Thermal imaging camera for HVAC: finding hot spots and losses
Thermal imaging camera for HVAC work: where it pays off on refrigerant, electrical and hydronic jobs, and how emissivity and reflections fool the reading.
At a Glance
A thermal imaging camera for HVAC shows you where temperature changes, not how hot something is: it finds a hot electrical lug, a refrigerant restriction or a leaking radiant loop in seconds. Its numbers only hold up once emissivity and reflected temperature are dealt with — and a contact probe confirms them.
An infrared spot thermometer gives you one number. An entry-level thermal imaging camera for HVAC work — a FLIR C5 with its 160 × 120 sensor, say — gives you 19,200 of them at once. Yet aim it at a new copper liquid line carrying 40 °C refrigerant and it may happily report the temperature of the ceiling. Both statements are true at the same time, and that is the whole story of this tool: it sees a great deal, and it measures badly until you help it.
This review is for the technician who leaves the camera in the bag, or pulls it out for dramatic pictures that prove nothing. It covers what the imager does well, where it pays for itself on refrigeration, electrical and hydronic work, and how to avoid the readings that get healthy parts replaced.
What is a thermal imaging camera actually good for in HVAC?
Short answer: a thermal imaging camera in HVAC locates a temperature difference; it does not measure a temperature. In seconds it shows which component runs hotter than its neighbours, where a line abruptly changes state, or where a heating loop is leaking. The exact number gets confirmed afterwards, with a contact probe.
The imager picks up infrared radiation coming off surfaces and turns it into a picture. The catch is that surfaces don’t all emit equally well — that’s emissivity. Matte electrical tape, paint, insulation and oxidized copper emit strongly; polished metal emits little and reflects the rest. Teledyne FLIR puts it plainly: a polished metal surface behaves like a mirror, and the camera records the heat reflected off it rather than the metal’s own temperature.
That gives the rule everything else hangs on: the image is reliable for comparing two areas of the same material under the same conditions; absolute values are reliable only after correction.
Six jobs where the imager earns its place
- Panels and starters. A loose lug, a tired fuse or an unbalanced phase heats up before it fails. The 2023 edition of NFPA 70B, the U.S. electrical maintenance reference, now requires an annual infrared inspection of electrical equipment, according to Fluke.
- Refrigerant circuits. Along a liquid line, the camera pinpoints exactly where temperature drops: a plugged filter-drier, a partly closed valve, a kinked tube. HVAC School describes it simply — you can see where a bright glow fades to a dimmer colour.
- Coils. A condenser or evaporator with a section that stays cold, or uneven feed between circuits, is obvious at a glance on a running rooftop unit.
- Hydronics. A radiant floor loop shows as warm lines; a slab leak forms a diffuse bloom spilling past the tubing layout. Cold radiators at the far end of a riser, stuck vents, zone valves that never open: same logic.
- Ducts and diffusers. Leaky joints in attic spaces, diffusers that aren’t delivering, missing insulation on ductwork running above a cold ceiling.
- Envelope. Thermal bridges, air leakage around windows and flat roofs. In Montreal, the 10 °C indoor-outdoor difference building thermography guides call for is reached on most nights from late October to April.
Reading a thermal image without getting fooled
Set emissivity before you touch the palette. On insulation, paint or painted ductwork, an emissivity of 0.90 to 0.95 works. On new copper, stainless steel or bare bus bars, no setting will rescue the reading: stick on a piece of matte electrical tape, wait for it to reach the surface temperature, then measure on the tape.
Enter the reflected temperature. Most service cameras ask for it. It’s the temperature of whatever the surface “sees”: the room, the sky, you. On a flat roof in summer, or in front of a panel in a hot mechanical room, leaving the default value skews every low-emissivity reading.
Compare against a twin. The most robust electrical reading compares three phases on the same feeder, or two identical breakers under similar load. Widely used severity tables, drawn from NETA maintenance testing specifications, put a 1–3 °C difference between similar components in “investigate” territory and 4–15 °C in “probable deficiency” territory. Those tables assume enough load — at least 40%, or the highest normal load available.
Lock the span manually. In auto mode the camera stretches its palette between the coldest and hottest points in the frame, so a 2 °C difference can glow bright red. Lock the span to a known range before comparing two images or two visits.
Confirm with a probe. When the number drives the diagnosis — temperature drop across a drier, coil delta, return water temperature — a pipe clamp or thermocouple settles it. The camera found the spot; the probe tells you how much.
Four mistakes that produce false diagnoses
- Mistaking a reflection for a source. A warm patch on a sheet-metal panel door may be you, or a light fixture. Take a few steps sideways: a reflection moves with you, a real hot spot stays put.
- Scanning without load. An idle contactor, a heat pump in defrost or a stopped hydronic loop shows nothing. Thermography is done with the equipment at steady state.
- Shooting the envelope in full sun. A sunlit wall gives back heat for hours and hides any insulation defect. Early morning, from the inside, gives the most honest images.
- Treating the picture as the report. A thermal image with no emissivity, reflected temperature, load or outdoor temperature recorded can’t be compared to anything next year. As with an ultrasonic leak detector, the method is what gives the instrument its value.
Laval field case: the drier that was choking a rooftop unit
July, a strip mall in Laval, flat roof, a roughly ten-ton rooftop unit serving a pharmacy. Classic heat-wave complaint: supply air “not cold enough” in the afternoon. The previous technician had called it low on refrigerant and recommended topping up.
The readings said otherwise: high superheat, but normal-to-high subcooling. Low charge would have shown high superheat with low subcooling; normal-to-high subcooling points to a liquid-line restriction instead. One pass with the camera, span locked, showed an even liquid line up to the filter-drier, then a distinctly darker shade at its outlet. Clamps on either side confirmed a 6 °C drop — and HVAC School holds that any confirmable temperature drop across a line drier means it is restricted.
The same pass caught one lug on the compressor contactor about 12 °C hotter than the other two, under the same load. Fixes: refrigerant recovered, drier replaced, system evacuated and weighed in; lug re-torqued, then the contactor replaced after its pitted contacts were found. On the follow-up check, the drop across the new drier was negligible and the phase-to-phase difference was under a degree. The recommended top-up would have added refrigerant to a circuit that was already correctly charged.
Field verdict: which imager for which job?
| Criterion | Native 160 × 120 sensor | Native 320 × 240 sensor |
|---|---|---|
| Typical use | Close-range electrical, line sets, radiant floors | Envelope, roofs, small or distant targets |
| Strength | Pocket-sized, always on the truck | Enough detail for a client report |
| Limit | A small lug at 2 m covers only a few pixels | Pricier and bulkier |
| Best fit | HVAC service technician | Portfolio manager, scheduled surveys |
Ignore the screen resolution and any software-”enhanced” figure: what matters is the sensor resolution on the spec sheet, along with thermal sensitivity (NETD) and measurement range.
For most technicians, a native 160 × 120 used with discipline does more work than a 320 × 240 left in auto mode. Add a roll of matte electrical tape and a pair of pipe clamps: that trio is what turns a pretty picture into a diagnosis.
Before the camera goes back in the case: log four values
On your next maintenance visit, run a thermal pass on each unit’s panel and liquid line with the span locked, and write emissivity, reflected temperature, load and outdoor temperature on the work order. That image becomes next year’s baseline.
For a property manager, a two-minute scan that heads off a needless refrigerant top-up or a melted contactor in the middle of a heat wave is worth more than any parts invoice. That evidence — image and number side by side — is what Montréal Combustion files in the preventive maintenance record for every unit it services.