How Consistent Is the Kaffelogic Nano 7? 29 Repeat Roasts

Every Kaffelogic review says the roaster is repeatable, ours included. We wanted to put a number on Kaffelogic Nano 7 consistency, so we pulled every recipe in our roast log that we have run more than once on a Nano 7e and measured how far apart the repeats landed.

A recipe here means the same bean, the same profile, the same level and the same batch size. That gave us 29 roasts across 9 recipes, each repeated two to five times.

The short answer: repeats of a recipe finished within 2.3 seconds and 0.8 °C of each other. The cup scores for those identical roasts still differed by up to 1.5 points out of 5.

How close the repeats were

These are standard deviations between repeats of the same recipe, pooled across the recipes that had log data (22 to 24 of the 29 roasts, depending on the measure).

A typical roast in this set runs about seven and a half minutes, so 2.3 seconds is about half a percent. Two of the recipes repeated to the same second.

The first-crack marker moved more than the finish did: 6 seconds against 2.3. The marker is the moment the bean temperature crosses a fixed value partway through the roast, so the roasts varied more in how they got there than in when they finished. It is another reason not to read too much into small differences in development time ratio, which is calculated from that marker.

The cup varied more than the roast

The same recipes, scored after tasting:

Three of the four Kenya roasts have log data: they finished within 7 seconds and 1.8 °C of each other. The four were rated from 3.5 to 5. The two Colombia roasts were 2 seconds apart and a full point apart.

The roaster did close to the same thing each time, so the difference has to come from somewhere else. The likely candidates are the ones the roast log does not capture: how many days the coffee rested before it was tasted, how it was brewed that day, and the taster. One recipe, the decaf at level 3.4, scored 4.5 three times running, so consistent scores are possible. They are just not guaranteed by a consistent roast.

What this changes about how to roast

Compare recipes, not batches. Overlaying two logs of the same recipe shows two lines on top of each other. The differences worth looking at are between recipes: a different level, profile or batch size on the same bean.

Do not re-roast to check the machine. If a recipe disappointed, running it again will produce nearly the same roast. Change something.

Taste more than once before you judge a recipe. With scores moving a point or more on identical roasts, a single cup is weak evidence either way. Taste a recipe at a consistent rest time, and more than once if you can, before keeping or dropping it.

Make level changes big enough to see. On an F-Series profile a 0.1 level step is 0.2 °C of end temperature (see Kaffelogic roast levels in degrees). Repeats of an identical recipe already differ by about 0.8 °C, so on that profile it takes a step of about 0.4 to move the end temperature by as much as two identical roasts differ. Judge smaller steps over several batches.

What this data cannot tell you

It is one machine, one roaster and 29 roasts, all at 100 g or 120 g. It says nothing about other units, about small batches, or about consistency across a wide range of room temperatures. It measures whether the same recipe repeats, which is the claim people make about this roaster, and on our bench it does.

For the rest of our experience with the machine, including its limits, see the Kaffelogic Nano 7 review.

Related Reading

• Kaffelogic Nano 7 review — a year of weekly roasts on the machine
• Kaffelogic roast levels in degrees — what a level step is worth on each profile
• Development time ratio on a Kaffelogic — why DTR depends on the profile
• Roast logs — the full attempt history for six beans