Choosing a DAQ System - Channels, Sample Rate, Sensor?
A DAQ (data acquisition) system turns analog sensor signals into digital measurements. This guide explains how to select by channel count, sample rate and sensor type, and how to amplify weak signals and route them to the ADC with low noise.
View DAQ systemsInhalt
How many channels and what sample rate does a DAQ system need?
Selection starts with the number of measurement points and the dynamics of the signal. The channel count follows the number of sensors read simultaneously, the sample rate follows the highest frequency you still want to capture correctly.
The Nyquist-Shannon sampling theorem requires a sample rate of at least twice the highest signal frequency. In practice you choose 5 to 10 times higher so that waveform and amplitude are reconstructed cleanly and the anti-aliasing filter has enough margin.
- Slow process values (temperature, pressure): 1 to 100 samples per second are enough.
- Vibration and acoustic measurement: several kHz up to more than 50 kHz per channel.
- Simultaneous sampling for phase relationship between channels, e.g. order analysis.
- Plan headroom: a few spare channels beat a later retrofit.
Which resolution and input range fit the signal?
The ADC resolution sets the smallest distinguishable voltage step. 16 bit divide a ±10 V range into about 65,000 steps, roughly 0.3 mV per step. For very small signals, 24-bit delta-sigma converters are common.
Match the input range to the signal amplitude. Too wide a range wastes resolution, too narrow a range clips the peaks. Use about 70 to 90 percent of the measurable range.
How are different sensor types connected and amplified?
The sensor type decides the required signal conditioning. Thermocouples deliver only millivolts and need cold-junction compensation, Pt100 resistance probes need a constant-current source in three- or four-wire connection, strain gauges need bridge excitation and amplification.
An instrumentation amplifier boosts small differential signals and rejects common-mode interference. Only then follow the anti-aliasing low-pass and the ADC. This order - amplify, filter, digitize - keeps the wanted signal well above the noise.
How do you digitize signals with low noise?
Most noise enters on the path from sensor to converter. Short, shielded and twisted leads, a single defined ground point and the separation of signal from power wiring are the most effective measures against interference pickup.
- Differential inputs instead of single-ended to reject common-mode interference.
- Connect the shield at one end only to avoid a ground loop (hum loop).
- Galvanically isolate the analog front end to keep out 50 Hz mains coupling.
- Anti-aliasing filter before the ADC to stop high frequencies folding back.
- Pick sample rates as a multiple of 50 Hz to average out mains hum.
Frequently asked questions
What is the minimum sample rate I need?
By the Nyquist theorem, at least twice the highest signal frequency. For a clean reconstruction of waveform and amplitude, 5 to 10 times higher is recommended in practice.
What is the difference between simultaneous and multiplexed sampling?
Multiplexing shares one ADC across several channels in sequence, causing a small time skew. Simultaneous sampling gives each channel its own ADC so all values are captured at exactly the same instant, which matters for phase relationship.
Is 16 bit enough or do I need 24 bit?
16 bit is enough for most voltage measurements in the ±10 V range. For very small signals such as from strain gauges or thermocouples, 24-bit delta-sigma converters deliver much more usable resolution.
Why do I need cold-junction compensation?
A thermocouple only measures the temperature difference between the measuring point and the terminal. Cold-junction compensation reads the terminal temperature and corrects the reading to the absolute temperature.
Looking for the right DAQ system?
We supply data acquisition by channel count, sample rate and sensor type - from thermocouple and Pt100 to strain gauge and IEPE, including signal conditioning.
Correctly sized
Channel count and sample rate matched to the task.
High resolution
16- to 24-bit converters for the smallest signals.
Low noise
Differential, shielded and galvanically isolated.
Expert advice
Measurement specialists help with the selection.


