How it works

A technical description for designers and tender participants. The numbers come from our own documents, not from a pitch deck.

Four layers and the override rule

This is the most important table in the whole system. It defines what must never move into software.

LevelWhere it runsWhat belongs thereWhy there
1hardware, outside the controllermanual-reset safety thermostat, motor and surge protection, pressure relief damper, actuator spring return, fire dampermust work even when the controller does not
2controller, on siteinternal pressure and its emergency path, wind and snow response, CO and smoke, temperature cascade, start sequences, generator start, ventilation interlockmust work without internet
3AERDOMA serverbookings, adaptive pre-heating, weather forecast, energy accounting, access rights, inspection records, notificationsneeds outside data, is not time-critical
4applicationcontrol screens, configuration, reports, customer bookings and ticketsthe human

The override rule: a lower number always overrides a higher one. The server can never allow something the controller has blocked, and the controller can never bypass the hardware path.

What we measure

PathMeasured continuously
Supplytemperature, pressure, air velocity, humidity, CO concentration, smoke detection
Extracttemperature, pressure, air velocity, humidity, CO₂ concentration
Hallmembrane pressure — operating and emergency sensor, temperature, humidity
Outsideweather station: temperature, humidity, atmospheric pressure, wind, precipitation, solar radiation

Plus the state of the unit: airflow pressure switch, the FAN / LIMIT / LIMIT2 thermostats, flue pressure switch, damper positions, protection devices and phase monitoring relay, motor thermal contact, drive status over Modbus, control power battery status, switchboard and plant room door contacts, lighting feedback, and the full status of the diesel generator including fuel level and running hours.

Control loops

  • Temperature — cascade

    Hall temperature is the master loop, supply air temperature the slave. This prevents overshoot given the large thermal inertia of the hall. The temperature the customer paid for stays within the safety limits — checked in the cloud at booking and again in the controller.

  • Internal pressure

    By fan speed and damper position. The emergency path on low pressure is hardware and does not depend on the control loop.

  • Wind and snow

    Pressure and fan speed raised along a pressure/wind curve. Above the second wind threshold the system cancels bookings and calls for evacuation. The controller switches snow mode on by itself when the forecast reports snowfall, the outdoor temperature is below the threshold and the precipitation sensor is wet — all three at once. The forecast alone does not control the hall; for strong wind it only alerts the manager. Evacuation is triggered by a measured gust above the threshold — immediately, without averaging. The pressure/wind curve only ever raises the pressure, it never lowers it.

  • Lighting

    Switched by sector through contactors, with feedback. No dimming — hall luminaires are switched by applying voltage and have no intensity control.

Entry by QR code

The controller does not verify a cryptographic signature. It holds a list of valid entries, which is both simpler and more robust.

  1. Every 24 hours — and on every change — the cloud sends the controller a list of hashes of valid tickets.
  2. The player holds the QR code to the reader at the door.
  3. The controller compares the hash against the list and checks the time window and the sector.
  4. If it matches and the hall is not in alarm, the lock gets a 3-second pulse.
  5. The event is logged locally and uploaded to the cloud at the next connection.

Permanent rights — manager, maintenance, service — stay on the list at all times, so staff can always get in. During a longer outage the validity of the last list is extended.

When the player has nothing to show the code on

Flat battery, broken screen, freezing weather. A PIN keypad does not solve it — the PIN would arrive by text message on the same dead phone.

  • Printed or photographed code

    The code is an image, and the reader at the door does not care whether it sees it on a display or on paper. Anyone heading out into the cold, or aware of a weak battery, prints it beforehand.

  • A new code on another phone

    The code exists nowhere else — the server holds only its fingerprint, so we cannot send the same one again. On another phone the player issues a new one, and the original stops being valid.

  • Opening from a distance

    Planned

    The operator unlocks from the application and the entry goes into the audit log: who opened, when and for whom. A phone number is on the door.

The hall has no staff, but the operator does exist. Players also usually arrive in twos — two dead phones at once are unlikely.

