Just Pulled Vacuum But It Won't Hold: Equipment Vs Leak Decision Tree

Why this matters

You pulled the system down to 500 microns, valved off the pump, and watched the micron gauge climb. The blower-door tech across the street would call that a leak and start hunting brazed joints. Half the time it is not a leak at all: it is the system outgassing trapped moisture, a tool leaking at a connection, or a pump that never actually got the system as clean as the readout suggested. Misreading a rising micron number sends you replacing a perfectly good coil or chasing a phantom leak for an hour. A disciplined decay test tells you, in five minutes, whether the rise is moisture (boil-off), a tool leak, or a real system leak, and lets you act with confidence instead of guesswork.

Symptom presentation

The classic pattern: the gauge pulls down to target (500 microns or lower), you close the valve isolating the pump, and the reading rises. What it rises to and how it rises is the whole diagnosis. A reading that climbs steadily and never stops points to a leak (air drawn in from outside). A reading that rises and then levels off at a stable plateau points to moisture or trapped refrigerant outgassing inside the closed system. A reading that shoots up instantly to several thousand microns on valve-off points to a gross leak or an open path. The shape of the curve, not the single number, is the tell.

Quick checks

Before condemning the system, eliminate the cheap suspects:

• Tool stack. Core depressors, hose gaskets, and the manifold itself leak more often than the system. A worn Schrader core tool or a cracked hose flare will mimic a leak indefinitely.
• Pump oil. Saturated, milky pump oil cannot reach a deep vacuum. If oil is contaminated, the pump pulls slowly and the readout lies. Fresh oil before a critical evacuation is non-negotiable.
• Gauge placement. Read the micron gauge at the system, not at the pump. A gauge mounted at the pump reads the pump's vacuum, not the system's, and masks restriction across the hoses and core.
• Ambient temperature. Below roughly 50 F, water boils off so slowly the system may never hit target. Cold equipment evacuates poorly and decays misleadingly.

Isolation tree

Run the decay test in defined branches. After hitting 500 microns and valving off the pump, watch the gauge for 5 to 10 minutes.

Branch A: rises and stabilizes below ~1500 microns, then holds flat. This is moisture outgassing. The water boiled into vapor, raised the pressure, then equilibrium stopped the rise. Re-open to the pump, pull back to target, and repeat. Each cycle the plateau should be lower and slower as moisture leaves. Two or three triple-evacuation cycles with dry-nitrogen sweeps between usually clears it. This is not a leak.

Branch B: rises steadily and does not level off (climbs past 2000, 3000, keeps going). This is a leak: air entering from outside. The system never reaches equilibrium because the source is unlimited. Proceed to leak isolation below.

Branch C: jumps instantly to 5000+ on valve-off. Gross leak or an open valve/port. Confirm all service valves and access ports are seated. A back-seated service valve, an open low-side port, or a missing core is the usual cause.

Branch D: never reaches 500 microns at all. Restriction or a saturated pump, not necessarily a leak. Check hose diameter (short, large-bore hoses pull faster), remove the core for the evacuation, and verify pump oil. A restricted path looks like a slow pull but tests clean on decay once you reach target.

To separate tool leak from system leak in Branch B: valve off the system at the manifold so only the hoses and pump side hold vacuum, and watch decay on that isolated section. If the isolated tool stack decays the same way, the leak is in your equipment, not the system. This single step saves the most wasted time.

Confirming diagnosis

Once the decay test points to a real system leak, confirm before opening anything:

• Standing vacuum hold. A system that holds below 500 microns for 15 minutes with no measurable rise is tight. This is the accepted pass criterion for a clean evacuation.
• Nitrogen pressure test. For a confirmed leak, pressurize with dry nitrogen to the equipment's test pressure (per nameplate and AHRI guidance), apply electronic detector and bubble solution to brazed joints, flare nuts, and Schrader ports. Pressure decay over time under nitrogen confirms a real leak independent of moisture artifacts.
• Compare curves. Document the micron-vs-time curve. A leak curve is linear and unbounded; a moisture curve is asymptotic. The graph is the proof.

Remediation

• Moisture (Branch A): triple-evacuate with nitrogen sweeps; replace the liquid-line filter-drier if the system was open to atmosphere for an extended period or shows persistent moisture. A fresh drier is cheap insurance against an acid-forming wet system.
• Tool leak (isolated stack): replace hose gaskets, the core depressor tool, and check the manifold valves. Retest the isolated stack to confirm before re-attaching to the system.
• System leak (Branch B/C): locate via nitrogen pressure test, repair the joint (re-braze under flowing nitrogen to prevent oxidation scale), re-pressure-test, then re-evacuate to 500 microns with a clean standing-hold.
• Restriction (Branch D): remove Schrader cores during evacuation, use short large-bore hoses, and verify pump oil and capacity for the system volume.

References

• EPA Section 608 Technician Certification requirements (refrigerant recovery and handling)
• AHRI Standard 700, Specifications for Refrigerants (purity and moisture limits)
• ACCA / ASHRAE refrigerant evacuation best practices (triple-evacuation, micron targets, standing-vacuum acceptance criteria)
• Equipment manufacturer installation instructions for nameplate evacuation target and nitrogen pressure-test values