Batch Size Scaling

Batch size changes how a Roest behaves mechanically, thermally, and diagnostically. The same profile can produce different sensor curves, crack behavior, pressure, roast time, weight loss, and cup profile when moved from 50g to 100g, 150g, 180g, or 200g. This page explains how to choose a batch size, how to scale profiles, and what secondary settings usually need to change with dose.

Core Rule: Batch Size Is Not a Linear Multiplier

Changing batch weight is not just “more coffee needs more heat.” It changes the air-to-bean ratio, how much of the drum is occupied, how well probes contact beans, how much hot air escapes to the exhaust, how much momentum the bean pile carries, and how aggressively the paddles move the coffee. Several contributors describe 50–100g as behaving more like a small air/fluid-bed roast, while 180–200g behaves closer to a fuller drum-style roast with more retained heat and more internal development 2 sources.

For that reason, profiles should not be transferred blindly across batch sizes. A 50–100g profile will not work unchanged on 180–200g, and a 180–185g profile can be too fast or too airflow-heavy when charged at 150g 2 sources. Batch changes should be treated as profile changes involving inlet temperature, power curve, airflow/pressure, and drum speed.

Practical Starting Points by Batch Size

Use this table as the canonical starting point for selecting a batch size. These are not universal recipes; they are community-supported working ranges and should be adapted to the machine, coffee, voltage, altitude, and airflow setup.

How to Scale a Profile

A profile should be rebuilt around the new batch size rather than linearly scaled. The most robust approach is to keep one variable fixed long enough to understand the new batch behavior, then adjust in a controlled order.

1. Choose the target batch size and stay there while dialing in. Changing even 5–10g can materially change timing and sensor behavior on some profiles source.
2. Match the profile type to the batch size. Use inlet/time or power/time for 50–100g when BT is noisy; BT/inlet profiles require enough coffee mass for more reliable BT feedback, with 120g+ commonly cited as the point where readings improve 2 sources.
3. Adjust heat application by batch size rather than applying a single percentage multiplier. Roest forum guidance quoted in the community suggests increasing a 100g profile by 5–10% power for 150g, but other experienced users warn that batch-size dynamics change the whole power/heat approach rather than simply scaling up or down 2 sources.
4. Re-check airflow and pressure for the new batch size. Fan percentages are not portable between machines; identical drum pressure may require very different exhaust percentages, even for the same nominal batch size 2 sources.
5. Re-select drum speed. For 100–120g, Roest guidance quoted by users recommends at least 55 rpm for energy efficiency; for 200g, guidance repeatedly drops toward 40–45 rpm to reduce beans getting stuck or entering ventilation exits 2 sources.
6. Confirm by taste, color, and weight loss rather than graph beauty alone. At 50–100g, good roasts may have unattractive BT/RoR curves because the probe is not consistently embedded in the bean pile 2 sources.

A useful working heuristic from community testing is that smaller batches need a gentler start and less dramatic late heat reduction, while fuller batches need more early energy but allow more reduction after drying because momentum carries forward 2 sources. For profile shape, one contributor describes peak power moving later for 50–100g, toward center/right for 130–150g, and earlier/left for 190–200g source.

What Changes When Batch Size Changes

Sensor behavior

Batch size strongly affects BT, ET, exhaust, and RoR interpretation. At lower weights, the bean probe may read a mixture of hot air and bean-surface temperature because the bean pile does not consistently cover the probe. This makes RoR especially noisy because RoR is derived from BT 2 sources.

At larger weights, BT often becomes more useful, but ET and exhaust can become less representative if beans expand into the probe area or alter airflow. Roest has been reported as saying that at 150g the air temp and exhaust sensors are no longer reliable; others report ET becoming useless around 180–200g or when beans touch the ET area 2 sources.

Yellowing temperature markers also shift with batch size. One commonly cited approximation is ~170–175°C BT for 100g, ~160–165°C for 150–165g, and ~150–155°C for 180–200g, while acknowledging that visual yellowing is subjective source.

