4 Radiators Balanced with a Fluke 62 MAX Infrared Thermometer on a Two-Storey Loop

Point the Fluke 62 MAX at the top-left corner of the hottest downstairs radiator and it will read something like 68 to 72 degrees Celsius on a system running at a 75 degree flow. Move it to the bottom-right and the number drops. That drop is the whole game. On a balanced radiator the difference between the flow side and the return side sits around 11 to 12 degrees. On an unbalanced one nearest the boiler, the return can still be running at 60 plus, which means water is racing through without giving up its heat, and the far radiators upstairs are getting the leftovers.

The two-storey loop makes this worse than a single-floor job. Gravity is not the issue on a pumped system, but pipe run length and the natural preference of water for the path of least resistance are. The radiator closest to the pump gorges itself. The one at the end of the upstairs run, forty feet of 15mm pipe later, gets a trickle.

Why the infrared reading beats guessing by hand

Back of the hand on a radiator tells you hot or not hot. It tells you nothing about the eleven-degree spread you are actually chasing. The Fluke 62 MAX has a distance-to-spot ratio of 10 to 1, so held about 150mm off the panel it reads a spot roughly 15mm across. Tight enough to compare the flow tapping against the return tapping on the same radiator without the readings bleeding into each other.

The trick is emissivity. Painted radiator surfaces sit around 0.95, which is close enough to the Fluke’s fixed 0.95 setting that you can ignore the correction entirely. Bare chrome towel rails are a different story, they reflect and the gun will lie to you by ten degrees or more. For those, a strip of matt masking tape over the spot you are measuring fixes the reflection problem, and you read the tape once it has settled.

Mark two spots per radiator with a chinagraph pencil or a bit of tape: one at the flow valve tapping, one at the lockshield return tapping, both about 40mm in from the end. Consistency matters more than absolute accuracy here. You are comparing radiators against each other and each radiator against itself, so as long as you read the same spots every pass, the drift in the gun’s absolute calibration cancels out.

The forty-minute pass

Start cold. Open every lockshield valve fully, all four radiators, then close each one back by a full turn. That gives you a rough starting point that is never right but is never disastrous either. Fire the boiler and let the system come up to temperature for a good fifteen minutes before you touch anything, because early readings while the loop is still filling with hot water will send you chasing ghosts.

Now the sequence. Read the flow and return on the radiator nearest the boiler first. If that return is only eight degrees below its flow, the radiator is hogging flow, so nip its lockshield in by a quarter turn and wait five minutes for the system to settle. The far upstairs radiator, the starved one, gets its lockshield opened wider. You are robbing the greedy to feed the far.

Work the loop in order of distance from the pump. Each adjustment shifts the balance of every other radiator, which is why one pass is never enough. The second pass is faster because the numbers are closer. By the third pass you are making eighth-turn tweaks and the four return temperatures should cluster within two or three degrees of each other, each sitting the target 11 to 12 degrees below its own flow reading.

Write every number down. A three-column notepad, radiator name, flow, return, one row per pass. Memory is useless here because the differences are small and the readings look similar. The paper trail is what tells you whether pass three actually improved on pass two or just moved the problem sideways.

When the boiler control fights you

A Viessmann Vitodens running weather compensation changes the flow temperature as the outside air changes. That is the point of weather compensation, but it complicates balancing, because a reading you took at a 70 degree flow means nothing an hour later when the controller has dropped flow to 55 because the day warmed up.

Set the balancing session up against a fixed flow. On a Vitodens you can force a constant flow temperature temporarily through the service menu, or simply balance on a cold still morning when the compensation curve is holding the flow high and steady. The 11 to 12 degree spread you are targeting is a differential, so it holds true whether the flow is 55 or 75, but you cannot compare a reading taken at one flow against a reading taken at another. Fix the flow, take all readings in one window.

The payoff with weather compensation is real. Once the four radiators are balanced, the return temperature stays low enough that a condensing boiler actually condenses. A Vitodens only hits its high-fifties efficiency when the return water drops below about 54 degrees, because that is the dew point where flue gases give up their latent heat. An unbalanced system with a fat 62 degree return from the near radiator never lets the boiler condense properly, so you burn gas for a lower efficiency figure than the badge on the front promises.

The frozen condensate that stops the whole thing

None of this balancing matters if the boiler locks out on a frozen condensate pipe at six in the morning. The Vitodens dumps acidic condensate through a plastic pipe, and where that pipe runs outside in 20mm rather than the recommended 32mm or larger, it freezes solid below about minus four degrees. The boiler senses the blockage and shuts down.

The field fix is a jug of warm water, not boiling, poured along the external run, or a covered hot-water bottle held against the frozen section for ten minutes. Reset the boiler afterward. The permanent fix is lagging the external pipe in Class O foam and, where possible, rerouting it internally to a soil stack.

Heat loss upstream of the radiators

Balancing shares the heat out fairly, but it does not create heat, and it does not stop heat leaving. A two-storey house loses a large chunk of its warmth through the roof, which is why loft insulation to 270mm of mineral wool is the single measure that changes how hard those four radiators have to work. Under many national schemes, households on qualifying income-related benefits fall inside loft insulation grant eligibility, and the survey that establishes it usually starts with a look at existing depth using nothing more than a ruler pushed into the joist gaps.

Cavity walls are the quieter loss. A thermal imaging camera, or even the Fluke 62 MAX swept across an internal wall on a cold day, shows cold patches where the cavity fill has slumped or was never installed. A properly filled cavity reads within a degree or two of the surrounding wall. An empty one reads noticeably colder near the floor, because the wall is bleeding warmth straight to the outside air. That cold wall is a load the radiators fight every hour the heating is on.

The interaction is the interesting part. Reduce the fabric heat loss and the balanced radiators start overheating rooms, because they were sized for the old, leaky, heat loss. Which raises a question the infrared gun cannot answer on its own: once the loft and cavity are done, are four radiators of that size even the right emitters for the rooms they now sit in, or has the balancing job just made an oversized system share its excess more evenly?