One of the undesirable characteristics of an oscilloscope is vertical noise found in its analog front-end as well as in its digitizing process. The measurement system noise degrades the actual signal measurement accuracy particularly when measuring low-level noise and signals.
In most cases, the vertical noise of an oscilloscope, which is a broadband measurement instrument, increases as the bandwidth rises. Thus, engineers should carefully evaluate the vertical noise characteristics before purchasing one.
Among other effects, vertical noise can induce uncertainty in the sin (x)/x waveform construction, induce errors in amplitude measurements, produce visually undesirable “fat” waveforms and induce jitter or timing errors as a function of input edge slew rates.
Although some oscilloscope vendors provide vertical noise characteristics/specifications in their data sheet, they are often incomplete and misleading.
This article aims to help oscilloscope users better understand the vertical noise characteristics of an oscilloscope.
White noise or random theoretically exhibits a Gaussian distribution or is unbounded. This implies that due to the random nature of the noise, the more data are collected from the noise characterization measurements will cause the peak-to-peak excursion to also grow higher. Hence, random phenomenon such as random jitter and vertical noise should be specified and measured as an RMS (one standard deviation) value.
Meanwhile, what is often known as the oscilloscope’s “baseline noise floor” is actually the level of noise of the oscilloscope when it is at the lowest V/div or most sensitive V/div setting. However, most oscilloscopes offered on the market today feature reduced bandwidth characteristics when operated at their most sensitive V/div settings. Since oscilloscopes are broadband instruments — where noise floor increases as the bandwidth rises — comparing the base-line noise floor characteristics of the oscilloscopes at their most sensitive V/div setting, may not yield apples-to-apples comparison as users may be comparing lower bandwidth oscilloscope against a higher bandwidth oscilloscope.
Most oscilloscope evaluators often commit the mistake of testing the baseline noise floor characteristic at the most sensitive V/div setting of the oscilloscope, and erroneously assume that this amplitude noise is true to all V/div settings.
It has to be noted that the oscilloscope’s inherent noise actually has two components. The first is a fixed level of noise primarily attributed to the oscilloscope’s amplifier and front-end attenuator. A good approximation of this component is the base-line noise floor found at the most sensitive full-bandwidth V/div setting of the oscilloscope. Although this component of noise is negligible when the oscilloscope is used on the least sensitive settings, it dominates at the most sensitive settings.
Determined by the specific V/div settings, the second component is a relative level of noise that is based on the dynamic range of the oscilloscope. Unlike the first, this component dominates when used on the least sensitive settings and is negligible at the most sensitive settings. Despite the noisy appearance of the waveform when set at high V/div settings, the amplitude of the noise can be relatively high.