From the tools: three analyser readings that made me stop and recheck

A flue gas analyser printout tells you more than pass or fail. The readings worth respecting are the ones that are in range but different: a ratio that has crept up since last year, a number too clean to be true, a result that doesn't match the appliance in front of you. Here are three from my own jobs.

Why reread an analyser printout at all?

Because the habit most of us learn is take the reading, staple it in, move on. The analyser said fine, so the job is fine. Most days that is true. But the printout costs you nothing to actually read, and about once a season it has told me something the pass mark did not. Ten seconds before you pack up, with the appliance still in front of you, because that is the only moment a strange number is cheap to chase. Back at home it is just a piece of paper. That is the whole discipline.

The ratio that was fine but rising

A combustion ratio inside the acceptable range, but noticeably higher than the same appliance read last year, is not a fail. It is a story. Something has changed, a bit of soot, a slightly disturbed burner, a flue starting to struggle, and the printout caught it before it became a defect. I logged the reading, told the customer in plain words what I had seen and why it was worth watching, and the next visit told me I was right. The detail of what the ratio means lives in the guide; the field lesson is to read this year against last year, not just against the threshold. A threshold tells you where you are. The trend tells you where you are going.

The reading that looked too perfect

Numbers that come back suspiciously clean can mean the probe is not where you think it is, or the analyser has not been zeroed properly in fresh air. I have caught both, and both produce a printout that looks like the healthiest boiler in Britain. When a reading does not match what my eyes and ears are telling me about the appliance, I trust the appliance and recheck the instrument: back out to fresh air, zero it again, get the probe seated properly, run it again. It is a few minutes, and either the second reading agrees with the first or you have just saved yourself from filing a number that was never real. A printout is only as good as the sample behind it.

The one I didn't like the feel of

Occasionally everything is in range and something still is not right: the smell, the flame picture, the way it is cycling. I have stood in front of an appliance holding a clean printout and not believed it. On those jobs the reading goes in the record and the judgement stays with me, which is exactly where it should be. The printout is one input, not the verdict, and nothing on it obliges you to switch your instincts off. The analyser measures; the engineer decides.

What I do with the printout now

It goes in the record, with the job, every time. Not because anyone will ever ask for it, though they might, but because next year's visit needs a baseline. The fine-but-rising ratio above was only visible because last year's number was sitting in the same place as this year's; a printout filed in a shoebox cannot tell you anything is creeping. Keep the reading where the next visit will actually see it, and the analyser stops being paperwork and starts being a memory. The customer gets a better service out of it too: watching this for you since last year lands very differently from a shrug and a pass.