Overcoming noise in data-acquisition systems
Rick Nelson, Chief Editor -- Test & Measurement World, 10/1/2005
Data-acquisition systems, whether operating near DC or upwards of 1 GHz, can suffer adverse effects from seven types of noise:
- quantization noise,
- internal analog-to-digital converter (ADC) noise,
- power-line noise,
- time skew,
- aliasing noise,
- common-mode noise, and
- radiated noise.
 |
| Sequential sampling (top) results in interchannel timing skew, which simultaneous sampling prevents. An archived Webcast provides details. |
To reduce quantization noise, for example, Chitalwala recommends employing increased data-converter resolution as well as AC coupling in the presence of offset voltages. You can minimize internal ADC noise, he says, by employing a high-quality ADC, reducing the bandwidth you use, and maintaining a stable temperature.
To compensate for power-line noise, Chitalwala recommends using an integrating ADC, a moving-average software filter, a power-line filter, or isolated inputs. To handle time skew, he recommends simultaneous, rather than sequential, sampling, despite the latter's lower cost per channel. He describes aliasing in detail and recommends hardware filters as the best approach to eliminating adverse effects.
He describes common-mode noise, which can plague thermocouple measurements of powered devices, and he recommends the use of an isolation barrier to eliminate it. Finally, Chitalwala recommends shielding techniques for dealing with radiated noise.
Chitalwala wraps up by presenting an example of fuel-cell impedance testing. The example, which measures impedance increase (which degrades performance) over time, involves isolated inputs and outputs, 16-bit resolution, AC/DC coupling, programmable gain, and hardware filtering.
An archive of the Webcast is available on the Test & Measurement World Web site at www.tmworld.com/webcasts.
| Reference |
|