Heat retention and exhaust behavior

With fewer beans, more hot air can pass through the chamber and out the exhaust; with more beans, the bean pile can restrict or “plug” the airflow path and retain more heat. This is why lower-dose roasts can show higher ET/exhaust temperatures even when less heat is reaching the bean mass 2 sources.

A fuller batch can also make the internal temperature environment more even by reducing empty air space. This is one reason some users report more body, sweetness, inner development, and less hollow/tea-like cup character when moving from 100g toward 150–200g 2 sources.

First crack detection

First crack sound and automatic detection are batch-size dependent. Small batches can amplify crack sound because the chamber is emptier, while 170–200g can dampen sound because the bean mass absorbs or blocks it 2 sources. At 50g, users may hear only a few scattered cracks; at larger batches, the machine may undercount cracks even when beans are audibly popping.

Because of this, several experienced users prefer manual marking, temperature-based dropping, or sensory/color/weight-loss validation over trusting crack count alone. See First Crack Management and Development Time and Drop Decisions.

Weight loss and color

Lower batch weights often show higher weight loss than larger batches under comparable roast goals. One user states that 100g roasts generally lose more than 180g roasts, and another reports lower weight loss at 190g than 100g 2 sources. Weight loss remains bean- and profile-dependent, so it should be interpreted alongside color, aroma, and cup results; see Weight Loss Targets.

Color comparison across batch sizes can be misleading. Roasts with the same or similar measured color can taste quite different, and 100g versus 150–190g batches may differ in body, sweetness, acidity, and aftertaste even when visual color appears close 2 sources. Use a consistent grind and meter workflow if comparing color; see Color Reading and Measurement.

Small Batches: 50g to 100g

Small batches are valuable for samples, experiments, and situations where green coffee is scarce. They are not simply “mini versions” of 150–200g roasts. Users repeatedly warn that 50–100g data can be janky, especially BT and RoR, and that cup results should be judged by cupping, brewing, aroma, color, and weight loss rather than graph appearance.

50g is widely framed as a cupping/sample size, not a brew-production size. Christopher Feran states that 50g is good for cupping only, and denis6004 says fast 50–100g sample roasts are for cupping/evaluation, not brewing/drip 2 sources. That said, several users have achieved useful 50g profiles, especially with inlet/time or carefully downscaled profiles.

100g remains common because many samples arrive in 100g quantities, but it is sensitive. At 100g, changes in airflow, inlet, RPM, and probe contact can produce large graph artifacts. For 100g workflows, some users ignore RoR, use longer smoothing intervals, or rely on inlet/time rather than BT/inlet feedback 2 sources.

Mid-Size Batches: 120g to 150g

The 120–150g range is the compromise zone. It gives better BT coverage than 50–100g while avoiding some of the mechanical and cooling complications of 180–200g. 125g is repeatedly cited as a useful or “ideal” size because it improves BT coverage, avoids impacting the ET probe, and works well for splitting samples 2 sources.

150g is frequently recommended for starting out because the graph data and roast behavior make more sense than at 100g, while remaining easier to manage than 200g. Patrickj2095 says 150g is nice for starting out, and other users report 120–150g as the range where graphs make the most sense source. If a 100g profile is being adapted to 150g, the community has quoted Roest forum guidance to add 5–10% power as a starting experiment, but this should be treated as a starting point rather than a universal scaling law source.

Larger Batches: 160g to 185g

The 160–185g range is often favored for drinking/brewing roasts. Users report better heat retention, more realistic BT data, more body and sweetness, and more traditional roast behavior than 50–100g. Denis6004 repeatedly describes 170–185g as his preferred range for personal drinking because it gives more consistent results and more inner development with less development time source.

For S100/L100-style machines, 180g ±5g is described by one experienced user as a sweet spot, while 180–185g is also described as a safe bet depending on bean size 2 sources. Another practical batch-planning compromise is 166.5g, which divides 500g and 1kg green quantities evenly and has been used as an adaptation of 185g profiles 2 sources.

