Settings
Everything that changes what a run does, grouped by what it affects.
Radio presets#
A preset is an agreed set of LoRa parameters for a territory, taken from the MeshCore community's own list. It is baked in rather than fetched, so the workbench works offline and a saved scenario keeps the table it was built with.
| preset | MHz | kHz | SF | CR |
|---|---|---|---|---|
| Australia | 915.8 | 250 | 10 | 4/5 |
| Australia (Narrow) | 916.575 | 62.5 | 7 | 4/8 |
| Australia (Mid) | 915.075 | 125 | 9 | 4/5 |
| Australia: SA, WA | 923.125 | 62.5 | 8 | 4/8 |
| Australia: QLD | 923.125 | 62.5 | 8 | 4/5 |
| Brazil | 923.125 | 62.5 | 8 | 4/8 |
| EU/UK (Narrow) | 869.618 | 62.5 | 8 | 4/8 |
| EU/UK (Deprecated) | 869.525 | 250 | 11 | 4/5 |
| Czech Republic (Narrow) | 869.432 | 62.5 | 7 | 4/5 |
| EU 433MHz (Long Range) | 433.65 | 250 | 11 | 4/5 |
| EU 433MHz (Narrow) | 433.65 | 62.5 | 8 | 4/8 |
| Netherlands | 869.618 | 62.5 | 7 | 4/5 |
| New Zealand | 917.375 | 250 | 11 | 4/5 |
| New Zealand (Narrow) | 917.375 | 62.5 | 7 | 4/5 |
| Portugal 433 | 433.375 | 62.5 | 9 | 4/6 |
| Portugal 868 | 869.618 | 62.5 | 7 | 4/6 |
| Switzerland | 869.618 | 62.5 | 8 | 4/8 |
| USA/Canada (Recommended) | 910.525 | 62.5 | 7 | 4/5 |
| Vietnam (Narrow) | 920.25 | 62.5 | 8 | 4/5 |
| Vietnam (Deprecated) | 920.25 | 250 | 11 | 4/5 |
EU/UK (Narrow) is the default. Set one with radio.preset, or per node in the Inspector.
Spreading factor is the dominant setting: each step up roughly doubles airtime and buys about 2.5 dB of link budget. A network on SF11 is slower and reaches further than the same network on SF8, and the two are not interoperable.
Provisioning#
What each node is told when its firmware starts. These are workbench settings rather than firmware settings: they decide which CLI lines are issued at boot.
| setting | what it issues | default |
|---|---|---|
| set name to the node's name on the map | set name <name> | on |
| set the node's position | set lat / set lon | on |
| set the clock, the same on every node | time <epoch> | on |
| define a transport region from the study area | region put / allowf / save | off |
| and make it this node's default scope | region default <name> | off |
| cap advert hops | set flood.max.advert <n> | on, 32 |
| stagger node start times | spreads boot across a window | on |
The clock comes from the scenario, not the host. MeshCore timestamps messages and judges freshness by them, so a mesh whose nodes disagree about the time behaves differently. Taking it from the scenario keeps runs reproducible.
Regions are off by default because they change which packets a node relays, and therefore change results. A node that already carries observed regions is always configured with them regardless of this switch.
Per node#
| field | notes |
|---|---|
| kind | repeater, advanced repeater, companion, room server, SDR observer, emitter |
| position | latitude and longitude; uncertainty is carried from an import |
| height above ground | matters more than power on most links |
| transmit power | capped by the board profile; a scenario asking for more is reported |
| board | sets antenna gain, feedline loss, noise figure, battery and panel |
| radio | frequency, bandwidth, spreading factor, coding rate |
| firmware role and version | which application, and which build |
| regions and default scope | what it relays, and what it scopes its own traffic to |
| forward flood traffic for any region | region allowf *; a fresh node already allows unscoped floods and the wildcard never matches a scoped one, so this changes nothing unless a node was told region denyf * |
| emitter duty | for interferers only: what share of the time it is keyed |
Simulation#
| setting | effect |
|---|---|
| seed | the run's randomness. Same seed, same scenario, same result |
| speed | how fast simulated time advances relative to the wall clock |
| step | the engine's tick, 10 ms by default |
| real firmware | whether nodes run MeshCore or inject traffic directly |
| excess path loss | a flat dB penalty, for calibrating against a known-bad reality |
Experiments#
| setting | effect |
|---|---|
| arms | what varies: firmware version, loop detection, path hash size, CAD mode |
| seeds | how many times each arm runs |
| senders | who originates; spread picks separated nodes rather than neighbours |
| scope | the transport scope the burst is sent on, written #sco |
| run for | simulated milliseconds per run |
| message size | payload bytes; 0 means a short label |
Energy#
Off by default, and not fully implemented yet: when on, a battery and panel are sketched from the board profile, but the numbers have not been validated against anything real. Do not use them for siting decisions.
Automation#
| setting | effect |
|---|---|
| let agents drive this workbench | opens the control socket at $XDG_RUNTIME_DIR/meshbench.sock |
| compute the link matrix in the background | keeps reachability current while editing |
MESHBENCH_CARTO_KEY in the environment | unlocks the CARTO basemaps and makes the dark map the first-run default; without it they answer every tile with an API-key watermark |
The control socket is how every scripted run, the test runner and the MCP server drive the application. See the control socket reference.