Stuck at a Decent Roast? The Level Isn't the Only Kaffelogic Dial

A Kaffelogic Nano 7 gets you to a decent roast very quickly. Pick a sensible profile, pick a level, and the first cups are good. Then the roasts stop getting better. You nudge the level, try another profile, and the cup stays roughly where it was.

I hit this on a natural Colombian Pacamara grown at 1800 m. This post is what my logs showed, and the way of iterating I now use instead: change one phase of the profile, not the level and not the whole profile.

The machine does what the file says

A Kaffelogic follows its profile closely. In the sample roast we ship with the log viewer, an F-WSH 1500-2200 roast at level 2.3, the bean temperature stays within about 1.3 °C (median) of the profile's target curve.

That is good engineering, and it means something practical: the curve in the file is, very nearly, the roast you get. If two roasts have nearly the same curve, they are nearly the same roast, whatever numbers you typed.

A level is not a roast degree

Each profile file carries a ladder of seven end temperatures, one per level from 0 to 6. The level you set picks a point on that ladder. On F-Series profiles one level step moves the end by 2 °C. On others it moves it by about 1.1 °C. The roast levels post has the full ladder story.

On my usual filter profile, levels 1, 2 and 3 end at 221.8, 222.9 and 224.0 °C. So moving from 2.1 to 2.4, which feels like a real step darker, moves the end by about 0.3 °C. That is effectively nothing.

Six roasts, one point tested

Here is what that did to the Pacamara. Six logged roasts, on two profiles (Bright Filter and Gentle Filter), at levels 2.1 to 2.4:

• end temperature 221.6 to 224.4 °C
• hot zone (time at 200 °C or hotter) 149 to 165 s
• 150 °C reached at 123 to 132 s

That is six roasts and one point tested. Every roast landed in the same small box. The cups agreed: fruity, but astringent as soon as the brew extracted a little more. One of those roasts has six logged brews that swing between fruit and astringency. When small brew changes swing the cup that much, the roast is the suspect, and I had never actually changed the roast.

Why a new profile doesn't isolate anything

The obvious next move is a different profile. The trouble is that a different profile changes everything at once.

Across our profile library, 87 profiles have both a curve and a level ladder. A target of a 216 °C end with a 120 s hot zone is reachable on 41 of them. But those 41 roasts take from 353 to 642 s in total, and reach 150 °C anywhere from 62 to 283 s. Same end, same hot zone, completely different roasts before 200 °C.

So if the cup improves on the new profile, you have learned that something improved. Not what.

It also helps to know what profile designers agree on. Of the 56 library profiles with a recommended level, filter profiles end at a median 215 °C with a 118 s hot zone in about 7:20. Espresso profiles end at 226 °C with a 202 s hot zone in about 9:44. Washed profiles (16) and natural profiles (7) come out at 214 °C / 101 s and 214 °C / 102 s. Designers separate by brew method, and hardly by process. So "pick the natural profile" is not much of a lever either.

Four dials, not one

A roast has three phases we measure, plus where it stops:

1. Start: charge to 150 °C
2. Middle: 150 °C to 200 °C
3. Hot zone: 200 °C to the end
4. End temperature

That is four dials. The level only touches the fourth, and on a flat-ended curve it doesn't even do that cleanly. A step worth measuring is about 4 °C of end temperature or 10 s or more of hot zone (the thresholds in our profile selection guide). My six Pacamara roasts never moved any dial by that much.

Published research backs the idea that more than the end point matters. Münchow and colleagues (Beverages, 2020) report that both roast colour and roast time drive flavour, and that fruitiness and acidity fall as development lengthens. Time is a dial in its own right.

One-dial variants

So we built a way to move one dial and leave the rest alone. Take a profile file and either stretch or shrink exactly one phase in time, keeping the shape of the curve, or move the end temperature (the whole ladder moves and the hot zone is kept).

Measured on two library profiles, in seconds for start / middle / hot zone:

Each change moves its own phase and nothing else. Compare that with the level: raising only the level by 4 °C on these two profiles also adds 50 to 60 s of hot zone, because the curve is nearly flat near the end. The level dial is not even a clean end-temperature dial.

There are limits built in. A phase stays between 60% and 160% of its length, the end temperature moves at most 8 °C, and the fan curve and boost zones move with the roast curve. The variant tool is available first to an AI agent connected to your BrewedLate account, which can make the variant and keep it with your profiles.

What I'm not claiming

Some honest limits:

• The variants are checked against the app's roast predictor, not yet proven on a roaster at the time of writing. The first roast on a variant is the real test.
• Shrinking a phase asks the heater for more power. A short phase on a full batch may simply not be reachable.
• Which dial fixes which fault is craft consensus, not proven. For "astringent when I extract a bit more", the usual suggestion is a longer hot zone at the same end temperature first, then a slower middle. That is the order I plan to try on the Pacamara.

How to iterate from here

• Measure the roast you have. Open your .klog in the log viewer and read the hot zone and the end temperature, and where the curve crosses 150 °C on the chart.
• Check whether your roasts actually differ. If your last few roasts sit within a couple of degrees and 10 to 15 s of hot zone, you have tested one point, however many levels you tried.
• Change one dial per roast. Keep the profile, move one phase or the end, and log it.
• Judge after day 12 of rest, on the same brew recipe. An early cup tells you little (see how long to rest coffee after roasting).
• Keep every roast and cup together. The roast tracker stores the time to 150 °C, the hot zone and the end temperature of every imported log next to its tastings, so you can see which dial moved the cup.

Related Reading

• Kaffelogic roast levels in degrees: why a level number only means something with its profile
• Roast or brew? Escaping the coffee death spiral: telling a roast fault from a brew fault before you change anything
• Kaffelogic Nano 7 profiles: the profile families and where to get them