We regularly encounter projects where HVAC controllers (Danfoss, Honeywell, Daikin VRV) use different protocols. One site — Modbus RTU on RS-485 for air handling units, BACnet/IP for chillers, a proprietary protocol for Daikin fan coils. Our mobile app works with normalized data through an API gateway, so we pay special attention to correct interpretation of source data. An error at the parsing stage can lead to emergency situations: for example, if the supply temperature is read as unsigned instead of signed, the system might interpret -10°C as 65436°C and shut down heating. Over the years, we have accumulated a database of typical configurations for most common controllers.
Key parameters and their sources
Minimum set for climate monitoring:
| Parameter | Source | Unit | Frequency |
|---|---|---|---|
| Supply/return temperature | PT1000/NTC sensors on pipeline | °C | 30 s |
| Zone air temperature | Room sensor or thermostat | °C | 1 min |
| Humidity | SHT31 or HIH6130 in zone | %RH | 1 min |
| Setpoint | Controller | °C | on change |
| Compressor status | Controller digital input | on/off | on change |
| COP (coefficient of performance) | Calculated on backend | — | 5 min |
Important nuance: temperature in Modbus Holding Registers often arrives as signed int16 in units of 0.1°C. If the controller outputs 0xFF9C, it is not 65436°C — it is -100, i.e. -10.0°C. Incorrect interpretation is a classic source of the "sensor shows -3200°C" error.
fun parseModbusTemperature(rawValue: Int): Double { // Convert unsigned 16-bit to signed val signed = if (rawValue > 32767) rawValue - 65536 else rawValue return signed / 10.0 } According to the Modbus Application Protocol Specification v1.1b, register addresses start at 40001, but in the gateway API they may be offset. Therefore we always verify register mapping against the controller documentation.
Android implementation: polling via Retrofit + coroutines
The gateway (Node-RED or custom Go service) provides a REST API. Polling with an adaptive interval — aggressive when the app is in the foreground, sparse in the background:
class HvacPollingService( private val api: HvacApi, private val repository: HvacRepository, ) { private var pollingJob: Job? = null fun startPolling(scope: CoroutineScope, foreground: Boolean) { pollingJob?.cancel() val interval = if (foreground) 15_000L else 60_000L pollingJob = scope.launch { while (isActive) { try { val data = api.getHvacStatus() repository.update(data) } catch (e: IOException) { // Log, don't crash — loss of gateway connection is normal } delay(interval) } } } } For sites with an MQTT gateway (Eclipse Mosquitto), we use org.eclipse.paho.client.mqttv3. Topics by zone: hvac/{buildingId}/{unitId}/temperature, hvac/{buildingId}/{unitId}/setpoint. MQTT delivers changes almost instantly — 10 times faster than polling with the same network load.
| Parameter | Polling (REST) | MQTT |
|---|---|---|
| Delivery latency | 1–15 s | < 0.5 s |
| Network load | High | Low |
| Reliability | Depends on interval | QoS 1/2 messages |
| Implementation complexity | Low | Medium |
Trends and history
A week-long temperature graph is a mandatory element. For long-period data, we request aggregation on the server (avg/min/max per hour), avoiding raw 30-second records. In the app we render via MPAndroidChart (Android) or fl_chart (Flutter):
LineChartData buildTemperatureChart(List<TemperatureReading> history) { return LineChartData( lineBarsData: [ LineChartBarData( spots: history.asMap().entries.map((e) => FlSpot(e.key.toDouble(), e.value.temperature)).toList(), isCurved: true, color: Colors.blue, dotData: FlDotData(show: false), ), ], titlesData: FlTitlesData( bottomTitles: AxisTitles( sideTitles: SideTitles( showTitles: true, getTitlesWidget: (value, meta) => Text(formatHour(history[value.toInt()].timestamp)), ), ), ), ); } Alerts on out-of-range values
Temperature exceeds limits — notification needed. We configure rules on the server (e.g., via Node-RED or TimescaleDB continuous aggregates), push via FCM. In the app, we store alert history locally in Room/SQLite — the user must see when and what happened, even if the notification was dismissed.
Why is correct protocol data interpretation important?
Even with correct register reading, discrepancies can occur: different controllers use different byte order (little-endian vs big-endian) and different data encoding. For example, some controllers transmit temperature in degrees Fahrenheit with a multiplier of 100, not 10. Our experience shows that 30% of projects have inconsistencies between documentation and actual protocol. Therefore, we always perform a test poll of all registers and cross-check against reference sensors. This ensures the app shows correct data from day one.
How do we ensure uninterrupted monitoring?
The key solution is communication channel redundancy. If the primary gateway via Modbus is unavailable, the app automatically switches to a backup BACnet/IP or Cloud API. We use the Chain of Responsibility pattern with timeouts. Additionally, we configure local data caching on the device for network loss. In one of our projects for a shopping center with 200 monitoring points, we guaranteed notification delivery within 30 seconds at 99.9% server uptime.
What is included in development?
Our work includes:
- Documentation of protocols and data points (register addresses, conversion coefficients).
- App source code with comments and tests.
- Integration with existing systems (BMS, SCADA) via API.
- Training of customer staff on using the app.
- Technical support for 3 months after launch.
Process: audit of existing controllers → integration architecture design → mobile app development (iOS/Android) → bench testing → on-site deployment. Timeline: 4 to 6 weeks for a typical project. Cost is calculated individually after analyzing protocols and number of monitoring points. Contact us to discuss details and get a consultation — we will analyze your infrastructure and propose the best turnkey solution.
Thus, we guarantee reliable HVAC monitoring with data accuracy up to 0.1°C and emergency response time under 30 seconds. Entrust climate control to professionals with 5+ years of experience and over 20 successfully completed projects. Get a consultation for your project today.