Mind Your Bitness: Understanding Specs and Measurement Quality

Mind Your Bitness: Understanding Specs and Measurement Quality

Quick, how big is the 65-inch television hanging on your wall? It sounds like a trick question, but chances are it’s not exactly 65 inches. Some TVs are marketed as “65-inch class,” but out-of-the-box actually measure 64.5 inches or 64 inches. ?

Similar to consumer electronics, in the world of data acquisition (DAQ), measurement quality has a parallel “class” specification known as resolution. Let’s examine DAQ spec terminology and practical tips to boost performance.?

First Things First: Definitions?

To understand DAQ specifications, it’s critical to level-set on measurement quality terms.?

Accuracy: How close is the measured value to the actual value??

Precision: How close are subsequent measurement values to each other? This is not the same as accuracy. The measurement can be inaccurate, but if each consecutive measurement is similarly inaccurate, then the measurement has high precision.?

Resolution: More bits of resolution yield smaller values than the device can report (more decimal points). This specification is inherited from the analog-to-digital converter used in the device.?

Sensitivity: The change in the measured signal needed for the instrument to recognize something in the signal changed. Higher sensitivity means that smaller changes can be detected.?

Bits of resolution are the “TV class size” specification and a convenient proxy for measurement sensitivity. Engineers sometimes refer to this spec as “marketing bits” because it documents how the converter reports data and not what you should expect regarding measurement sensitivity.?

Noise is the phenomenon that impacts all specifications above and is the big difference between “marketing bits” and reality. Noise is any undesirable electrical addition to your measurement that interferes with your signal of interest. It can be constant, sporadic, or transient, but it’s always there.?

Sources of noise include:?

  • Internal components used in the measurement equipment, like power supplies?

  • Inherent noise from instrumentation design (the analog signal path)??

  • Crosstalk from adjacent signals (when the signal on Channel A interferes with the signal on Channel B)?

  • Running AC power cables near signal wires (AC power has a strong 50/60 Hz component)?

Practical Tips to Get the Most Out of Your DAQ?

Check the specs. Higher-resolution measurement devices don’t guarantee higher accuracy. Read the specifications documents and look for accuracy calculations that account for noise and time/temperature deviation from the last calibration. Vendors that focus on quality will document their methodology for absolute accuracy calculations and recommend calibration frequency. ?

Choose simultaneous sampling DAQ devices that use one analog-to-digital (ADC) per channel. This design eliminates inherent noise associated with systems that switch multiple signals through a single converter. Simultaneous sampling devices cost more than single ADC designs, but it’s a worthy investment to reduce noise.?

Warm up your measurement equipment to a steady-state temperature.?Accuracy drifts with temperature, so a stable environment will help ensure a stable measurement. Many manuals prescribe a warmup time needed before performing a “self-calibration” or making measurements. Watch out for large temperature changes during tests caused by moving between rooms or going outside after equipment setup.?

Use differential measurements for the best signal integrity and to avoid ground loops. This measurement mode removes uniform noise that appears on both signal wires in a twisted pair, as opposed to single-ended modes where the signal wire paths may be subjected to different noise levels or sources.?

Sampling at a higher rate helps reduce uniform noise. By taking more samples and averaging them, the uniform noise cancels out. ?

Software or hardware filtering can remove signals in known frequency ranges. ?

Avoid long signal wires. They act like antennas picking up environmental noise. Instead, use a longer USB cable (that still adheres to the USB spec) or Ethernet cable to cover the distance.?

Use twisted pair or shielded wires to minimize cross talk and EMI. Simply twisting (braiding) two signal wires can help.?

NI Knows Data Acquisition??

By understanding noise, accuracy, and resolution, you can choose the right DAQ device and optimize your system for the best combination of cost and quality for your needs.???

NI is the industry leader and your go-to technical resource for data acquisition hardware and software. To learn more about improving the quality of your measurements, check out the Engineer’s Guide to the Digitization of Analog Signals .??

Lincoln Varnum

Instrumentation Engineer, E.I.

1 个月

Great intro. My two-cent contribution for new engineers interested in measurement systems: Understanding these specifications is important because they will contribute to your statement on the quality of the data obtained with the system you've built. It is not enough to obtain an observation, you must be able to quantify how certain you are that the observation represents the real world. Go forth, and science!

回复

要查看或添加评论,请登录