A cold storage alarm has two ways to fail. It can ring so often that staff learn to ignore it, or it can stay silent through the one night that matters. Both come from the same few decisions: which limits, how long a delay, where the sensors sit and who receives the alert. This guide goes through those decisions for cold rooms, walk-in coolers, freezer rooms and cold storage warehouses.
If you are still deciding between a data logger and a live system, start with cold room monitoring: loggers vs real-time. This article assumes the alarm exists and asks how to set it up so people trust it.
Start from the product, not the room
A room does not have a correct temperature. The product does. The US Food and Drug Administration puts refrigerated food at or below 4°C (40°F) and frozen food at -18°C (0°F). Pharmaceutical stores work to the range on the label; the WHO guidance on temperature mapping of storage areas gives -25 to -10°C, 2 to 8°C and 15 to 25°C as typical examples. Your HACCP plan, customer specification or product label sets the range. The alarm is a tool for staying inside it.
Three habits make the limits useful:
- Put the alarm limit inside the product limit. If the product limit is 4°C and the alarm also sits at 4°C, the alarm tells you the product is already out. A margin inside the limit gives the shift time to close a door or call a technician before the record shows an excursion.
- Set a low limit as well as a high one. Chilled rooms freeze product too. The air leaving the evaporator is the coldest in the room, and fresh produce or vaccines stored in that air stream can freeze while the room average looks fine.
- Give each point its own limits. A meat room, a vegetable room and a freezer room on the same site need different numbers. So do two points in one room when one sits by the door and one sits in the product zone.
The delay decides whether people trust the alarm
Every cold room leaves its range in normal operation. A door opens for unloading and the air beside it warms. A freezer evaporator defrosts on its schedule and the air around it rises for a while. An alarm that fires on each of those events is noise, and noise gets muted.
The answer is an allowed time out of range for each point. The reading may leave the range, and the alarm fires only if it stays out longer than you allow. In OmniCloud this is set per sensor: a minimum, a maximum and the time a reading may spend outside them before an alert opens.
How long? The data answers that better than a rule of thumb. Let the system record a week of normal operation, then look at each point:
- How long does a normal unloading peak last at the door sensor?
- How long does the defrost peak last near the evaporator, and how high does it go?
- How quickly does the room recover after each one?
Set the delay just above the longest normal peak, and keep it well below the time the product can tolerate. For reference, the WHO mapping guidance accepts readings outside the limits for a maximum of 30 minutes after a door opening when a study includes door openings. A point that keeps breaching its delay during normal work is telling you something about the room, not about the alarm.
Where the sensors go
One sensor describes one point. Temperature in a cold room varies with distance from the door, the evaporator and the ceiling, and with how pallets block the airflow. The way to see that variation is a temperature mapping study: many loggers placed across the room for a period, then the hot and cold spots read from the results.
The WHO technical supplement on temperature mapping gives a practical grid for such a study:
- Across the floor: a logger every 5 to 10 metres along the length and width, and up to 20 or 30 metres in very large facilities.
- In height, up to a 3.6 metre ceiling: three levels at each grid point, for example at floor level, 1.2 metres and 3.0 metres.
- In taller spaces: vertical arrays at the bottom, middle and top. For a 6 metre store it suggests 0.3, 1.8, 3.6 and 5.4 metres.
It also warns that high bay pallet racking is particularly prone to temperature stratification and should be mapped over its full working height, and it notes that a two-season study is typically not necessary for cold rooms and freezer rooms, which sit inside a building and depend less on the weather outside.
A mapping study is temporary. The permanent alarm sensors then go where the study says the risk is, and the WHO guidance lists repositioning the monitoring sensors as one of the normal outcomes of a mapping report. In most cold rooms that means:
- The warm spot, usually near the door or in the far corner behind stacked pallets.
- The cold spot, near the evaporator outlet, where it matters if product can freeze.
- The product zone, at the height where the most valuable stock sits, not on the wall at eye level where a thermometer is easiest to read.
With RS-485 sensors this is a cabling question more than a cost question. Each THS-100 temperature and humidity sensor has its own Modbus address (1 to 254), and several share one bus that can run up to 2,000 metres with proper cabling and termination, so adding the third point in a room means one more sensor on the same cable.
Freezer rooms: keep the electronics outside
A freezer room is colder than most electronics are rated for. The Omni Genesis controller is rated for -20 to +70°C, so in any room held near or below -18°C it is mounted on the outside wall at normal temperature, and only the sensors go inside. The cable enters through an existing pass-through in the panel or a small hole sealed after the cable is pulled.
Two sensors cover most freezer and cold room points:
| THS-100 | DS18B20 probe | |
|---|---|---|
| Measures | Air temperature and relative humidity | Temperature |
| Temperature range | -40 to +80°C | -55 to +125°C |
| Stated accuracy | ±0.3°C at 25°C; ±3% RH | ±0.5°C from -10 to +85°C |
| Connection | RS-485 Modbus RTU, addressed | 1-Wire |
| Typical use | Air at the door, the corner and the product zone | Inside a product sample or where a sealed probe is needed |
Read the accuracy line carefully. Both figures are stated at room temperature or above, not at -20°C. For a freezer point, check each sensor against a reference thermometer at the room’s working temperature after installation. The THS-100 has a temperature offset register for exactly this kind of field trim, so a small difference against the reference can be corrected in the sensor itself.
