To choose a data acquisition system, define your requirements channel by channel before looking at any hardware: what each sensor outputs, how fast and how accurately it must be measured, what conditioning it needs, and where it is installed. Then add timing, environment, software and integration needs. The right DAQ hardware and software usually become obvious once that list exists; skipping it is the most common reason DAQ projects run over budget or deliver data no one trusts.
This checklist is part of our complete guide to data acquisition (DAQ), which explains the components and hardware types in detail. Here we focus on the decision process, with a worksheet you can fill in and three worked examples.
The seven-step DAQ selection process
- List every signal and fill in the channel worksheet.
- Match each signal type to the specifications that matter for it.
- Define timing: sample rates, synchronization, triggering and any control loops.
- Describe the environment and installation: distances, noise, power and safety.
- Define data handling, software and integration with other systems.
- Shortlist hardware platforms that meet all of the above.
- Check budget and lifecycle cost, then decide between off-the-shelf and custom.
Step 1: Fill in a channel-by-channel worksheet
One row per signal is the single most useful document in a DAQ project. It forces the questions that change the hardware choice, and it is exactly what an integrator needs to quote accurately. A filled-in example:
| Signal | Sensor and output | Range | Bandwidth / sample rate | Accuracy needed | Conditioning |
|---|---|---|---|---|---|
| Oven temperature (×8) | Type K thermocouple, ~41 µV/°C | 0–600 °C | 1 Hz / 10 S/s | ±2 °C | Gain, cold-junction compensation, filtering |
| Press force | Load cell, 2 mV/V | 0–5,000 lb | 50 Hz / 1 kS/s | 0.25% of full scale | 10 V excitation, bridge input, remote sense |
| Spindle vibration (×3) | IEPE accelerometer, 100 mV/g | ±50 g | 10 kHz / 25.6 kS/s | ±5% | IEPE current, AC coupling, simultaneous sampling |
| Line pressure | 4–20 mA transmitter | 0–300 psi | 5 Hz / 50 S/s | ±0.5% | 250 Ω shunt, loop power |
| Part present | 24 V digital sensor | On/off | 1 ms response | n/a | Isolated digital input |
Add a column for future channels: plan 20–30% spare capacity, because systems almost always grow. For help with the conditioning column, see signal conditioning by sensor type; for the sample rate column, see how to choose a sampling rate.
Step 2: Match each signal type to the specs that matter
| Signal type | Specifications to check |
|---|---|
| Thermocouples and RTDs | Input range in mV, resolution (usually 24-bit), cold-junction compensation accuracy, open-sensor detection, isolation |
| Strain gauges and load cells | Excitation voltage and stability, bridge completion, remote sense, shunt calibration, low-noise gain |
| Vibration and acoustics (IEPE) | Simultaneous sampling, sample rate per channel, built-in anti-aliasing filters, IEPE current, dynamic range |
| 4–20 mA and process signals | Current input or precision shunt, loop power, channel-to-channel isolation |
| High voltage and current | Input range, isolation rating (working and withstand), safety category |
| Fast transients and pulses | Analog bandwidth, sample rate, onboard memory, triggering and pre-trigger |
| Digital, counters and encoders | Logic levels, input speed, counter/timer resolution, isolation |
| Outputs and control | Analog output range and update rate, digital outputs, loop timing |
Resolution and accuracy are different specifications: a 24-bit converter can still be inaccurate if gain error, offset and drift are poor. Compare the accuracy tables in engineering units at your actual range, not the headline bit count.
Step 3: Define timing and synchronization
- Per-channel sample rate. Multiplexed devices share one converter, so divide the aggregate rate by the channel count.
- Synchronization. Do channels need to be sampled at the same instant (vibration, power, phase measurements)? Across one device or several?
- Triggering. Continuous logging, or capture around an event with pre-trigger data?
- Control. Does the system act on measurements? A loop faster than about 10 ms usually needs a real-time controller or FPGA rather than a desktop PC.
- Data rate. Channels × sample rate × bytes per sample. If it is in the MB/s range, read our high-speed data acquisition article before choosing a bus.
Step 4: Describe the environment and installation
- Distances. Are sensors within a few meters of one point, or spread across a machine or building? Long analog runs favor distributed DAQ nodes near the sensors.
- Electrical noise. Motor drives, welders and switching supplies call for differential inputs, shielding and often isolation.
- Grounding and safety. Sensors at different ground potentials, or touching hazardous voltages, need isolated inputs; see ground loops and galvanic isolation.
- Physical conditions. Temperature, vibration, dust, washdown and the enclosure rating required.
- Uptime. A lab test can be restarted; a 24/7 production monitor needs local buffering, watchdogs and a plan for power loss.
Step 5: Define data, software and integration
- Who uses the system: engineers exploring data, or operators who need a simple pass/fail screen?