Cooling and maintenance become more important above 150g. Stock cooling is described as acceptable for 50–100g, but too slow for 150–200g by one user; another user reports effective vacuum cooling of 170–185g batches in 30–45 seconds 2 sources. curated If using vacuum or external cooling, keep hot beans and chaff out of household vacuum motors and plastic dust containers unless the setup is specifically heat-safe and uses an appropriate separator or metal collection path. See Cooling and Between-Batch Protocol.

200g and Above

200g is possible, but it is not a neutral upgrade. It changes drum fill, airflow, RPM constraints, sensor usefulness, cooling, and risk of beans entering the chute or exhaust. Some users report excellent 200g results, more stability, clearer cracks, and deeper development; others call 200g suboptimal or too risky for their machine and workflow 2 sources.

For 200g, the dominant practical constraint is drum speed. Roest guidance is repeatedly quoted as recommending 40–45 rpm or below 40 rpm for 200g to keep beans from getting stuck or entering the exhaust/ventilation exit 2 sources. Some users report successful 200g batches at 35 rpm or 39 rpm, while others report beans in the chute or stalls depending on bean size and machine 2 sources.

200g also changes sensor interpretation. ET can become less useful because beans may cover or contact the sensor area, while BT can become very stable in some setups 2 sources. If using 200g, validate by taste, weight loss, color, and mechanical inspection rather than assuming the 100g or 150g graph logic applies.

Conflicts and Limits

CONFLICT (Unresolved): ideal batch size. Roest is quoted as recommending 125g as an ideal or best weight ratio, while some experienced users prefer 150g, 160g, 170–185g, or 200g depending on purpose. Christopher Feran describes the machine as best around 120–150g and 200g as suboptimal in his opinion; denis6004 often prefers 160–185g or 180g ±5g for drinking roasts; sorinl has reported preferring 200g for stability and complexity in some contexts 2 sources. The practical resolution is use-case based: 50g for cupping samples, 120–150g for reliable small-batch profiling, and 160–185g for many drinking/brewing workflows.

CONFLICT (Unresolved): 200g viability. Tom Roest and several users report 200g can work, especially with counterflow or correct RPM, while other experienced users warn that 200g causes beans in the chute, ET loss, stalls, and cooling problems on some setups 2 sources. Treat 200g as an advanced batch size rather than a default.

CONFLICT (Unresolved): scaling by percentage. Roest/forum guidance quoted by users suggests adding 5–10% power when scaling a 100g profile to 150g, but other trusted contributors argue that batch size changes the whole thermal and airflow system and cannot be treated as a linear up/downscale 2 sources. Use percentage changes only as a first test, then rebuild around sensory and log results.

Troubleshooting Batch-Size Problems

Batch Planning and Green Quantity

Batch size also affects how green coffee is divided. 100g is convenient for 1kg lots and common sample bags, while 125g divides 1kg into 8 batches, 150g works well for 1lb blocks as 3 × 150g, and 166.5g divides 500g or 1kg evenly into 3 or 6 parts 2 sources. Some users prefer 165–185g but accept odd splits or blending leftovers to avoid very small remainder batches 2 sources.

When learning, it is better to buy enough of one green coffee to repeat and compare roasts rather than buying many tiny lots. One recommendation is to buy 5kg of the same green instead of 5 × 1kg so the roaster can learn from repeated trials on one coffee source.

Related Pages

Batch-size decisions interact strongly with Airflow and Fan Settings, Pressure Management, Drum Speed / RPM Settings, Inlet Temperature Management, Bean Temperature Profiling, Cooling and Between-Batch Protocol, and Roast Defects Troubleshooting.

Example: 150g High-Density Washed Profile Adjustment

One shared reference profile was described as being intended for 150g batches of high-density washed coffees from origins such as Kenya, Ethiopia, Rwanda, and Colombia. In that context, the suggested target was reaching first crack in roughly 5:00–5:30. If the coffee does not crack in that window, the recommended adjustment was to raise all inlet-temperature points by about 10–15°C rather than changing only one segment of the profile. Treat this as a profile-specific tuning heuristic, not a universal rule, because roast timing will still depend on machine generation, airflow, batch behavior, and green-coffee condition. source

