Finding Refrigerant Leaks: Methods Compared
Electronic detectors, bubble solution, UV dye, and nitrogen pressure testing — ranked for how they actually perform on a service call, and how to combine them so you never recharge a system twice.
Found the leak? Charge it right.
After the repair, weigh in the exact nameplate charge and verify it with our refrigerant charge calculator.
In This Guide
Why Leak Detection Is Non-Negotiable
A refrigerant leak is the second most common reason an AC system loses capacity, right behind restricted airflow. The symptoms line up quickly: low suction pressure, high superheat, low subcooling, bubbling in the sight glass, and warm air off the supply registers. The temptation is to slap on a set of gauges, weigh in a pound, and move to the next call. Don't.
A sealed refrigeration system should never need refrigerant added. If it's low, it leaked. Under EPA Section 608 of the Clean Air Act, knowingly venting refrigerant is illegal, and for comfort-cooling appliances above a threshold charge, a documented leak that exceeds the annual leak-rate trigger must be repaired — not just topped off. Beyond the regulation, recharging without finding the leak is a callback waiting to happen. You will be back in weeks, and this time the customer is angry.
Never top-and-run
Adding refrigerant to a leaking system is a temporary fix that wastes increasingly expensive refrigerant, violates recovery rules, and guarantees a return trip. Find it, fix it, verify it.
Read the Gauges First
Before you spend an hour hunting, confirm the system is actually undercharged and estimate how badly. Your readings also tell you where to look — a system that's bone dry probably has an accessible, fast leak, while one that's a half-pound low over a season points to a slow seep in the coil.
The undercharge signature
- Superheat: high, above 20°F on a TXV system (target is 8–14°F) — the evaporator is starved.
- Subcooling: low, below 5°F (target 10–18°F) — not enough liquid stacked in the condenser.
- Suction pressure: below the normal saturation for a ~40°F evaporator.
Both symptoms together — high superheat and low subcooling — are the classic undercharge fingerprint. If subcooling is high instead, you have an overcharge or a condenser problem, not a leak. Get those numbers straight before you break out the detector.
The Four Methods, Compared
1. Electronic Leak Detector (the daily driver)
The electronic sniffer is what you reach for first on most calls. The two common sensor types are heated diode and infrared (IR). Heated diode units are extremely sensitive and cheaper, but they false-alarm around moisture, cleaning chemicals, and even some ozone; they also need periodic sensor replacement. Infrared sensors are more stable, drift less, and hold calibration longer, which is why they've taken over the premium tier. A good detector resolves leaks down to roughly 0.1 oz/yr.
Technique matters more than the tool. Move the probe slowly — about an inch per second — and keep it below the suspected joint for heavier-than-air refrigerants like R-410A and R-454B, since the vapor sinks. Kill fans and wait for still air. Follow the piping methodically: service valves, Schrader cores, brazed joints, the reversing valve, and every return bend on the coil.
Field caution
Wind, blower operation, and residual refrigerant in the air produce false positives. Confirm any electronic hit with bubbles before you cut into the system. The detector tells you where to look, not where to braze.
2. Bubble Solution (the confirmer)
Nothing beats watching a bubble grow. Purpose-made leak-detection solution clings and foams at pressures where dish soap barely reacts. It's cheap, unambiguous, and it pinpoints the exact hole — no guessing. The limitation is access: you can only test where you can reach and see, and the joint must be under positive pressure. On a dead- empty system that means adding a nitrogen holding charge first. Bubbles are the perfect second step after an electronic hit, and the primary method during a pressure test.
3. UV Fluorescent Dye (the slow-leak catcher)
UV dye is injected into the system and circulates with the oil. When it weeps out at the leak it leaves a bright yellow-green stain visible under a UV lamp. Dye shines for the intermittent, slow leak you can't catch in a single visit — you dye it today and inspect on the next service. Downsides: it takes time to migrate, it can mask small leaks if overused, and some equipment manufacturers frown on dye for warranty reasons (always check). Use manufacturer-approved dye and the correct dose; too much dye can affect oil return.
4. Nitrogen Pressure Test (the ground truth)
When the electronic detector is chasing ghosts, or the system is already empty, dry nitrogen is the definitive answer. Pressurize the isolated system with regulated nitrogen, sweep the joints with bubbles, and — if nothing shows — run a standing pressure test: record pressure and temperature, walk away, and come back to see whether the pressure held. It's the only method that proves a system is tight rather than just failing to show a leak. It's also the slowest and requires recovering any remaining charge first.
Safety: pressure limits
Always regulate the nitrogen and never exceed the low-side design pressure stamped on the data plate. Uncontrolled nitrogen from a full bottle can hit 2,000+ psig and rupture a coil or launch a fitting. Never pressure-test with oxygen or any flammable gas.
