The physical reality
By Vimal Bhaya, Founder and Lead Systems Architect · Last reviewed 7 September 2026
Most cold room monitoring that fails does not fail in the software. It fails because a walk-in is a metal box, because a coin cell cannot push a radio in the cold, or because the sensor ended up on the wrong wall.
This is a six-part tutorial on those constraints. Every one of them is settled physics rather than a matter of opinion, and every one of them is decided before you ever see a dashboard. Read them in order, or jump to the one that matches the argument you are having.
The 6 parts, in order
By the end you will know why Wi-Fi dies inside a freezer and what does not, how much margin each radio actually has, what it takes to cover a whole site, which batteries survive -20°F (-29°C) and why reporting interval decides that, where in the box the sensor belongs, and how little the probe type matters next to the rest.
-
1. Will wireless work through a freezer wall?
How well do your phone calls go in an elevator? A walk-in panel is the same steel box: a 4 inch foam core with a sheet of metal on each face. The link has to cross it in both directions, and at 2.4 GHz both directions fail.
Read part 1 -
2. LoRa, 433MHz, Wi-Fi and Zigbee compared
The same comparison in numbers. LoRa can afford to lose about 55 decibels more signal than Wi-Fi and still be read, which is roughly three hundred thousand times more margin. Also what LoRaWAN adds to LoRa, and where Wi-Fi HaLow fits.
Read part 2 -
3. Metal walls, long spans and multi-room sites
On a real 190 by 95 foot production floor the far freezer is 208 feet and ten walls from the network cabinet. Why adding access points does not fix that, and what each approach actually costs to install.
Read part 3 -
4. Battery or wired: what survives -20°F
The coin cell in your remote is rated colder than your freezer and still dies in it, because a rated range is measured at a trickle. Which chemistry works, and why reporting every 2 minutes costs seven and a half times as much battery as every 15.
Read part 4 -
5. Where to place a sensor, and where not to
A walk-in is not one temperature. Five places a sensor ends up, what each one is really measuring, and the quiet failure where a weak radio slowly pushes the sensor to where the signal is rather than where the food is.
Read part 5 -
6. Probe types: thermistor, RTD and 4-20mA
Two of those are sensors and one is a wire. What each actually is, when platinum and a calibration certificate are worth paying for, and why placement moves your reading about fifty times further than probe tolerance does.
Read part 6
How to use this
Read them in order if you are starting from scratch. That is the usual path. If you already know which piece you need, jump to it from the list above.
The figures in this series come from published datasheets and standards, from an annotated floor plan of a working food production site, and from our own Philadelphia freezer and Mississauga reefer case-study exports. Every source is listed at the foot of each part.
This tutorial follows the alerting problem, which covers what to do with the readings once they are arriving. This one is about getting them to arrive at all.