Housing matters too. The THS-100 comes with seven probe housings. The vented PVC cap responds fastest but is not waterproof, while the sealed polymer, brass, stainless and IP67 sintered housings protect the element where condensation forms, at the cost of a slower response to fast humidity changes. Rooms that see frost, washdown or warm humid air at the door usually call for a sealed one.
Who receives the alarm
An alarm only helps if it reaches someone who can act, at the hour it happens. Decide that before the first night shift, not after the first loss.
In OmniCloud, an alarm goes out as a notification to the users on the account. Each user can set quiet hours and mute alert types they do not handle. A sensor can be marked as critical, and a critical alarm comes through quiet hours, so the freezer room reaches the person on call at 3 a.m. while a humidity drift in the dry store waits until morning. Until someone reads the alert, it repeats, every two hours by default. When the reading returns to range, a second notification says so, which closes the loop for whoever is driving in.
Practical points that save trouble later:
- Name every point by room and position, such as “Freezer 2, door”. An alert that says “Sensor 3” sends someone to the wrong room.
- Keep the user list current. When a shift manager leaves, remove them and add the replacement the same day.
- Mark as critical only what needs a night call. If every point is critical, nobody sleeps and the alarms lose their meaning.
When the monitoring system goes quiet
The worst case is not a loud alarm. It is a silent one, because the system itself stopped working. Plan for three cases:
- Power cut. The Genesis controller runs on two rechargeable 18650 batteries, so it keeps measuring and sending alerts through the outage. Afterwards the history shows when the room started to warm and how fast.
- Lost connection. The controller reports over 4G-LTE with a SIM included, or over Wi-Fi where a stable network reaches it. It logs readings locally and syncs them to the cloud when the connection returns, so the record has no gap.
- A controller that stops reporting. OmniCloud opens a controller offline alert when a controller misses its expected reports, and a low battery alert before the batteries run down. A dashboard with no new readings is treated as a problem, never as a room in range.
For sites with no network or mains power at all, the same controller runs on battery and solar; see wireless temperature monitoring.
Records: what to keep and for how long
The alarm protects the product. The record protects you in an inspection or a dispute over a warm delivery. Requirements depend on your country and product, so check what applies to you. One clear example is the EU rule for quick-frozen food, Regulation (EC) No 37/2005:
- Warehousing and storage of quick-frozen food must have recording instruments that monitor air temperature at frequent and regular intervals.
- The instruments must comply with EN 12830, EN 13485 and EN 13486, and the operator keeps the documents that show it.
- Temperature records are dated and kept for at least one year, longer where the product’s shelf life calls for it.
- Retail cold stores under 10 m³ may use an easily visible thermometer instead, where the authority allows it.
If your store falls under a rule like this one, check which instrument standard it names before you choose the system that will hold the official record. For food businesses working to a HACCP plan, our guide to temperature monitoring for food safety and HACCP covers what the plan asks for, and the temperature log form covers rooms still recorded by hand.
Test the alarm before you need it
An alarm nobody has seen fire is an assumption. Test it when it is installed, and again after any change to users, limits or sensors:
- Warm one sensor by holding it in your hand or moving it outside the room, and leave it there longer than its delay.
- Confirm the alert arrives on the phones of the people who should receive it, with a name they recognise.
- Read the alert and return the sensor, and confirm the recovery notification follows.
- Look at the history weekly. A point that drifts up over several days, with no single alarm, often means an icing evaporator or a door seal that has started to fail.
A setup checklist
| Decision | What to set |
|---|---|
| Limits | From the product range, with the alarm a margin inside it; a low limit as well as a high one |
| Delay | Just above the longest normal door or defrost peak, well below what the product tolerates |
| Points per room | The warm spot, the cold spot and the product zone; high and low in tall rooms |
| Freezer rooms | Controller outside, sensors inside, each point checked against a reference at working temperature |
| Who is alerted | Named users, critical only where a night call is needed, user list kept current |
| System health | Offline and low battery alerts on, battery backup in place |
| Records | Retention and instrument standard checked against your rules |
| Testing | At install, after every change, and a weekly look at the history |
Conclusion
A cold storage alarm earns trust when it rings for real problems and only for real problems. That comes from limits taken from the product, delays measured from the room’s own history, sensors placed where mapping shows the risk, and an alert that reaches a named person and repeats until someone answers. The same setup applies to a single walk-in freezer and to a warehouse with a dozen rooms; only the number of points changes. For refrigerators and freezer cabinets rather than walk-in rooms, see refrigerator temperature monitoring, and for trucks see refrigerated truck temperature monitoring.
The cold room temperature monitoring solution shows what gets installed. Tell us how many rooms you run, at which temperatures, and where a controller can be mounted, and we will plan the measuring points with you.