Community Experiments and Research

Community experiments are most useful when they turn informal curiosity into repeatable comparisons. This page describes how Roest users structure tests, choose baselines, avoid confounded results, and decide when a finding is mature enough to trust. Detailed parameter guidance remains on the specialist pages for each variable, such as Inlet Temperature Management, Drum Speed RPM Settings, and Airflow and Fan Settings.

Research Mindset

Roasting experiments work best when they begin from a known baseline rather than from a complicated custom profile. A test profile can be used first to learn the machine, then profile development proceeds by changing one parameter at a time 2 sources. This keeps the result attributable: if the cup improves or worsens, the changed variable is easier to identify.

The opposite pattern—changing several settings at once—makes results harder to interpret. When a roast has moved from one setup to another by changing many variables, the useful next step is to return to stock or known settings and restart with a single controlled change source. This applies whether the variable is heat, fan behavior, drum speed, counterflow, charge temperature, or another profile control.

Community learning is practical and iterative. Asking someone with the same roaster can provide a starting point, but users still need to test on their own machine and taste the results. For shared starting profiles, see Profile Sharing and Starting Points.

A Repeatable Experiment Protocol

Use this as the canonical workflow for community-style Roest experiments.

1. Define the question.
   State the single thing being tested: for example, whether higher RPM shortens roast time, whether a counterflow profile needs lower inlet, or whether a different drum temperature changes the cup.

2. Choose a baseline.
   Start from stock settings, a stable personal profile, or a simple test profile. On Ultra, one experienced approach is to keep stock settings and build profiles around those settings, especially because Ultra hardware changes are more than just interface and screen updates source. For broader Ultra context, see Roest Ultra Guide.

3. Change one variable.
   Make one intentional change and leave the rest of the profile alone. Avoid building a profile that depends on many manual interventions during the roast; a new-machine profile with 20 manual step changes is difficult to repeat reliably source. curated For a clean comparison, keep the coffee itself and batch conditions the same wherever possible: same green coffee, same batch size, similar warm-up/preheat state, and similar ambient conditions. Otherwise it is easy to attribute a difference to RPM, inlet, fan, or PID when the coffee or setup changed instead.

4. Choose an increment large enough to observe.
   Very small changes can make experimentation slow, but very large changes can move the roast outside a useful range. Community examples include testing a +10 RPM change, trying a 10°C temperature change that was already tasteable, or making a 40°C decrease to reach a workable counterflow range before testing other variables 2 sources.

5. Roast and record the result.
   Record the changed parameter, batch size, profile behavior, and any observed change in timing, temperature, or power demand. If testing RPM, one proposed check is whether a +10 RPM change makes the roast faster at the same inlet; if it does, the beans may be reaching higher temperatures under that condition source.

6. Cup against the baseline.
   Sensory comparison is the final filter. A change that looks promising on a graph should be cupped against the previous setting or baseline before it is treated as an improvement. For evaluation methods, see Cupping and Sensory Evaluation.

7. Decide, then continue.
   Keep the change, revert it, or test the next variable. If the result is unclear, repeat before stacking additional changes.

Choosing Variables to Test

The most common community experiments change heat application, air movement, drum behavior, or control logic. Each should be treated as its own experiment rather than bundled into a multi-variable rewrite.

Working With Stock Settings

Stock settings are useful because they provide a stable reference point. They do not need to be treated as permanent roasting doctrine; they are a controlled starting condition. A beginner or a user exploring a new machine can learn from a simple profile first, then branch out through deliberate single-variable changes.

This is especially important when the machine itself changes. Ultra should not be assumed to behave like an older Roest model with only a new screen, because community observations describe broader part changes source. When comparing old and new hardware, keep the baseline conservative and document the machine version.

PID and Profile Following

PID experimentation belongs in the same controlled framework as roast-profile experimentation. If a user wants the machine to follow a profile more closely, PID parameters may need adjustment, but each adjustment should be isolated and tested. One community PID experiment changed P from the default 6 to 4 and then to 2, with the tester later saying 4 and 6 worked best in that context source.

Treat PID findings as machine- and profile-specific unless repeated. A setting that improves following for one profile may not be a universal recommendation.

Counterflow Experiments

Counterflow testing is a good example of why experiments need a workable range before fine comparisons begin. One counterflow profile was adjusted with a 40°C decrease specifically to reach a usable range before continuing with other variables source. Another 100g comparison between counterflow and classic drum direction reported substantially lower inlet and power requirements in counterflow, but that result should be treated as a test observation rather than a universal conversion factor source.

Counterflow control can also raise workflow limitations. One tester wanted the ability to change counterflow in the middle of a profile, which indicates that some experiments may be constrained by what the software can adjust during the roast source.

Extreme Tests and Fine Tuning

Large changes can reveal whether a variable matters, while smaller changes refine the final setting. Some experimenters prefer to move “more to extremes” first to make differences visible, then fine tune afterward, rather than spending many roasts on tiny increments source. This approach is useful when the experiment’s goal is discovery, but the result still needs to be brought back into a practical range and verified by tasting.

Machine-tilt experiments fall into the exploratory category. One proposed test involved tilting the machine 20–40 degrees for 50–100g batches source. Such tests are not starting-profile guidance; they are examples of community research intended to expose mechanical or heat-transfer effects. curated Mechanical experiments such as tilting an operating roaster are advanced and safety-sensitive: for normal use, keep the roaster level and stable, and do not run it tilted unless the setup is manufacturer-approved, mechanically secured, and assessed for instability, hot bean or chaff spillage, altered exhaust or airflow behavior, ventilation, fire, burn, tipping, and warranty risks.

Comparing Machines or Roaster Models

Machine comparisons require much more evidence than one or two roasts. A meaningful comparison project can take months or hundreds of roasts, especially when one roaster has been in use for 8–10 months and another has not yet been thoroughly tested source. Early impressions can be useful, but they should not be treated as settled conclusions.

For practical comparison work, the same rules apply: establish a baseline, control variables, repeat results, and evaluate by cup quality rather than profile shape alone.

Sensory Validation

The community’s experimental approach ultimately works backward from the cup. There is no single finalized consensus profile that every coffee should follow, so profile decisions need sensory confirmation. Cupping recent changes against older settings is a practical way to determine whether the experiment improved taste inferred.

For structured tasting, defect identification, and comparison procedures, use Cupping and Sensory Evaluation and Roast Defects Troubleshooting.