Milesight Neigungssensor mit aufgedruckten X-, Y- und Z-Achsen an der Innenseite eines geöffneten Metalldeckels

Tilt Sensor: How Does Tilt Monitoring Work?

After a heavy storm, the wooden pole at the edge of your property looks a little more crooked than it did last week. Or does it? And the lid on the job-site toolbox: is it actually shut after hours? Tilt monitoring exists for exactly these questions. Here is the short answer: an electronic tilt sensor uses an accelerometer to measure which way gravity points relative to its housing and calculates angles for the X, Y and Z axes from that. It raises an alarm when one of those angles crosses a limit you set, measured against a stored reference position.

The catch is in the details: limited accuracy, a reference position that has to be right, and a wireless sensor that does not report every second. Using the Milesight EM310-TILT, a three-axis LoRaWAN sensor, as the example, you will see realistic numbers, where to mount it and how to pick useful thresholds.

Milesight tilt sensor with printed X, Y and Z axes mounted on the inside of an open gray metal toolbox lid
Screwed down firmly, the tilt sensor follows every movement of the lid. The printed axes show which angle changes when the lid opens.

How does a tilt sensor work?

Inside most modern tilt sensors is a MEMS accelerometer, a tiny chip that measures acceleration along three axes at right angles to each other. When the sensor sits still, the only force acting on it is gravity, roughly 9.81 m/s². Depending on how the housing is tilted, that force is split differently across X, Y and Z. The firmware calculates the tilt angles from the ratio of the three components.

The EM310-TILT works exactly this way: its built-in three-axis accelerometer delivers angles for X, Y and Z, each from −90° to +90°, and the axis directions are printed on the housing.

Two limits follow from this principle. First, rotating around the vertical axis does not change the direction of gravity relative to the housing. Second, the chip cannot tell gravity apart from other accelerations. If a machine vibrates or a pole sways in the wind, that motion leaks into the reading. The angle is only reliable once the object has come to rest.

You will see several names for the same principle: tilt sensor, inclinometer, tilt meter or angle sensor. Note that “angle sensor” also covers rotary encoders on shafts, a different class of device.

How accurate is a tilt sensor?

Two datasheet numbers matter here and are often confused: resolution and accuracy. The EM310-TILT reports angles with a resolution of 0.01° but specifies an accuracy of ±1°. The second decimal place tells you how finely the value is displayed, not how close it is to reality.

What ±1° means in practice: the table lists how far the top of a rigidly tipping pole moves sideways.

Tilt Top displacement, 5 m (16 ft) pole Top displacement, 10 m (33 ft) pole
1° about 9 cm (3.4 in) about 17 cm (6.9 in)
2° about 17 cm (6.9 in) about 35 cm (13.7 in)
3° about 26 cm (10.3 in) about 52 cm (20.6 in)
5° about 44 cm (17.2 in) about 87 cm (34.4 in)

A shift of a few centimeters at the top of a 5-meter pole falls inside the sensor’s uncertainty. A lean of three or five degrees, which can move the top by half a meter, is detected reliably. Zeroing removes a fixed offset, but the datasheet promises nothing better than ±1°.

Error sources outside the chip matter too, such as a bracket with play. Milesight’s user guide states plainly that the sensor must sit flush against the surface being measured, with one of its three axes parallel to the object’s measuring axis, or the result may be wrong. Milesight also notes it is not intended as a reference sensor. For surveying work you need a calibrated inclinometer; for the question “has this moved noticeably?” a wireless tilt sensor is enough.

Accelerometer vs tilt sensor: what’s the difference?

Technically, a MEMS tilt sensor contains an accelerometer. The difference lies in what the device does with the raw data. An accelerometer outputs values in m/s² or g, often many times per second, which makes it good at catching shocks and vibration. A tilt sensor focuses on the static component, gravity, and outputs angles in degrees.

