The term “data point” is a fundamental element when designing and specifying measurement and control systems — and control systems in general — used to determine the system’s size. From the perspective of a programmable controller or software, we are not particularly concerned whether the controlled pump at a heat exchanger has a power output of 250 W or 2 kW, or whether the valve on the heating branch connected to the controller output has a DN 25 or DN 80 fitting. What matters is the type of control signal required for its operation.
The signals connected to the inputs and outputs of the control system determine the types and quantities of these inputs and outputs — that is, the data points. From the designer’s point of view, a data point is an input or output of the control system, such as a measured temperature, humidity, or gas concentration; a signal indicating level status; a relay for switching a fan or pump; or an output for controlling an air-handling damper. The number of data points defines the size of the control system. A typical heat-exchanger station may contain 20–50 data points, a larger boiler room with heating circuits 200–300 data points — and an entire building up to several tens of thousands of data points.
Data points can have different types. The basic division is into continuous and two-state (binary) types and into inputs and outputs of the control system:
Sometimes also referred to as “UI,” Universal Input (not to be confused with “Universal Inputs/Outputs” below). This is a control-system input used for reading a continuous value such as voltage, current, or resistance. The signal type and measurement range depend on the hardware design of the input circuit and are usually switchable — either by hardware (jumper, DIP switch) or software from the configuration program, or by a combination of both. A typical voltage-input range is 0…10 V DC, though there are inputs that measure negative ranges (−10…10 V) or smaller voltages (±0.1 V, 0.5 V, 1 V, etc.). Most sensors used in building-automation systems provide a standard 0…10 V output signal, often with a switchable measurement range. For example, a measured pressure 0…100 Pa corresponds to an output 0…10 V, but the sensor can also be switched to another range such as 0…1000 Pa again corresponding to 0…10 V. Therefore, during commissioning, it is necessary to check the sensor range and set in the software the conversion from input voltage to the measured-value range accordingly. This, of course, also applies to current and resistance inputs.
For current inputs, the ranges 0…20 mA or 4…20 mA are used almost without exception. In the latter case, 0 mA at the input can be interpreted as a sensor fault, which — after evaluation in the control system — provides additional diagnostic information. Current inputs are more resistant to electrical interference and are therefore often used in industrial environments; in building-automation systems, the 0…10 V voltage signal is more common, mainly because of its lower cost and easier commissioning and diagnostics.
The most interesting inputs, or ranges, are those for measuring resistance. They are mainly used for measuring temperature with passive sensors. These are relatively inexpensive and therefore the “first choice” for temperature measurement unless there is a reason to use sensors with voltage or current outputs. Such a reason may be a large distance between the sensor and input, where cable resistance would be significant or there would be a risk of interference. Resistance inputs are the most sensitive to interference — they use a relatively small current (about 0.3 mA) so the sensing element is not heated by the passing current, which would introduce measurement error. The measured resistance is then converted to temperature using a linearization function. Each type of sensing element (Pt1000, Ni1000, etc.) has its own linearization function — but that’s another story.
These are terminals where the control signal appears for continuously controlled dampers, valves, frequency converters, or other peripherals. Voltage signals (0…10 V DC) are used, and in rare cases, current signals (0(4)…20 mA). Outputs are typically short-circuit-resistant, and the maximum current for voltage outputs is about 10 mA. Sometimes, several peripherals are controlled by a common signal — for example, in an air-handling unit, both the supply and exhaust dampers may share one control signal, i.e., one analog output. However, in such cases, it is important to carefully consider whether it might be necessary under certain conditions to control each one separately.
The name might be a bit misleading, as it may suggest some kind of data line. A more accurate term would be “two-state input,” sometimes referred to as “binary input” (BI), which is actually the most descriptive term: this input is designed for signals that can take one of two values — off or on. The signal source is a switching or breaking contact, such as a thermostat, pressure switch, or other “...stat,” a switch, a push button, an auxiliary contact of a contactor, or a contact of an isolating relay. Electrically, it can be either a so-called voltage-loaded input, to which voltage must be applied (usually the supply voltage of the module or PLC), or an input for a potential-free contact.
Indicator LED on binary inputs and outputs, Domat PLC
Binary inputs are usually equipped with indicators — when active (under voltage or with a closed contact), the corresponding LED lights up. In the PLC program, the input logic can usually be inverted (negated), so when the input is active, the logic in the program reads as “0.” Negated signals are used to monitor signal loop interruptions: when the cable is broken, the signal has the same value as in an alarm state (logic 0), alerting the operator that the peripheral device is malfunctioning.
