Kamado thermometers and probes: which to buy and how to calibrate

Behind that chromed dial sits a bimetallic coil: two metals with different rates of expansion wound into a spiral that turns a needle as it heats. It is cheap, robust and free of electronics, which is why every lid carries one. It is also imprecise. Manufacturers rarely quote a tolerance for these dials, and the ones that do sit around ±10-15 °C. Fifteen degrees of error is irrelevant grilling vegetables and devastating at hour nine of a smoke.

The underlying problem, though, is not accuracy but position. The stem reaches a few centimetres into the highest part of the dome and reads the air circulating up there, not the air around the food. In an indirect cook, hot air climbs the sides, washes over the dome and leaves through the chimney while the grate sits shielded behind the deflector: the gap between what the lid reads and what you actually have at grate height usually runs to 10-25 °C, and not always in the same direction. In low & slow the lid tends to read high; searing over live fire, the surface of the grate is far hotter than the air the dial can see.

None of which makes it an ornament. It is the only figure you can read with the lid shut without switching anything on, and it is good for what it is good for: seeing the trend. If the needle is climbing slowly, the chamber is climbing; if it has been parked in the same spot for twenty minutes, you have found your vent balance. That trend reading is exactly what you use to work the dual airflow.

What you cannot do is cook by it. The two figures that decide are different ones: the ambient temperature at grate height, which a chamber probe clipped to the grill measures, and the temperature at the centre of the meat, which a probe pushed into it measures. Everything else is context.

The instant-read spike is a very fine probe with the sensor housed in the last few millimetres. It does not stay inside: you push it in, read in two to four seconds and pull it out. The good ones quote ±0.5 °C or better, and that accuracy matters precisely where the margin is narrow: a chuletón goes from red to pale pink in three degrees, a fish loin in two. It is the tool you will reach for most often across the year, and a benchmark instant-read such as the Lavatools Javelin PRO Duo covers most of what a domestic kamado cooks without pairing anything to your phone.

The wired leave-in probe is the instrument for long cooks. It stays inside with the lid shut and sends its reading out along a thin cable to a base unit that alerts you when the target arrives. What matters is not the gadget but the fact that almost all of them ship with at least two probes: one goes into the meat, the other clips to the grate. That second one, the chamber probe, is what genuinely replaces the thermometer on the lid.

The cable-free wireless probe is a cylinder of ceramic and steel that goes into the meat whole. It carries two sensors: one at the tip, reading the centre of the cut, and one at the tail end, reading the air around it. No cable crosses the gasket and you can turn the meat without untangling anything, which is precisely the problem it was built to solve.

And there is a fourth device sold on the same shelf that does not measure the same thing: the infrared gun. It reads the temperature of a surface from a distance, without contact. It is perfect for knowing how hot the pizza stone or a cast-iron plancha is, and completely useless for knowing whether a chicken is done, because it sees through nothing.

Quoted accuracy is the figure everyone looks at and the one that decides least. Between ±0.5 °C and ±1 °C there is little practical difference on a brisket that will spend ten hours inside and that you are going to pull by feel as much as by number. Where it does show is on short, expensive cuts: fish, solomillo (fillet), magret. Look at response time too, because a tip that takes fifteen seconds to settle forces you to hold the lid open for fifteen seconds, and in ceramic you pay for that.

The number of probes is the most practical decision. Two is the functional minimum, because one of them has to go to the chamber. Four start to make sense when you put two cuts in at once, or when you watch a rack of ribs at both ends. And a question almost nobody asks before paying: whether the manufacturer sells replacement probes separately. The probe is the consumable in the system; if the only thing on sale is the complete pack, you have bought a disposable device.

Maximum ambient temperature is the figure that saves the most grief, and it is worth reading closely, because many spec sheets give two different numbers: one for the probe and one for the cable. The lower one rules. Wired probes in everyday use quote something like 300 °C; wireless ones, between 275 and 290 °C ambient. Translated: neither of them stays inside while you bake a pizza at 380 °C, nor while you sear with the vents wide open. They come out first.

The rest are details you will be grateful for six months in. That the base unit has a screen of its own and does not depend on the app alone, because a router going down should not leave you without an alarm. That alarms can be set per probe. That the body survives a splash. And that the probe is wiped with a damp cloth without wetting the connector: submerging the cable is the number-one cause of dead probes.

The ranges printed on the box — ten metres for class 2 Bluetooth, fifty or a hundred metres on models with a repeater — are measured in open ground with no obstacles. That is the small print that changes everything, because between your probe and your phone sits the worst thing you can put in front of a 2.4 GHz signal.

Inside a closed ceramic the direct link between probe and phone struggles, and it is not the device's fault: the wall of a kamado is two or three centimetres of dense ceramic, with steel bands around it and a metal lid on top. It is a cage. Which is why wireless systems include a repeater block that has to stay plugged in and reasonably close — a few metres away, and better with one window in between than two partition walls — and why real usable range bears little resemblance to the spec sheet. Models that also go up to the cloud, like the MEATER Pro 2 Plus, solve long range but move the dependency to your router: if the wifi drops, the alarm drops with it.