The comparison below sums up the practical differences:

Feature Accelerometer Tilt sensor
Output Acceleration in m/s² or g Angle in degrees per axis
Typical question Was there a shock or vibration? Has the position changed?
Sample rate High, often many readings per second Low, readings on events or at intervals
Motion during measurement Is the actual signal Distorts the angle
Example Impact on a shipping container Open lid, leaning pole

So if you want to know whether a lid is open or a pole is leaning, use a tilt sensor; if you want to know whether someone kicked a container, you need raw accelerometer data at a high sample rate. Simple ball-type tilt switches only report “tipped” or “not tipped,” with no angle.

Tilt sensor alarm: which threshold makes sense?

A sensible threshold sits well above the normal drift of the reading and well below the angle at which things actually get serious. As a starting point, don’t set it tighter than two to three times the sensor’s accuracy. With the EM310-TILT at ±1°, limits below a 2 to 3° deviation tend to produce false alarms rather than insight.

Lids are easy because the movement is large. Closed, the deviation is 0°; open, it might be 70 to 90°. A threshold at 15 to 20° separates the two states cleanly and ignores the few tenths of a degree the closed reading wanders. Poles and walls leave less room: collect readings for a few days first, then set the threshold at roughly twice the largest swing you observed.

On the EM310-TILT, you set an upper and a lower limit per axis, called “Over” and “Below” in the ToolBox app, both relative to the stored reference position. An alarm condition combines the axes with “and” or “or.” It is read left to right; blank means “or.” A pole that can lean in any direction needs something like X or Y; a hinged lid usually needs just one axis.

When a threshold is reached, the sensor sends an alarm packet immediately. It then checks again at the configured alarm reporting interval and, as long as the condition persists, sends further alarms up to the configured count. A return to the allowed range is reported immediately too. The guide describes no hysteresis setting, so a resting value close to the threshold makes the status flip back and forth. Whether an alarm escalates only after a certain duration, and who receives it, is something you configure afterward in your network server or IoT platform.

Milesight tilt sensor on a wooden workbench next to a spirit level and a spiral notebook
Before you store the reference position, check with a spirit level that the object really is in its normal position, and write down the date and readings.

Wireless tilt sensor: when is LoRaWAN worth it?

A wireless tilt sensor pays off whenever there is no power or network cable at the measuring point and one message per event plus a regular heartbeat is enough. LoRaWAN was built for exactly that: a few bytes, long range, years on a battery. According to Milesight, the EM310-TILT reaches up to 15 km (9 mi) line of sight and runs for more than five years on two 3,500 mAh ER17505 lithium batteries when it reports six events per hour. That is a lab figure; cold and frequent alarms shorten it.

Poles, trees, retaining walls or lids on scattered enclosures can all be monitored without running cable. The housing is rated IP67 and specified for −20 to +60 °C (−4 to 140 °F). Milesight’s own datasheet lists tree monitoring, pole lean detection and landslide monitoring as applications.

You need a LoRaWAN gateway in range and a network server that accepts the data. The sensor runs as Class A and can only receive commands right after one of its own transmissions. If you change thresholds via downlink, the change takes effect after the next uplink, which with factory settings could be hours away. For initial setup, use NFC instead: the Milesight ToolBox app reads and writes settings when you hold your phone against the housing. Change the default password 123456 while you are at it.

Wireless is not worth it if you need continuous high-rate logging or a wired controller already sits next to the object; then a wired inclinometer is the better choice.

Where should you mount a tilt sensor?

One rule applies everywhere: the sensor must be rigidly attached to the part whose position you care about, and it must sit flat. The EM310-TILT is screwed onto flat surfaces; the two mounting ears on the housing take the screws, and caps go over them. Zip ties or tape give way, so you end up measuring the mount instead of the object.

On a pole, screw the sensor onto a flat surface or a rigid mounting plate with one axis parallel to the pole. If the pole tips as a whole at its foundation, the angle is the same at the bottom and the top. If it bends, a sensor mounted higher up will show more tilt.

