Diagnosing and Reducing Measurement Noise​

When trying to determine the source of noise in your signal, start with two quick diagnostic steps: 

1. Create an FFT graph—This step helps you visualize the frequency content of your signal and identify any frequency peaks that may stand out.  
2. Survey your setup and environment—Nearby equipment, signal sources, or even the type of cable you’re using can introduce noise.  

By combining FFT analysis with a physical inspection of your setup, you’ll gain the most comprehensive understanding of your signal behavior and potential noise sources. Continue reading to find guidance on interpreting frequency spikes and addressing common sources of interference, along with additional insights into environmental factors.  

Electrical AC signals and the magnetic fields they generate can couple onto your signal wires through cables, instruments, or even the infrastructure of your building.  

Common-mode noise is a type of AC interference that appears equally on both conductors of a differential signal pair. It typically arises from external electromagnetic fields or differences in ground potential, and is often introduced through power lines, nearby equipment, or improper grounding. Although it affects both lines identically, it can still impact measurement accuracy if not properly rejected by the system. 

Normal-mode noise refers to unwanted AC signals that appear as a voltage difference between the two conductors of a signal pair—essentially affecting the signal itself rather than both lines. This type of noise is typically introduced by external electromagnetic interference or coupling from nearby power sources and equipment. Unlike common-mode noise, normal-mode noise directly distorts the measured signal and is visible in the differential voltage. 

Let’s break these signals down into three categories based on their frequency. 

Low Frequency (50 or 60 Hz)

An FFT of your signal showing a spike at 50 Hz or 60 Hz (or even 100 Hz or 120 Hz) is an indicator that electrical or magnetic fields from power lines are being coupled onto your signal. This coupling could be from the power lines themselves, synchronous rotating equipment, or even fluorescent lighting. 

Ground loops may also be a cause for low frequency spikes. Ground loops occur when multiple ground paths exist between the signal source and the data acquisition (DAQ) device. Ground loops are a common issue in systems where the signal source and DAQ are powered from different outlets or grounding schemes. 

Depending on the source, these cabling and setup techniques may be effective to reduce the impact of low frequency noise and ground loops:

• Shielded EPM cables improve noise immunity by separating analog and digital signals and blocking external electromagnetic interference. When grounded, the shield prevents unwanted current from flowing through the signal path, helping maintain clean and accurate measurements. 

• Twisted pair cables reduce electromagnetic interference by ensuring both conductors are equally affected by external noise. When paired with differential inputs, this common-mode noise is canceled out, improving signal integrity and minimizing magnetic field pickup.
• Use isolated DAQ hardware or differential inputs to eliminate the electrical connection between grounds. This disconnect may break the loop. Also, ensure that all equipment shares a single-point ground whenever possible, and avoid connecting grounds to multiple points in the system.

Medium to High Frequency (1 kHz to hundreds of kHz)

Medium- to high-frequency noise, typically in the range of 1 kHz to several hundred kHz, can be difficult to diagnose because it often appears as an alias. In these cases, the frequency you observe may not reflect the actual source of the noise, which makes debugging more challenging. If you notice the noise shifting when you change your sample rate, that’s a strong indication that aliasing is occurring. To better understand the true frequency content, try capturing the signal with as much bandwidth as possible.

This type of noise is commonly emitted by high-speed electronics such as switch-mode power supplies (SMPSs). Even if these devices aren’t directly connected to your system, they can still interfere with your signals by radiating energy through the air. Their presence in the vicinity of your setup can be enough to introduce unwanted noise.

To reduce the impact of this kind of interference, consider taking the following actions, keeping in mind that even unconnected devices nearby can affect your signal:

• Shield the device emitting the noise to contain its electromagnetic output.
• Use shielded cables to prevent external noise from coupling onto your signal wires.
• Inspect your system for unwanted current paths as noise may be induced on the ground line, so you may need to check for ground loops or improper grounding.
• Adjust your sample rate or applying filters can also help isolate and identify the true source of the noise.

Very High Frequency (MHz to GHz)

Very high-frequency noise, typically in the MHz to GHZ range, can be especially difficult to manage, particularly if your lab is located near a radio or TV broadcast antenna. This type of RF noise interference could appear as broad-spectrum noise, raising the overall noise floor of your measurements.

Shielding may help, but it tends to be less effective at these higher frequencies compared to lower-frequency noise sources. Instead, consider the following techniques to reduce RF interference: 

• Use twisted pair shielded cables and ground them properly to block RF pickup.
• Keep cables short and well-routed, away from power lines and noisy electronics.
• Try analog filtering to limit bandwidth to just what your signal needs.

If you’re still seeing elevated noise levels, especially compared to your device’s spec sheet, try disconnecting components one at a time or shorting the DAQ device input to the device’s ground to establish a baseline. This technique can help you isolate the source of interference without diving into more advanced techniques right away.

DC-coupled noise will appear as an offset between the expected value and the measured value. If your application performs averaging, you may be unintentionally masking an AC source of noise, so it’s important to confirm the offset by taking single-point measurements or plotting raw waveform data without any processing. Compare the offset to your device’s accuracy and resolution specifications. If the offset is within specification, it may be a limitation of the hardware rather than external interference.

Keep in mind that DC-coupled noise could also be seen in the form of common-mode or normal-mode noise. 

For DC-coupled noise, although there may be environmental causes such as cabling issues, it could be more helpful look for direct paths where an unintended signal could couple onto your measurement system:

• Is your signal source grounded or floating? Match it with the correct DAQ device input mode to reduce noise and improve accuracy.
  • Differential mode for floating sources
  • Referenced single-ended or non-referenced single-ended for grounded sources

• If your sensor is mounted to a conductive surface, such as an accelerometer screwed to a metal chassis, ensure the monitored device has a solid path to ground. Poor grounding can introduce unwanted DC offsets. Also, avoid creating ground loops by ensuring the sensor and DAQ device share a single ground reference.

• If your cable has exposed leads or connectors, such as the plug of a BNC cable, check that nothing conductive is touching it. Even incidental contact can introduce a DC bias. 

• Verify that the DAQ device or measurement is properly grounded. Floating or poorly grounded instruments can cause DC offsets due to potential differences between the device and signal source.

• Minimize thermal drift by allowing equipment to warm up before taking measurements, especially in precision applications. Temperature changes can cause small shifts in voltage levels that appear as DC offsets over time.

• Categorize the noise by frequency.—Identify whether the noise is low (50/60 Hz), medium/high instrument (kHz), very high (MHz–GHz), or DC coupled—each has distinct sources and mitigation strategies.

• Use proper cabling and grounding.—Utilize twisted-pair shielded cables and single-point grounding along with avoiding ground loops to reduce both AC and DC noise. 

• Configure DAQ device to match signal source.—Use differential inputs for floating sources and single-ended for grounded ones to prevent offsets and interference.

• Apply filtering.—Use analog/digital filters to suppress unwanted frequencies. Isolated DAQ devices help eliminate ground loops and common-mode noise. The PXIe-6381 has a 40 kHz selectable filter to remove noise from the signal you are measuring.
  
  
  
• Use isolated measurements.—Isolated DAQ devices help eliminate ground loops and common-mode noise. Several modules for NI CompactDAQ, NI CompactRIO are isolated including the NI-9239 (24-bit 10V input) , NI-9219 (Universal analog input), and NI-9212 (thermocouple).
   

• Inspect and troubleshoot systematically.—Check for exposed conductors, poor grounding, EMI sources, and thermal drift. Use FFTs and raw data to isolate and diagnose noise issues.