Ventora — Your AI HVAC Companion
PT charts for R-410A and R-454B, superheat and subcooling calculators, and AI troubleshooting — right in your pocket. Free on iOS.
Side-by-Side Comparison Table
No single method wins every situation. Here's how they stack up on the factors that decide which one you pull off the truck:
| Method | Sensitivity | Speed | Pinpoints Leak? | Best For |
|---|---|---|---|---|
| Electronic (IR / diode) | Very high (~0.1 oz/yr) | Fast | General area | First-pass hunting |
| Bubble solution | Moderate–high | Fast | Exact spot | Confirming accessible joints |
| UV dye | Moderate | Slow (return visit) | Exact spot | Intermittent / slow leaks |
| Nitrogen pressure test | Definitive (pass/fail) | Slow | With bubbles: yes | Empty systems / proving tight |
The takeaway: electronic detection finds the neighborhood, bubbles find the house, dye watches over time, and nitrogen delivers a verdict. The best techs layer them rather than betting on one.
Nitrogen Standing Pressure Test, Step by Step
This is the procedure to fall back on when you need certainty — a slow leak that hides from the sniffer, or a system that came in empty. Work it in order.
Step 1 — Recover the charge.
Pull any remaining refrigerant into a recovery cylinder per EPA 608. Never vent. Note how much came out — it tells you the leak size.
Step 2 — Pressurize with dry nitrogen.
Through a regulator, bring the system up to the low-side test pressure on the data plate — commonly around 150 psig for R-410A/R-454B equipment. Never exceed the low-side design pressure.
Step 3 — Bubble the suspect joints.
Brush leak solution on brazed joints, flares, Schrader cores, and service valves. Growing foam = your leak. A trace of dish soap won't do; use real leak-detection solution.
Step 4 — Run the standing test.
If nothing shows, record the exact pressure and ambient temperature. Isolate and wait — 15 minutes catches gross leaks; several hours (or overnight) catches slow ones.
Step 5 — Correct for temperature.
Nitrogen pressure moves with temperature (~2–4 psi per 10°F). A drop that tracks a cooling ambient may be fine; a drop with steady temperature is a real leak.
Step 6 — Repair, re-test, evacuate, recharge.
Fix the joint, re-pressurize to confirm it holds, then evacuate to below 500 microns, confirm a standing vacuum, and weigh in the nameplate charge.
Worked example: reading a pressure drop
You charge to 150 psig at 75°F and return two hours later to find 144 psig at 72°F.
- Temperature fell 3°F, which alone accounts for roughly 1 psi of the drop.
- That leaves about 5 psi unexplained — the system is leaking. Go back to bubbles or dye.
- Had it read ~149 psig at 72°F, the whole change would be thermal and the system is tight.
A2L Refrigerants and Detector Choice
As the industry transitions from R-410A to low-GWP A2L refrigerants like R-454B (GWP 466) and R-32 (GWP 675), leak detection picks up a safety dimension. A2Ls are mildly flammable, classified under ASHRAE 34. New A2L equipment ships with built-in refrigerant sensors and mitigation logic, and ASHRAE 15 addresses charge limits and detection for these systems.
For your handheld tools: make sure the electronic detector is rated for the refrigerant you're testing. Many modern IR detectors read R-454B and R-32 directly, but confirm compatibility — a detector tuned only for legacy HFCs may under-respond. R-454B operates within about 3% of R-410A pressures, so your gauge targets and standing-test pressures are nearly identical, but the handling and jobsite ventilation rules are stricter. Keep ignition sources away and ventilate before brazing.
A2L reminder
With A2L systems, a leak isn't just a performance problem — it's a flammability consideration. Purge with nitrogen before applying heat, and follow the manufacturer's A2L service procedures.
A Practical Field Workflow
Here's how to sequence the methods so you spend the least time and never braze the wrong joint:
- Confirm undercharge with gauges — high superheat, low subcooling, low suction.
- Visual + electronic sweep. Look for oil stains (oil travels with refrigerant, so an oily spot is a leak flag), then run the detector slowly along all joints and the coil.
- Confirm with bubbles anywhere the detector alarms before you commit to a repair.
- Can't find it? Recover, pressurize with nitrogen, and run a standing pressure test.
- Slow / intermittent leak? Add approved UV dye and schedule a follow-up inspection.
- After the repair: pressure-test to prove it holds, evacuate below 500 microns, and weigh in the correct charge.
Pro tip: follow the oil
Refrigerant carries a little oil everywhere it goes, so a greasy, dust-caked spot on a fitting or return bend is often your fastest lead — before you even switch on the detector. Wipe it clean, then re-check; if the oil comes back, so does the refrigerant.
Diagnose Faster with Ventora
Ventora — the AI HVAC assistant in your pocket. Instant PT charts for R-410A and R-454B, superheat and subcooling calculators, code references, and AI troubleshooting so you find the leak the first time.