Vacuum Decay Test & Evacuation Criteria

No, and this is the single most important thing to know about this test. It is the most repeated number in the trade, and the manufacturers' own documents do not support it as one criterion. Trane's Installer's Guide asks you to pull to 350 microns first and then passes the system only if it does not rise above 500 microns in one minute. Goodman asks you to hold for ten minutes and calls the system leak-free under 1000, with 2000 as the line where a leak is present. Carrier states that a tight dry system will hold 1000 microns after approximately 7 minutes. Rheem asks for 500 or less held fifteen minutes. Daikin's VRV commissioning guide asks for a full hour. York gives no time at all. Daikin's residential multi-zone manual does not use microns for the hold at all: it asks you to watch a compound gauge needle for 4 to 5 minutes and confirm it does not swing back. That is 7 criteria covering Amana, Carrier, Daikin, Goodman, Rheem, Ruud, Trane, York. A calculator that hard-codes 500 as the pass mark fails a system Carrier's own manual calls tight and dry. A calculator that has heard "2000 means a leak" does the opposite, and passes a Goodman system at 1800 microns that Goodman sends back to re-evacuate. Pick your equipment above and read your own.

Not reliably, and this tool will not pretend otherwise. The rule the whole trade teaches is that a steady continuous rise means a leak and a rise that levels off means moisture. York's installation manual publishes it in those words, so it is not folklore. But it has a boundary nobody states. Simulated against a mixture model, 400 trials per setting: against a PURE leak with nothing else going on it finds the leak 27.5 percent of the time at 8 readings over 30 minutes, dropping to 7.2 percent at 5 readings over 15, and only reaching 92.2 percent at 12 readings over a full hour. Against a leak COEXISTING with ordinary outgassing it found the leak in ZERO of 400 trials at every setting. The reason is arithmetic rather than instrument quality: over a finite window a leak's linear ramp and the early part of a saturating exponential are mathematically confounded, so no estimator separates them from one decay series. Two other things break the same rule in the other direction. Refrigerant coming out of the POE oil and nitrogen trapped in a dead leg both push a sound system past the water line, and a genuinely wet system often plateaus far BELOW it, because water boiling into a vacuum cools itself toward ice. So this tool judges your readings against your manufacturer's criterion and names no cause. If you need the leak question answered, a nitrogen pressure test answers it and a decay curve does not.

Because two points define a straight line. Separating a curve that is decelerating toward a plateau from one that is climbing without bound is a question about curvature, curvature is a second difference, and a second difference needs a third point. A tool that takes two readings and names a cause is not measuring the shape at all: it is comparing your level against a threshold and attaching a story to it. That is why 3 is the minimum here. More readings over a longer hold make the answer stronger, and they cost nothing but the time you were already going to spend watching the gauge.

Because water collects at the cold spot and its vapor pressure is set there. The water line on this page is the saturation vapor pressure of water at the system's own temperature, and if a line set runs through a cold crawl space while the condenser sits in the sun, the crawl space governs. Entering a temperature warmer than the true cold spot raises that line and can hide free water underneath it. It is worth a minute to get right.

Then this tool has not read your manual and it will not guess at your number, which is the whole design. It covers Amana, Carrier, Daikin, Goodman, Rheem, Ruud, Trane, York, and coverage is per DOCUMENT rather than per brand: Trane's ducted split criterion and Trane's mini-split criterion are genuinely different procedures, so one does not stand in for the other. For anything unread, open the installation instructions for the unit in front of you, find the evacuation acceptance step, and type its two figures in yourself. The tool then judges your readings against that. What it will never do is fill in a house number and let you assume a manufacturer stands behind it.

It will, once it has the hold criterion for your equipment, either because you picked your unit from the list or because you typed the figures in from your own manual. Without that number it can still tell you a reading is past what water can do at your temperature, which is a comparison against the vapor pressure of water and needs no threshold. What it will not do is invent a pass mark, because a pass is the claim that your system stayed under the figure your manufacturer specifies, and those figures genuinely disagree across manufacturers.

Because both of those numbers rest on assumptions the readings themselves leave open. Turning a rate of rise into a leak size needs the volume of the system, which the tool does not know, and it assumes the isolation is sound, which is one of the things a rise puts in doubt. A figure computed from an assumption its own answer disputes is worse than no figure, because it looks like a measurement. What the tool reports instead is what you measured: where you isolated, how high it went, over how long, and where the water line sits for your temperature.

Yes, in four ordinary ways. The tool asks you to confirm three of them rather than assuming, and refuses outright on the fourth. A micron gauge left near the pump reads the deepest, cleanest part of the setup and can stay low while the system sits far higher, which is the false pass. Readings taken with the pump still pulling describe the pump, not the system. Valve cores left in the circuit restrict the path so the pump pulls the hose down while the system stays high, and the equalisation afterwards looks exactly like a decay curve. And below freezing, water in the system is ice, whose vapor pressure is far lower than liquid water's, so a wet system can sit at a deep flat reading that looks like a pass. The tool refuses below freezing, though the bound itself comes from the temperature window the water line can be derived over rather than from the ice argument.

No, and this is worth being precise about. A pass here means your readings met the hold criterion you entered, over the time you entered. A decay test pulls the system inward, and a joint can seal under vacuum and open under a few hundred PSI of operating pressure. That is why manufacturers specify a nitrogen pressure test as a separate step rather than as an alternative. This test is good evidence that the system is dry and tight enough to charge; it is not a substitute for pressure testing it.