Interfaces

InterfaceUsed for
MQTT / TLScontroller ↔ AERDOMA server, telemetry every 5 s
HTTPS + WebSocketcontrol screens served from the switchboard, no cloud needed
Modbus TCP / BACnet IPexposing parameters to the customer building management system
Modbus RTU (RS485)variable speed drives, electricity meter, weather station
Modbus TCP to the air handling panelunit status and faults; writing only within the operator’s scope
HTTPS with the hall certificatedaily logs and the nightly controller backup to the AERDOMA server; the hall opens the connection
Front panel switcheslocal operation and manual mode, no display
SQLarchive of measured data

Modbus TCP and BACnet IP are off by default and enabled only where the customer asks for them. When running, they are bound to the switchboard network.

Network and service access

There is always our router in the cabinet. Only what sits on its WAN side changes: our SIM card, or a cable into the customer network.

RuleWhy
The tunnel is always initiated outbound by our routerthe customer opens nothing and configures nothing
No inbound portnothing that can be scanned from outside
The router does not bridge the customer network and our tunnelwe must not be a door from the internet into the customer network
A separate key per hall, /32 routeone hall cannot see the others
The tunnel is for service, not for dataoperational traffic goes over MQTT; the tunnel is up only for diagnostics
The switchboard is its own network segmentif the customer network fails, the hall keeps running
Firmware is signed and updated in the night from Sunday to Monday, in the hall's local time, only when there is no booking before the end of the windowthe controller accepts no foreign code and does not restart in the middle of a game
One signed release serves all halls of the same type; the hall’s identity is in its configuration, not in the firmwarean update reaches all halls at once and a hall never reports under someone else’s name
Once a day the controller sends its logs to our server and deletes them only after a verified writewe tune against real operation and the controller’s memory does not fill up
Every night the controller sends a backup, and when it does not arrive our service team gets in toucha failed controller is replaced and restored from the backup

I/O scope

Indicative, from the reference diagram. Exact counts depend on the number of sectors and air handling units.

ConfigurationDIAIDOBuses
1 air handling unit, 1 court4617134× Modbus RTU, 2× Ethernet
2 air handling units733034—

The electricity meter, weather station and drives sit on Modbus RTU, not on physical inputs. Of the 46 DI, 13 are front panel switches and 13 are diesel generator monitoring.

Archiving

The archive and the live control screen are not the same thing. Mixing them means either a flooded database or an unusable control screen.

QuantityInterval
Live telemetry to the control screenevery 5 s
Temperature, humidity, pressure and air velocity — supply and extract15 min
Supply fan power, CO and CO₂15 min
Outside temperature, humidity, atmospheric pressure, solar radiation60 min
Hall pressure and wind speed — for tuning the pressure/wind curve1 min
Phase voltages and mains frequency1 min
Wind speed and directionon a change of 1 m/s
Precipitationon occurrence
Alarm states and faultson occurrence

The controller

A freely programmable controller to IEC 61131. Configurable controllers running a manufacturer application are out of the question — adaptive pre-heating, booking logic and the energy model cannot be put into one.

The controller holds the safety and control logic. The control web interface, the archive, the MQTT bridge and the optional Modbus/BACnet server run on an industrial computer in the same switchboard. The split follows the layer table above.

Alarms

The catalogue has 29 entries across four priorities. Four critical ones shut down or evacuate and require a manual reset. The critical mains and wind gust alarms report and do not shut the hall down.

FaultSystem responsePriority
Low internal pressureemergency path, evacuation, manual resetcritical
Smoke detected in supply airimmediate burner shutdown, evacuationcritical
CO in supply air above thresholdburner shutdown, full ventilation, evacuationcritical
Safety thermostat trippedburner shut down in hardware, restart blocked, service recordcritical
Wind above the second thresholdbookings cancelled, full pressure, evacuationhigh
Measured wind gust above the evacuation thresholdinstruction to clear the hall immediately, without averaging or waitingcritical
Mains failurelights and heating off, air handling restarts step by step once voltage returnscritical
Undervoltage, overvoltage or frequency deviationreport naming the phase, the hall keeps runningcritical
Loss of switchboard supplydiesel generator starthigh
Flow / speed deviationsuspected broken belt or blocked filtermedium

A selection from 29. The full catalogue is part of the datasheet.

+421 907 762 373 Assess my hall