On a retaining wall, the usual question is whether it is leaning forward. Mount the sensor on the top of the wall or on its face, and orient the housing so that one axis lies in the plane the wall would tip in, meaning across the wall, not along it.

On a lid, the sensor goes on the lid itself, not on the frame. Align one axis at right angles to the hinge so the opening movement shows up fully on that axis. Some lids swing open past 90°; because the sensor only covers −90° to +90° per axis, a lid that is flipped all the way back can produce a reading similar to a half-open one. Keep the threshold well below 90°. Metal also blocks radio signals. Milesight recommends keeping the sensor away from metal objects. Mounted on the inside of a steel lid, the closed box becomes a small Faraday cage, so test the radio link with the lid shut before you rely on the messages. The guide also warns against shocks and impacts, which a slamming lid delivers.

How do you set a tilt sensor’s reference position?

With a wireless tilt sensor like the EM310-TILT, “calibration” in practice means setting the reference position. The user guide calls it the “Relative Initial Surface.” You mount the sensor, put the object in its normal position and tap “Setting” in the ToolBox app. From then on, the sensor measures every angle as a deviation from exactly that position, and the thresholds refer to it as well. “Clear” resets the reference to the factory default of (0.00°, 0.00°, −90.00°). Read the values back to confirm it was saved; a downlink command does the same remotely.

What counts as the normal position is the crucial part. For a lid, it is fully closed and latched, not the almost-closed state it is in while you stand there with your phone. Set it with the lid slightly raised, and every threshold shifts by that amount. For a pole or wall, it is the state on installation day. If you need to know whether the object is actually plumb, check it with a spirit level first; the sensor only tells you how the position changes from now on.

Three mistakes come up again and again. You set the reference right after tightening the screws, before anchors and mounting plate have settled; it is better to wait a day and set it again. You move the sensor to a new spot later and forget to reset the reference. Or you don’t record when and how you set it; a short note with date, object state and readings saves guesswork later. Calibration in the metrology sense, meaning adjustment against a reference inclinometer, is not part of this device’s setup, so the ±1° accuracy still applies.

How often does a tilt sensor report?

On the EM310-TILT, two mechanisms decide how often it reports. The first is the regular reporting interval: out of the box, it sends its three angles and battery level every 1,080 minutes, which is every 18 hours. You can set the interval anywhere from 1 to 1,080 minutes. The second is the threshold report: when an angle reaches a limit, the sensor transmits immediately, regardless of the interval. That packet includes a per-axis flag showing which threshold was reached.

For alarms, the long default interval is no problem. It becomes one when the sensor goes silent. A dead battery or a lid that blocks the radio only shows up when the next regular report fails to arrive, up to almost a day later. So choose the interval based on how long an outage may go unnoticed. For a pole you check after every storm anyway, a few hours is fine; for a lid that must be shut every evening, about an hour. Treat a sensor that has missed two regular reports as unknown, not as normal.

Every message costs battery. The manufacturer’s figure of more than five years assumes six triggers per hour, which works out to the same message rate as a ten-minute interval. With a one-hour interval and rare alarms you have headroom; with a weak radio link and cold weather, less.

Where can you buy a tilt sensor?

If you are shopping for a tilt sensor, work through these questions. Does the accuracy match the smallest movement you need to catch? At ±1°, you will reliably detect a clear lean or an open lid, but not centimeter-scale movement in a structure. Can you set the reference position on the device itself, and can you configure upper and lower limits per axis and how they combine? Does the sensor transmit immediately when a threshold is crossed, and what regular interval do you need to notice an outage in time? Is there a LoRaWAN gateway in range? Do the ingress rating, temperature range and screw mounting suit the location, and will the battery last several years at your reporting rate?

Sort these out up front and your tilt monitoring will produce alarms that actually mean something. If you are looking for a suitable tilt sensor, you will find our models in the tilt sensors category.

Our latest news

Scroll to top