- What happens to the data: files for analysis, a database, reports, or results sent to MES, SCADA or a PLC?
- Traceability and compliance needs: calibration records, audit trails, data retention.
- Expected lifetime and who maintains the software.
These answers drive the software platform more than any hardware spec; our comparison of data acquisition software covers ready-to-run tools, LabVIEW, Python and custom applications.
Step 6: Shortlist hardware platforms
| If your requirements look like this | Start your shortlist with |
|---|---|
| Under ~32 channels, near a PC, modest rates | USB DAQ devices |
| Sensors spread over a machine, line or facility | Ethernet or distributed DAQ nodes |
| Many synchronized channels, mixed signal types, high rates | Modular chassis systems such as PXI Express |
| Deterministic control, microsecond timing or edge processing | Real-time controllers and FPGA-based hardware |
| Long unattended recording with few channels | Standalone data loggers |
The guide’s section on DAQ hardware types compares these platforms in detail.
Step 7: Check budget and lifecycle cost
Hardware is often the smaller part of the cost. Include sensors and cabling, conditioning, software development, integration with existing systems, validation and calibration, operator training, and support over the system’s life. A cheaper device that needs months of custom software can cost more than a higher-end platform that fits out of the box.
Three worked examples
1. R&D test bench
Requirements: 16 thermocouples and 4 strain gauges, one sample per second, in a lab next to a PC, with frequently changing test setups.
What fits: a USB DAQ with thermocouple and bridge modules, and configurable logging software with simple test templates.
Why: slow signals, short cables and a flexible user mean simplicity and low cost matter more than performance.
2. Production line monitoring
Requirements: 64 mixed channels (temperature, pressure, flow, digital status) across several machines, running 24/7, with data on a central server and alarms for operators.
What fits: Ethernet DAQ nodes at each machine, a central server and database, and integration with SCADA or MES for dashboards and alarms.
Why: distance and uptime dominate; local nodes cut analog cabling and buffer data if the network drops.
3. High-performance test system
Requirements: vibration and acoustic measurements on dozens of channels sampled simultaneously, real-time analysis, and automated test sequences with pass/fail results.
What fits: a modular chassis with simultaneous-sampling analog modules, shared timing and triggering, and custom test software, often in LabVIEW, with FPGA processing where needed.
Why: synchronization, data rate and automation drive the choice; see high-speed data acquisition for the architecture options.
When a custom DAQ system makes sense
- Your mix of signals, speeds and environment doesn’t match any single off-the-shelf product.
- The system must integrate tightly with motion, vision, PLCs or plant systems.
- Operators, not engineers, will run it every day.
- Regulations require specific data handling, traceability or validation.
- The system will run for years and must be supported and extended.
See the DAQ projects we have delivered for examples of each.
What to send for a quote
If you work with an integrator, this is what lets them quote accurately and quickly:
- The channel worksheet from Step 1, even if some cells are estimates.
- Sensor models or data sheets, if sensors are already chosen.
- A short description of the test or process, and what decisions the data supports.
- Timing needs: sample rates, synchronization, triggers and control loops.
- Installation details: location, distances, environment and safety constraints.
- Software and integration needs: users, reports, databases, MES/SCADA/PLC links.
- Timeline and any budget range.
Have your requirements list ready?
Send it to Sciotex. We design, program and build custom data acquisition systems to your requirements, and support them in the US, as we have since 1997.
Frequently asked questions
Start with a channel-by-channel list of sensors, ranges, sample rates, accuracy and conditioning. Then add timing, environment, software and integration needs, shortlist platforms that meet them, and compare lifecycle cost.
It depends on the signals. Per-channel sample rate, input range, accuracy (not just resolution), the right conditioning for each sensor, synchronization and isolation usually matter most.
USB suits smaller systems near a PC, Ethernet suits sensors spread over a machine or facility, and PXI Express suits large synchronized systems with mixed signals and high data rates.
Plan for 20 to 30 percent spare capacity, or choose a modular platform that can be expanded, because DAQ systems almost always grow after installation.
Channel count and signal types, required speed and accuracy, conditioning and isolation, software development, integration with other systems, and validation. Software and integration often cost more than the hardware.
Related articles
- What Is Data Acquisition (DAQ)? The Complete Guide
- Signal Conditioning: What It Does and What Each Sensor Needs
- Sampling Rate, Nyquist and Anti-Aliasing Filters Explained
- High-Speed Data Acquisition: Sample Rates, Throughput and System Architecture
- Data Acquisition Software: Ready-to-Run, LabVIEW, Python or Custom?
- Ground Loops and Galvanic Isolation: Causes, Diagnosis and Fixes
- Thermocouple Data Acquisition: Cold-Junction Compensation, Accuracy and Wiring
