Skip to content
AkerminMeasure. Monitor. Improve. Compare Products
Learning Center

What a data logger tells you that a live display cannot

The difference between a monitor and a logger is not the sensor, which is often identical. It is that one quantity is witnessed and the other is recorded with a time stamp. This article sets out the four questions only a record can settle, works memory depth into usable deployment spans, and shows what the published figures on four loggers do and do not cover.

As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.

A live display answers exactly one question: what is the value at the moment somebody happens to be looking at it. A record answers four questions that no amount of looking can settle, which are how long a condition persisted, how quickly it changed, whether it repeats on a cycle, and whether two quantities moved at the same time. The sensing element inside a monitor and inside a logger is frequently the same part. What changes is that one value is witnessed and forgotten while the other is written down with a time stamp against it, and almost every useful conclusion about a building depends on the time stamp rather than on the value.

A display answers now, a record answers how long

Standing in a room with an instrument in your hand, you can establish the present state to whatever precision the instrument offers. You cannot establish anything about the hours you were not there, and those hours are where the condition you are investigating actually lives. A bedroom that reads comfortably at six in the evening may have been far outside any sensible band at four in the morning. A cellar that looks dry on a bright afternoon may run wet for days after rain. A workshop that reads well while the extract is running tells you nothing about the overnight state with everything off.

There is a second effect that catches people out, which is that the observer is part of the measurement. A person standing close to a carbon dioxide monitor is a carbon dioxide source, and in a small room the act of walking over to read the number raises it. Body heat and breath moisture do something similar to temperature and humidity at close range. An instrument left alone and interrogated afterwards removes both the gaps and the interference. The definition of an environmental data logger sets out the parts that make this possible, principally a clock, a memory and a scheduler.

Four questions a glance at a screen cannot settle

Duration above a threshold is the first and the most frequently needed. Almost every guideline in building physics is written as a value held for a period rather than a value touched once, and a single reading cannot distinguish a brief excursion from a sustained problem. Rate of change is the second. How quickly a room recovers after a window is opened is a direct measurement of its ventilation, and the recovery curve is only visible if you have a sequence of readings rather than two isolated numbers.

Periodicity is the third. A condition that repeats on a daily cycle has a daily cause, usually occupancy, heating control or solar gain, while a condition that tracks the weather has an external cause. Separating those two from a record is straightforward and from a single reading impossible. Coincidence is the fourth. When two channels move together you have a candidate mechanism, and when they move independently you have ruled one out. A rise in humidity that follows every fall in temperature is condensation behaviour rather than a moisture source. The piece on using carbon dioxide data to evaluate ventilation works one of these shapes through in detail.

How deep the memory has to be for the interval you want

Memory depth and interval together set the span, and the arithmetic is worth doing before you leave rather than after you come back. Take the figure published for the HOBO MX1101, which states 128 kilobytes of storage holding up to 84,650 measurements alongside a stated range of minus 20 to 70 degrees C with an accuracy of plus or minus 0.21 degrees C in standard conditions and 1 to 90 percent relative humidity with plus or minus 2 percent typical accuracy. The bullets do not say whether that count treats a temperature and a humidity reading taken at the same instant as one measurement or two, so the honest approach is to work both out.

Interval Sample sets per day Span if the count treats each channel separately Span if the count treats a pair as one
10 seconds 8,640 About 4.9 days About 9.8 days
30 seconds 2,880 About 14.7 days About 29.4 days
1 minute 1,440 About 29.4 days About 58.8 days
5 minutes 288 About 147 days About 294 days
15 minutes 96 About 440 days About 880 days

The practical lesson is that the interval decision is usually a memory decision in disguise, and that the ambiguity in a published count is worth a factor of two. Plan against the pessimistic column and you will never come back to a truncated file.

