How long should you leave an HVAC data logger in place?

Long enough to cover at least one full cycle of the phenomenon you're chasing. For a fault that returns every night, two to three days is often enough, and a 48-hour window is a common starting point. For a weekly fault, or one tied to a mid-winter thaw, aim for one to two weeks. The rule: a recording that runs too long beats a window that misses the event and sends you back empty-handed.

Which data logger should I use on a commercial boiler?

It depends on which quantity betrays the fault. For a temperature drift or a comfort complaint, a standalone temperature/humidity logger is enough. To link a shutdown to a component's demand, a multi-channel model with a current clamp captures both temperature and the burner or motor current. For a supply issue — voltage sags, phase imbalance — you need a three-phase energy logger. You pick the instrument from the symptom, not from the catalogue.

Can a data logger replace a multimeter?

No — they do different jobs. A multimeter gives an instant value, ideal for a permanent fault you can measure on site. A logger captures the time dimension: it shows what happens when you aren't there, and above all at the exact instant the system trips. On an intermittent fault, that timeline — impossible to get with a multimeter — is what reveals the cause.

How do I know whether a fault is intermittent or permanent?

A permanent fault repeats every time you operate the equipment under the same conditions; you measure it and fix it on site. An intermittent fault vanishes the moment you arrive and refuses to reproduce on demand, usually because it depends on something you don't control — outdoor temperature, time of day, the building's electrical load. If you've reset a unit twice and found nothing, treat the problem as intermittent and deploy a logger.

HVAC Data Logger: Catch an Intermittent Fault

“No fault found.” Three words every property manager has read on a work order — and they mean nothing. The fault was there; the technician simply wasn’t in the mechanical room when it happened. That is exactly the gap an HVAC data logger fills. Instead of burning hours trying to reproduce a fault that only shows up at 3 a.m. at −22 °C, you let an instrument stand watch in your place and time-stamp the event. Here is a field review of the tool: its role, when to deploy it, how to read what it reports, and the pitfalls that turn thousands of data points into a false trail.

What does an HVAC data logger do about an intermittent fault?

An HVAC data logger is a standalone instrument that measures and time-stamps one or more quantities — temperature, humidity, current, voltage — at a set interval, over hours or days. On an intermittent fault, it captures the exact instant of the fault and the state of the system around it. That is something no single spot reading can do.

Role: turning a phantom fault into time-stamped proof

A multimeter answers “what is the value right now?” A logger answers “what happened, and at exactly what time?” The difference is decisive, because the faults that cost the most in callbacks are precisely the ones that refuse to occur while you’re standing there.

A loose terminal that opens under overnight thermal contraction, a flame-sensing signal that dips just below threshold at re-light, a voltage sag when a large motor starts elsewhere in the building: these faults sometimes last only a handful of seconds. They trip a safety lockout, the system re-arms, and everything looks normal by the time the technician arrives. The logger, though, was there to the second. It doesn’t replace the technician’s reasoning — it finally gives that reasoning the hard evidence it was missing.

When to pull the logger out of the truck

The tool only earns its keep on a specific class of problems. Deploy it when:

a safety lockout keeps returning at random and no on-site reading explains it;
a comfort complaint comes and goes with the hour or the weather (a space that’s cold only on the coldest mornings);
a breaker or high-limit trips once a week for no visible reason;
a temperature drifts slowly over days without crossing a threshold you can catch in one visit.

If you’ve already reset a unit twice and found nothing, you’re on an intermittent fault: instead of a third improvised emergency repair, deploy a logger. The approach complements a structured recurring-lockout diagnosis — the logic tree says what to check, the logger says when the fault strikes.

Three families of loggers, three uses

Not all loggers measure the same thing. Choosing the right one means first identifying which quantity betrays the fault.

Logger type	What it captures	When to choose it
Standalone temperature/humidity (e.g. HOBO MX1104, Testo 176)	Temperature, humidity, one analog input	Comfort complaint, temperature drift, abnormal cycling
Multi-channel with current clamp (e.g. HOBO MX-HVAC kit)	Temperature + AC current of a motor or burner	Linking a shutdown to a component’s real demand
Three-phase energy logger (e.g. Fluke 1730, 600 V CAT IV)	Three-phase voltage, current and power	Voltage sag, phase imbalance, starting inrush

The HOBO HVAC kit, for instance, bundles a Bluetooth logger, three temperature probes and an AC current sensor: enough to track supply temperature and burner current in parallel on the same time axis. For an electrical-supply problem, the three-phase energy logger records voltage and current on all three phases over time — something neither a handheld thermometer nor a clamp meter will do over a shift.

How to read the data: correlate, don’t just record

A logger that captures a single quantity only moves the question. The diagnostic value comes from correlation: overlaying at least two channels on the same time axis and watching what moves together at the moment of the fault.

Field case. A hot-water boiler in an office building on the South Shore was locking out once or twice a night, never during the day. Three resets, no anomaly measured on site. We deployed a multi-channel logger: supply temperature, return temperature, and a current clamp on the burner, at one-minute intervals over five days. The plot spoke for itself. Every shutdown lined up with the night setback: return temperature plunged, the burner restarted under heavy demand, and the current trace showed a one-second dropout at ignition. An air pressure switch whose contact chattered when cold was opening the circuit for a fraction of a second. Invisible in daylight, obvious on the graph. We replaced the switch, ran five nights of confirmation logging, and the lockouts stopped.

Two settings make or break the exercise. First the sampling interval: too coarse and it steps over a two-second event and you see nothing; on a fast electrical fault, drop to the finest step the device allows. Second, time-stamping: logger clocks drift, and one channel offset by a few minutes against another destroys any correlation. Synchronize the devices before you place them.

The reading errors that cost a second callback

A good instrument used poorly sends you back on site as surely as no measurement at all. The recurring traps:

One channel, no reference. Logging temperature without current, or current without temperature, leaves the fault with no context. You need at least one “cause” quantity and one “effect” quantity.
Interval too long. A five-minute step misses a one-second dropout. Match the cadence to the phenomenon, even at the cost of battery life.
Ignoring the Montreal context. Here a fault may only surface at the first hard freeze or during a thaw that cycles the equipment more often. Without outdoor temperature on the same graph, you miss the trigger.
Bypassing a safety device to “see.” You log a safety’s behaviour; you never jumper it. The CSA B149.1 code, in force in Quebec, governs these installations: a defeated lockout is a carbon-monoxide or overheat risk, not a diagnosis.

Field verdict

An HVAC data logger isn’t a commissioning gadget: it’s the tool that ends the callback loop. The moment a fault refuses to reproduce, it shifts diagnosis from guesswork to proof. The next time a unit locks out “for no reason,” don’t reset it a third time: place two relevant channels, let it run as long as it takes, and read the graph.

For a manager, a well-placed logger replaces a run of “no fault found” invoices with a dated cause and a repair that holds. That shift — from billed time to a solved problem — is what Montréal Combustion looks for on every phantom fault in a Greater Montreal boiler room, because an intervention that fixes the fault on the first try is worth far more to the owner than its hourly cost.

HVAC Data Logger: Catching an Intermittent Fault