Connection of a digital input with external voltage
In the image, we can see a digital input DI7, to which a pressure switch S4 on the AHU filter is connected. In the default (idle) state, the pressure-switch contact is closed, so the input is under voltage (24 V AC, G vs. G0) — active. When the pressure increases (indicating a clogged filter), the contact opens, and the input changes to “false.”
The input current draw is approximately 10–15 mA to prevent unwanted activation by induced voltage. The input circuits include surge protection and debounce filtering — very short signals in the order of milliseconds are ignored.
Sometimes also referred to as BO (Binary Output). Similar to DI, a digital or binary output represents a two-state signal, usually in the form of a switching or changeover relay contact. In some cases, you may encounter triac outputs or DC voltage outputs (24 V DC) switched by a transistor on the output card. A relay typically has a load capacity of 230 V / 5 A and may be used to directly switch smaller pumps and other devices, or as part of a control chain to switch the coil of a power contactor. The contactor, together with motor protection, then provides power supply and switching for single-phase or three-phase motors, electric heaters, or similar equipment.
Transistor outputs are designed to control DC relays or contactors. Outputs equipped with transistors (typically 24 V DC with a load capacity of 0.5 A) need to be supplemented with power relays, so the question arises whether these cheaper transistor outputs are overall more cost-effective than I/O modules with 230 V relays.
While mechanical relays have a limited lifespan defined by the number of switching cycles — typically in the hundreds of thousands — triac outputs, using semiconductor power components, have virtually unlimited switching life as they cannot wear mechanically. Therefore, they are often used for so-called quasi-continuous control of electric heaters or thermal valve actuators. This involves periodic switching with a period of tens of seconds and a duty cycle proportional to the required average heating power or valve opening degree.
In addition to these four basic types of data points, we may also encounter a counting input:
CI, or Counting Input, is a special type of binary input characterized by high-speed processing — up to the kHz range. In building control systems, the maximum pulse frequency for counting inputs is typically around 50 Hz, which is still much faster than the typical response of a digital input (DI), which may be up to 1 second. Counting inputs are used to count pulses from water meters, gas meters, and electricity meters. Even though we generally prefer communicative meters, sometimes a pulse signal is the only way to obtain consumption data: for controlling demand peaks, consumption pulses and quarter-hour synchronization pulses are a standard way for a billing meter to communicate with the building control system.
Although some digital inputs in software include a rising-edge counter, we use dedicated counting modules for data acquisition from meters. Pulses may be only a few tens of milliseconds long, and ordinary digital inputs might miss some of them.
The distance between the pulse source (meter) and the counting input should be as short as possible to prevent interference — counting inputs respond even to very short pulses, and voltage spikes induced on the line may be incorrectly interpreted as valid signals.
Some control systems, such as Siemens Desigo, offer universal input/output cards or modules that can be configured in software as inputs or outputs (DI, AO, AI for voltage, current, or resistance). This provides great flexibility, especially when the system is later expanded — see the section on Reserves.
Data points in a measurement and control (MaR) project are clearly visible in the process diagrams. The table at the bottom of the drawing summarizes individual data point types by rows, and on the right we can see their totals (per diagram / overall), which then serve as a basis for sizing I/O modules or PLCs.
Process diagram with data points
Note that one peripheral device may contain several data points. For example, each frequency converter includes one DO (run enable), one AO (frequency control), and one DI (fault indication). Analog inputs are labeled as UI (“Universal Inputs”) and divided into two groups:
· UIa – 0…10 V inputs
· UIr – inputs for passive resistance sensors.
This is because some control systems have certain inputs dedicated for specific purposes (e.g., resistance measurement only), and the general label “UI” would not provide sufficient detail for control-system specification.
So far, we have considered only hardware inputs and outputs of the control system. While these hardware, or “physical,” data points are essential for sizing the control system, when creating and especially when licensing software, we must also account for additional, “virtual” data points. These are represented by variables in the control software of programmable sub-stations. Let’s look at how virtual data points can affect the total number of variables using a typical control application — a heating circuit with weather-compensated regulation:
AI: Outdoor temperature, heating-water temperature (2)
AO: Heating circuit control valve (1)
DI: Circulation pump run signal (1)
DO: Circulation pump switching (1)
Parameters of the weather-compensation curve, four-point, separate for day and night (4 × 2 × 2 = 16)
Calculated heating-water temperature based on the weather-compensation curve (1)
Time program (considered as 1, but actually a rather complex data structure)
Alarm – pump fault (1)
Circuit run enable (1)
…and a diligent programmer would surely find a few more in the code.
In this admittedly extreme example, the total number of variables is five times the number of physical data points. However, virtual data points must be considered when estimating software development effort and especially when sizing the license for the visualization program (SCADA). In practice, for visualization licensing purposes, the “number of tags” (i.e., variables in the visualization) is considered to be three to four times the number of physical data points.