What these loggers publish about memory, interval and accuracy

Instrument Interval range stated Memory stated Accuracy stated Alerting stated
HOBO USB temperature and humidity logger 1 second to 18 hours Not stated in the listing bullets Plus or minus 3.5 percent relative humidity from 25 to 85 percent Programmable alarm
HOBO MX1101 Not stated in the listing bullets 128 kilobytes, up to 84,650 measurements Plus or minus 0.21 degrees C, plus or minus 2 percent relative humidity typical Not stated in the listing bullets
Autopilot desktop carbon dioxide monitor and logger Not stated in the listing bullets Adjustable 24 hour or 7 day chart on the unit Not stated in the listing bullets Audible alarm for low or high carbon dioxide
Lascar wireless carbon dioxide and air quality logger Not stated in the listing bullets Not stated in the listing bullets Not stated in the listing bullets High and low alarms with notifications

Two of those rows describe instruments that straddle the boundary between a monitor and a logger. The Autopilot desktop unit keeps a trend chart on the device itself over an adjustable window and lets you view it by minute, hour, day or week, which is a record you read on the instrument rather than a file you take away. The Lascar wireless unit goes the other way, publishing dual power from a mains supply with battery backup and an app for viewing and downloading from any connected device, which makes it a logger you never physically visit. Choosing between those two shapes is covered in the guidance on carbon dioxide loggers for long term work and, for the temperature and humidity case, in the temperature and humidity logger guidance.

Turning a downloaded file into a decision

Look at the shape before you calculate anything. Plot the whole deployment on one axis and let your eye find the repeating pattern, the step changes and the flat sections. A flat section in a building record is usually a sensor stuck at a limit or a unit that stopped logging, not a remarkably stable room, and it is far easier to spot visually than in a summary statistic. Once the shape makes sense, annotate it with everything you know about what the building was doing, because an unexplained feature with a known cause beside it stops being unexplained.

Only then reduce the record to numbers, and reduce it to the numbers that answer the question you were asked. Time above a threshold, worst sustained period, recovery time after a known event, and the difference between two locations are all defensible. An average across the whole deployment almost never is, because it blends the conditions you care about with the hours in which nothing was happening. The write up of the HOBO logger goes through what the download step looks like in practice for one of these units.

Frequently Asked Questions

If the sensor is the same, why does logging change the answer?

Because the limitation was never the sensor, it was the observer. A person reads a display when they happen to be present, which is almost never when the interesting thing happens. A bedroom at four in the morning, a loft during a cold snap, a cellar over a holiday weekend: those are the conditions that matter and nobody is standing there. The record removes the requirement that a human be in the room.

Does my presence actually change a reading?

On some quantities very much so. An adult exhales carbon dioxide continuously, so walking up to a monitor to read it raises the number you came to read, and the effect is worst in exactly the small enclosed rooms people most want to assess. Body heat and breath moisture do the same to temperature and humidity at close range. A logger left alone and downloaded afterwards sidesteps the whole problem.

What interval should I use if I am not sure what I am looking for?

Five minutes is a reasonable default for building conditions because it resolves anything driven by occupancy, heating or ventilation while still spanning weeks of memory. If you suspect something fast, such as a plant item cycling or an extract fan that runs briefly, drop to one minute for a short reconnaissance deployment first and then go back out with a slower interval and a longer span.

Is a logger with a screen better than one without?

A screen earns its place at deployment and retrieval rather than during the record. It lets you confirm the unit is awake, sitting at a plausible value and still logging before you walk away, which prevents the single most expensive failure in this work, which is a blank download after several weeks. During the deployment itself the screen is irrelevant, and on some units it shortens battery life.

How long should a first deployment run?

Long enough to contain at least two full cycles of whatever you think drives the condition. For an occupancy or heating effect that means a minimum of a full week so that both weekdays and a weekend are present. For a weather driven effect it means covering a spell of the relevant weather, which can mean leaving the instrument out for several weeks and accepting that one quiet fortnight proves nothing.

Can I trust a logger that has been sitting in a drawer for a long time?

Check the clock and the battery before anything else. A real time clock that has drifted or reset makes every time stamp suspect, and a record you cannot align with an event log or a weather record is worth much less. Beyond that, sensors drift at their own pace, so a quick side by side comparison against a second instrument in a stable room is a cheap sanity check before a deployment you intend to rely on.