nitrogen the same morning, nitrogen pressure test held and documented, so step 1 passed on paper rather than on memory. Pump blanked off reached well below target in under two minutes with fresh oil. Cores pulled at both service ports, half-inch hose from the pump to the outdoor low side, micron gauge on the evaporator side, which is the far end.

First pull-down: 3,400 microns at 20 minutes, 2,100 at 40, and then it sat. At 65 minutes it was still reading 1,800 and had not moved in a quarter of an hour. Under step 4 that is a stall, not a slow system.

Step 7 FAILED on the first attempt. The isolation valve was closed at 1,800 microns and the reading climbed steadily: past 3,000, past 5,000, still climbing at 12,000 when the test was called at eight minutes, with no sign of a plateau anywhere. Steady, non-levelling rise is the leak signature, and this one had no plateau at all, so it was not read as moisture. Under the stop rule that sends the job back to a pressure test rather than to a third pull-down.

Re-pressurized with dry nitrogen to the plate's stamped test pressure, and a bubble solution found it in about four minutes: the flare nut at the outdoor low-side service valve, tight to the hand but not to the wrench, from the changeout that morning. The joint was broken down, the flare re-checked for a crack, remade and re-torqued, and the pressure test re-run and held.

Second evacuation reached 350 microns in 38 minutes, faster than the first attempt got to twice that figure, which is itself consistent with the leak having been the whole story. Isolated at the valve, the reading rose to 420 microns over 15 minutes and levelled there for the last five. That is below the 500 micron criterion in that manufacturer's installation manual for the full hold, so it passed, and the small levelling rise reads as normal outgassing rather than water: a wet system would have kept climbing toward a plateau far higher, in the thousands. Charge was released with the vacuum still held.

What the failure teaches: the first pull-down produced a number a tech under pressure could have accepted. 1,800 microns is not a passing figure at any manufacturer, but a stalled gauge invites the belief that the pump or the hoses are the problem, and the reflex is to change oil and pull longer. The decay test settled it in eight minutes and pointed at the right half of the system, and the shape of the rise, not its size, is what separated a leak from water.

References

• Manufacturer installation instructions for the target micron level, the hold duration and the acceptance criterion, which govern because they are what the warranty is written against
• 40 CFR Part 82, Subpart F, for recovery requirements before the system was opened
• AHRI and trade-standard practice for deep vacuum technique: gauge remote from the pump, cores removed, isolation valve for the decay test
• See related: Brazing With a Nitrogen Purge Standard; Charge Verification by Weight and Superheat