Somewhere between physical and virtual data points lies another category — data points obtained through the integration of third-party or proprietary devices via communication buses. These include photovoltaic inverters, battery systems, room controllers, as well as heat pumps, diesel generators, and similar equipment. From the commissioning workload perspective, this is something of a “gray area.” In the best-case scenario, the technician connects to a functioning external device, establishes communication on the first try, and within half an hour integrates dozens or hundreds of variables. Other times, it can take several days of troubleshooting just to integrate five data points — in which case, wiring them analogically might have been easier in the end. Much depends on the technician’s previous experience and the degree of cooperation from the third-party system vendor.
Of course, when determining the license size for the visualization system, it’s necessary to calculate the actual number of integrated variables. In the offer phase, this is difficult to predict, so it’s safer to include a margin. Typical values are as follows:
Device
Number of Integrated Data Points
PV inverter
15–20
Frequency converter
8–15
Diesel generator
10–30
Battery management system
40–60
| Device | Number of Integrated Data Points |
| PV inverter | 15–20 |
| Frequency converter | 8–15 |
| Diesel generator | 10–30 |
| Battery management system | 40–60 |
The advantage of integration is that some data simply cannot be effectively transmitted in analog form — for example, cumulative values such as meter readings, operating hours, or even fault codes and multi-state mode indications.
In some cases, communication-based integration eliminates the need for physical inputs and outputs in the control system, which means savings on hardware (though not necessarily on application software or graphical-interface licensing). This factor must also be taken into account during system design.
For room controllers, things become even more interesting. A typical controller has at least one output (for a thermal valve), while the temperature sensor and the setpoint adjustment knob are built into the device and are not connected separately, simplifying commissioning. More complex systems that combine heating, cooling, and possibly even lighting and blind control, however, can include dozens of data points per room — plus complicated setup or even application programming in the main PLC. For visualization license calculations, we consider approximately 10–12 data points per room for HVAC control (without blinds or lighting):
Setpoints for comfort – setback – off modes, both for heating and cooling (6)
Actual room temperature, user offset “−3…+3 K,” calculated control temperature (3)
Heating, cooling, and fan-speed outputs (3)
Actual and commanded operating modes from the visualization (2)
Optionally, window-contact / occupancy-sensor states (2).
However, due to the repetitive nature of room configurations, commissioning work can benefit from economies of scale: the commissioning effort per zone may be reduced to 5–7 data points, or even 3 data points in exceptional cases when most or all testing is performed by the installation company. (It should be noted that “1:1” testing — i.e., thorough verification of all peripherals against the control system during commissioning — should never be underestimated. Every wiring error in zone control systems becomes a time bomb that may “explode” months or years later.)
In the Czech Republic, this is usually not the case, but for projects originating from Austria or Germany, a data point table is often included. This table is part of the MaR (measurement and regulation) implementation project, similar to the technical report, cable list, or wiring diagram. It provides a clear summary of all inputs, outputs, and their types and can be used to prepare indicative offers. The format is even standardized in EN 16484-3: Building Automation and Control Systems, Part 3 – Functions:
Data point table. Source: www.rr-schema.de
Each row represents one peripheral device, while columns show data point types and additional functions (manual operation, PI control, limit alarms, etc.). In the table, analog inputs corresponding to sensors are highlighted in red — each sensor represents one analog input.
If maintained properly, the data point table serves as an excellent tool for estimating a project’s software workload and can also help in resolving potential disputes regarding required and finally implemented system functions. Unfortunately, it is still rarely used in domestic projects.
When sizing the control system, we should not forget about possible future expansions. This includes not only the addition of I/O modules themselves but also ensuring there is enough physical space in the control cabinet and that connections to peripherals remain feasible — meaning we should plan for sufficient terminal space and adequate 24 V power supply capacity.
The client may specify in the technical report a requirement for, for example, a 10% reserve of data points — either in total (in which case all unused inputs and outputs can be counted together regardless of type) or separately for each data point type (a stricter requirement). Missing or unimplemented reserves can cause issues during project handover, so when selecting I/O modules, it’s important to verify whether and what level of reserve is required. The same applies to software data points. Additional data points can usually be purchased through license expansion, but this represents extra costs — which may be unpleasant, especially if you slightly exceed the existing limit.
Data points are the fundamental measure of a control system’s size, and therefore, attention should be paid to their structure and quantity already during the offer stage — and certainly during system design. Properly sizing the control system with respect to both physical and virtual data points saves time for installers and programmers alike and prevents unpleasant surprises during software licensing or even at the final project handover.
