Vacuum won't hold (microns rise), but nitrogen holds and no leak is found outside
First look at HOW the microns rise with the pump valved off. Rises and stops — that's gas from the oil and moisture, not a leak (the level depends on how much moisture and the temperature): warm the compressor, triple evacuate, every round gets better. Climbs steadily and never stops — that's a leak nitrogen didn't show in half an hour: a micron gauge is far more sensitive (1 psi = 51,700 µ). Look at the compressor terminals, the shell seam, the brazes under the sound blanket. And check your own rig first: hoses and cores are the most common cause of a false rise.
Likely causes, in order
1. Old oil and compressor internals outgassing: dissolved refrigerant and water from POE boil under vacuum — a rise that levels off. Not a leak
Tools: Micron gauge on the system (not the pump), clock, heat gun, nitrogen
- What happens: under vacuum the refrigerant dissolved in the compressor oil boils out, and water boils out of POE oil (POE pulls moisture from air within minutes). An old compressor with a quart of dirty oil is a “gas cylinder” inside the system. A new compressor arrives with dry oil under a nitrogen holding charge — nothing to outgas, the vacuum holds.
- Curve-shape test: pull to 300–500 µ, valve off, watch 15–30 minutes. Oil gas/moisture: fast rise, then the curve LEVELS OFF and sits (the level depends on how much moisture and the temperature; with free water at 72–80 °F up to 20,000–25,000 µ, frozen water about 3500–4500 µ). Leak: a straight line climbing steadily with no stop, all the way to atmosphere.
- Second round: pull again. With moisture/oil each next round rises less and slower — the gas is leaving. With a leak — the same every time.
- Speed up outgassing: warm the compressor crankcase and receiver with a heat gun (not above 150 °F), nitrogen sweep “triple evacuation”: vacuum → 5 psi nitrogen for 10 minutes → release → vacuum again, three cycles. Change pump oil if it clouds.
- Burnt compressor (acidic oil, smell): without a new filter drier and, on a bad burnout, a suction-line clean-up drier, the vacuum will “not hold” for hours — acid and water leaving the oil and windings.
- Bottom line: a vacuum that levels off — charge when the 10-minute rise is within target (POE 500 µ, mineral 1000 µ). No need to chase zero.
Normal: Curve levels off, each round better than the last, holds target for 10 minutes.
If not: Steady never-stopping rise → leak (see the next card).
2. A micro-leak nitrogen won't show in 30 minutes: a micron gauge sees it in minutes (compressor terminals, shell seam, stubs under the sound blanket)
Tools: Nitrogen (to the plate/manual test pressure, often 150 psig, never above design low), accurate gauge, micron gauge, soap, electronic leak detector
- Different sensitivity. Nitrogen at the plate/manual test pressure (often 150 psig) on an analog gauge: a visible drop is 2–5 psi in 30 minutes, i.e. a noticeable leak. A micron gauge: 1 psi = 51,700 microns. A micro-leak that won't move the needle in half an hour of nitrogen gives +2000 µ in 10 minutes under vacuum. That's why the vacuum test finds what the nitrogen test misses.
- Where such micro-leaks hide on an old compressor: the glass insulators of the electrical terminals (Fusite) — cracks/corrosion under the terminal cover; the shell weld seam; the factory brazed suction and discharge stubs UNDER the sound blanket where nobody puts soap or a detector; the process-tube plug.
- What to do before calling it a “leak”: pull the blanket off the compressor, remove the terminal cover, soap the terminals and seams at 150 psi nitrogen, detector slowly (1″/s) around the whole shell from below — refrigerant/trace gas is heavier than air.
- Nitrogen test done right: at least 30 minutes, better 1–2 hours, a digital gauge reading 0.1 psi, temperature correction (Tools → Nitrogen). A 1 psi drop per hour on a temperature-corrected digital gauge is already a leak.
- Directional leaks exist: a crack that seals under internal pressure and opens under vacuum (gaskets, a cracked terminal insulator). So “nitrogen holds” by itself is not proof.
- It holds after the compressor swap because the old shell's terminals, seam and factory brazes left with it and the new brazes are yours. That's the normal outcome, not a mystery.
Normal: Nitrogen with no drop for ≥1 hour on a digital gauge AND vacuum holding target.
If not: Nitrogen holds, vacuum climbs steadily → micro-leak in the compressor shell/terminals or in brazes under insulation.
3. Moisture in the system after sitting open / after a burnt compressor
4. Your rig: hoses, Schrader cores, manifold, micron gauge on the pump instead of the system
Tools: 1/4″ caps, micron gauge, core tool
- Put the micron gauge on the system (the second service port or a tee at the port), not on the pump: the pump always shows 50 µ, the system shows the truth.
- Hoses: old rubber hoses leak under vacuum by themselves; Schrader cores in the ports and in the hoses are the main source of false rise. Pull the cores with a core removal tool under vacuum or use core-less hoses.
- Rig test: close both manifold valves, cap the hoses, evacuate the manifold with hoses and gauge to 200 µ, valve off the pump — it must hold. It rises → replace gaskets/hoses, don't chase a leak in the unit.
- Ball-valve core removal tools beat a manifold: closed — the system is isolated from pump and hoses, the micron gauge sits on the system.
- Only when the rig holds — draw conclusions about the system.
Normal: Manifold with hoses holds 200–300 µ on its own.
If not: Replace hoses/gaskets/cores, repeat.