Anecdotal 50g Profile Note

One reported 50g profile attributed to Luca was described by Denis6004 as producing very good results, with the comment that it roasts “stupid good.” The same report claimed that, across all beans tried with that profile, first crack occurred very consistently at about 5:30, within approximately ±3 seconds. Treat this as profile-specific anecdotal guidance rather than a general rule for all 50g batches, since first-crack timing can still vary with bean density, moisture, processing, machine setup, and environmental conditions. source

Community Profile for Tiny Sample Amounts

A community-shared ROEST profile has been recommended as a starting point for sample roasting when only very small amounts of a coffee are available: https://front.roestcoffee.com/sharedprofile/f7631b2f-8844-49ff-bbea-60cc8507dbc0/  . It was described as created by Luca and reported to have worked well for several people. Treat it as a practical starting point rather than a graph-perfect reference; the same discussion cautioned that the resulting graphs may not necessarily “make sense,” which is consistent with the general warning that very small batches can produce unusual or less interpretable sensor curves. source

Practical Starting Point for Easier Dial-In

One practical beginner-friendly approach is to avoid starting with very small batches or difficult coffees. A suggested low-hassle starting point is a not-too-expensive washed, high-altitude coffee paired with an existing 150g inlet-temperature profile. One example mentioned is Denis 150g InletT washed. This was described as a relatively safe bet for achieving a good initial result. source

Smaller batches may be more difficult to get right, especially when using ET profiles. For early learning or lower-friction profile development, a 150g inlet profile may therefore be easier to work with than smaller-batch ET profiles. source

Beginner Batch-Size Starting Point

Community advice often points beginners toward about 150g as a practical starting batch size on ROEST. One experienced user summarized that “for starting out 150g is nice,” while another cautioned against starting below 150g because smaller batches can be harder to dial in. The stated reason was that, under roughly 150g, the bean probe may not be fully immersed in the bean pile, making the BT trace and related graph behavior less representative. source

A related practical observation is that graphs may be easiest to interpret around the 120–150g range. This does not mean smaller batches are unusable: at least one experienced user reported commonly roasting 100g power profiles, but the same discussion still framed 150g as the easier beginner starting point. source

Field Note: 50g Sample Profiles and First-Crack Detection

A reported test of the “50g sample Luca/Denis” profile on a washed Sidamo Gr2 grown around 2000m found that automatic first-crack detection did not behave as intended. The user reported early random cracks, stopped two roasts too soon, and judged those roasts clearly underdeveloped. On a third attempt they ignored the automatic FC trigger; the sustained/real cracking arrived a little before 5:30, and the roast was dropped at 6:00. The same report noted green moisture of 12.3%, weight loss of 13.6%—judged probably a little high—and an irregular RoR, which the user considered expected for such a small batch. source

Practical takeaway: with very small batches such as 50g, do not rely only on automatic first-crack detection or a smooth RoR curve. Treat isolated early pops cautiously, confirm FC by the pattern of repeated cracking and other roast cues, and avoid ending the roast solely because an automatic FC marker appears early. source

BT Reliability Around 100g

BT readings generally become more useful when the batch is large enough to keep the bean probe well covered and produce a stable, repeatable signal. One experienced user summarized this as: BT is “great when you use a large enough batch to get reliable BT readings,” and reported that it is “mostly working at 100g.” source

Treat 100g as a practical lower-middle range rather than a guaranteed threshold. Another user reported that, in their experience, profiles were still “wonky with BT/inlet at 100g,” suggesting that 100g can work for some setups and profiles but may still produce unstable or hard-to-interpret BT/inlet behavior for others. source

Practical implication: if BT-driven profiling is unstable at 100g, first verify probe coverage, airflow/fan behavior, and profile design before assuming the profile itself is wrong. If repeatability remains poor, test a larger batch size to see whether the BT signal becomes more reliable.

Community Example: Small Temperature Point Boost on a 100g Profile

One user reported raising the temperature points by about 3–5°C on a 100g profile. In that trial, first crack occurred around 9 minutes, while the desired target was closer to 8 minutes. This is a useful reminder that modest point increases may still leave first crack later than intended, depending on the original profile, bean, airflow, and machine behavior. source