Wired systems do not have that problem because the probe transmits nothing. The cable carries the signal out, and the base unit, sitting a metre from the kamado in open air, does have a clean line to you or to the router. An Inkbird IBBQ-4T with four probes lets you follow the brisket from indoors for the same reason any smart plug works. The toll is physical: a cable crossing the gasket every time you cook.

On battery life there is no real contest either. Wireless spec sheets quote something like 24 hours per charge, ample for a smoke as long as you start with the probe full; a base unit on alkaline batteries lasts months. If you cook for twelve or fourteen hours, charging the probe the night before stops being fussiness and becomes part of the plan.

The principle is always the same, and it explains nearly every mistake: the sensor is not spread along the probe, it sits in the last three to five millimetres of the tip. You are looking for the geometric centre of the thickest part, which is the point that cooks last. Go in from the side and horizontally: that way the body of the probe stays inside the meat instead of exposed to hot air, and you can pull back a centimetre hunting for the lowest reading, which is the true one. Three things falsify the measurement: bone, which conducts and reads high; seams of fat, which read low; and pockets of air.

In a brisket, the probe goes into the thickest part of the flat, entering from the side at mid-height, never into the fat seam that separates flat and point, nor into the point itself, which always runs ahead. And a warning that saves disappointment: the final number matters less than feel. The probe has to slide in like butter; if at 96 °C it still meets resistance, the meat is asking for more time.

In a whole chicken or a turkey, the thickest part of the thigh, from the outside and aimed at the centre, dodging the femur, just behind it. The breast is measured separately, from the side and parallel to the board. Two readings, because they are two different targets, and the breakdown by cut is in the temperatures and times cheat-sheet.

In a chuletón or a tomahawk it goes in through the edge, horizontally, to the centre, well clear of the bone and the outer fat; on a reverse sear it stays in through the low phase and comes out before searing. And the honest limit: in a large whole fish you measure alongside the backbone, in the thickest part of the loin, but in a sardine, a gambón (large prawn) or a thin fillet there is nowhere to house five millimetres of sensor. There, eye and clock rule.

The ice test is the more reliable of the two and the one to do first. Fill a glass to the brim with crushed ice, top it up with cold water until the ice is covered, stir and leave it two or three minutes to settle. Put the tip into the middle of the ice, without touching the bottom or the sides, and stir gently. It has to read 0 °C within whatever margin the spec sheet quotes, and the beauty of the method is that this zero depends neither on altitude nor on the weather.

The boiling water test checks the other end, and there geography does rule. Water boils at 100 °C at sea level and loses one degree for every 300 metres of altitude: in Torrevieja, down at beach level, the target is 100 °C; in Madrid, at 667 metres, about 97.8 °C. Get it to a rolling boil, put the tip into the middle of the pan without touching the bottom — the metal is hotter than the water — and keep the handle and the electronics out of the steam.

With the deviation in hand there are three routes. Many digital instant-reads have an adjustment or factory-reset function described in the manual. The ones that do not are used just the same, by noting the error on a label stuck to the handle: one that always reads 1.5 °C high is perfectly usable, because a known error stops being an error. And the bimetallic dial on the lid can almost always be corrected by turning the nut behind the face with a spanner, with the stem sitting in the boiling water.

How often: when it is new, after any knock or fall, at the start of the season, and before a long cook where you will have no chance to correct course. And one sign of death: if the reading wanders when you move the cable, or takes time to settle where it used to be instant, the probe is finished and no calibration will save it.

There are four profiles, and almost everyone is in one of them. If your kamado mostly sees live fire — chuletón, fish, vegetables, burgers, pizza — the instant-read spike is not the first step: it is the only one you need. Those cooks last between six minutes and an hour, you are standing right there, and a leave-in probe with an alarm gives you nothing that opening the lid and spiking the meat does not. Climbing to the premium tier here gives nothing back.

If you are starting out and know you will smoke something now and then, the right set is an instant-read plus a two-probe wired kit with a base unit that has a screen. One to the chamber, one to the meat, and the number governing the cook stops being the one on the lid. It is the biggest jump in results a single accessory can deliver, which is why the probe appears among the essential accessories from day one.

If low & slow is your ground — racks of ribs, brisket, paletillas (pork shoulders), turkeys — and you leave the house while the kamado works, then it does pay to move up to four probes with wifi. Not for accuracy, which is the same, but for the number of points you are watching and for being able to look at the curve from work. Logistics justify the jump, not thermometry.

And if the cable gets in your way, if you turn cuts, if you cook in a cocotte or if you are in the habit of moving meat between zones, the cable-free wireless probe solves a real problem rather than a whim. It is only worth remembering what it does not do: it does not stay inside above its ambient limit, each cut needs its own tip, and all of it depends on a charged battery. The sensible order, if you are not buying everything at once, is instant-read first, leave-in later. Plenty of people do it the other way round and end up with a handsome gadget and no way to check a chuletón.