Oscilloscopes and ENOB
When oscilloscope users choose which oscilloscope to use for critical measurements, knowing the quality of the scope’s measurement system is paramount. While banner specs like bandwidth, sample rate, and memory depth provide a basis of comparison, these specifications alone don’t adequately describe oscilloscope measurement quality. Seasoned scope users will also compare a scopes update rate, intrinsic jitter, and noise floor, all of which enable better measurements. For scopes with bandwidth in the GHz range, another quality metric involves characterizing a scope’s analog-to-digital converter (ADC) using effective number of bits (ENOB). When selecting which scope to use, how important is ENOB and how effective is ENOB at predicting a scope’s measurement accuracy?
Designing oscilloscope architectures for measurement accuracy involves both front-end and ADC technology blocks. A scope’s front end conditions a sampled signal so that the ADC can properly digitize the signal. The front end consists of attenuator, pre-amplifier, and path routing.
Engineers who design scopes spend significant effort designing front-ends that have flat frequency responses, low noise, and desired frequency roll-offs. Due to unique requirements for ADC technology, each scope vendor designs their own ADCs. Development of a new front end or ADC requires significant investment. Therefore, the resulting technology blocks will typically be used across multiple scope families and generations. Scope design teams maximize a scope’s accuracy when these technology blocks induce the least change to the measurement of sampled signals.
While users can characterize the combination of the ADC and front-end, users can’t easier characterize the technology blocks individually. There are many ways to measure an oscilloscope’s front end measurement quality. Oscilloscope vendors typically will use noise measurements and ENOB as useful characteristic for determining how well a scope’s front end and ADC are designed. It is often beneficial to consider the entire oscilloscope performance, instead of evaluating just ENOB or noise floor in isolation.
Characterizing an oscilloscopes noise floor at different vertical settings and offset provides an excellent criterion in determining a scope’s measurement quality. These measurements tell the user how effective the scope’s design team was in designing a quiet front-end and ADC converter. Oscilloscope noise adds unwanted jitter and erodes design margins. Typically the higher the bandwidth of the oscilloscope the more internal noise the oscilloscope produces as the scopes are accepting cumulative noise from higher frequencies that are rejected by the lower frequency roll-off of lower-bandwidth scope. A straightforward method of characterizing a scope’s noise is to disconnect all inputs and see measurement the RMS voltage readings while varying both vertical sensitivity and offset.
IEEE defined a method for determining the goodness of ADCs using ENOB. Today’s oscilloscopes typically will use to ADC architectures, pipelined or flash. Pipelined ADCs use two or more steps of subranging to achieve higher sample rate, for instance the 90000A Series oscilloscope has a 20GSa/s ADC, which combines 80 subranges of 256MSa/s to achieve the high sample rate. Interestingly and contrary to common wisdom, some scopes provide more accurate measurements when not running at fastest sample rate, due to additional interleaving distortion that can occur at the fastest sample rates and the addition of high frequency noise. Flash ADCs have a bank of comparators sampling the input signal in parallel, each firing for their decoded voltage range. The comparator bank feeds a logic circuit that generates a code for each voltage range* Each ADC technology has its own inherent limitations, for instance flash ADCs are more prone to linearity errors, while pipelined ADCs typically will have more interleaving error. IEEE created the ENOB standard to help users determine the goodness of various ADCs.
Scope vendors will internally characterize standalone ADCs. They also characterize overall ENOB of a scope system. The resulting system ENOB will be lower than the ENOB of a standalone ADC. As a scope’s ADC is part of an overall system, and can’t be used independently, only ENOB results from the overall system are useful.
Users will generally use less than the full 8-bits of a scope’s ADC. For example, to take advantage of the entire 8-bit vertical range, users would have to scale waveforms to consume the entire vertical range. This makes reading a signal more difficult, and the user runs the risk of driving the ADC into saturation, which causes undesired effects. For a signals that is scales to take 90% of the vertical range, the user reduces the scope’s 8-bit converter to 7.2 bits (90%*8 bits). Front-end noise, harmonic distortion, and interleaving distortion will further reduce the effectiveness of the scope’s ADC.
What is ENOB and how is it measured?
ENOB is measured as a fixed amplitude sin wave is swept in frequency. The resulting voltage measurements are captured and evaluated. Using time-domain methods, ENOB is calculated from subtracting the theoretical best fit voltage versus time from what was measured. The difference is noise. This noise can come from the front-end of the scope from attributes such as phase non-linearities and amplitude variations over frequency sweeps. Noise can also come from interleaving distortion from ADCs. Evaluating the same signal in the frequency domain, ENOB is calculated by subtracting the power associated with the primary tone from the entire broadband power. Both techniques provide the same result.
If you are making your ENOB measurements or analyzing ENOB measurements your scope vendor has previously made consider the following. ENOB results will be impacted by the spectral purity of the source being used. First, the source and accompanying filters should ensure that the sources ENOB is larger than the scope’s ENOB. Second, ENOB values will be dependent of the amplitude ratio of the source signal to the scope’s full screen amplitude. ENOB values will be different if the source was 75% of full screen or 90% of full screen. The JDEC standard uses 90% of full screen as their recommended amplitude for determine ENOB. Any comparisons of effective bits specifications or testing must take into account test signal amplitudes as well as frequency.
What does ENOB do well?
ENOB can be a good measure for determining the goodness of a scopes ADC. If a scope has a good ENOB, it will have minimal timing errors, frequency spurs (usually caused by interleaving distortion), and low broadband noise. If your application relies primarily on sin waves ENOB provides an effective criteria for scope selection.
What does ENOB exclude?
While ENOB is one measure of ADC and front-end “goodness,” there are several attributes that it omits. ENOB does not account for offset, phase irregularities, nor frequency response distortion. Figure 2 shows an input signal, and display of this signal on two different scopes. Despite the fact that both scopes have the same ENOB, one scope displays a dramatically more correct representation of the input signal as shown in Figure 2.
Fig 2: While ENOB provides one basis for scope evaluation, ENOB computations don’t include the effect of magnitude or phase flatness. Both scope 1 and scope 2 have the same ENOB, but scope 2 has offset and phase distortion errors that limit its ability to correctly display the input signal.
ENOB doesn’t take into account offset errors that the scope may inject. Two scopes with equal ENOB may show identical wave shapes offset by differences in absolute voltage. Adjusting offset and measuring noise or evaluating DC gain specifications would provide a better evaluation metric.
To make it easier to select the right scope, ideally, all scopes would have a flat phase and frequency response and identical roll off characteristics. However, this isn’t the case and phase and frequency response plots aren’t generally found in vendor datasheets. As well, ENOB doesn’t take into account frequency response flatness or phase irregularities. And, every scope model will have different frequency response and phase irregularities. For example, two scope models both rated to 6 GHz will produce different wave shapes when looking at a 2.1 GHz sine. One scope might have a slower bandwidth roll off and minimal phase correction algorithms, while the other scope may have a frequency response that peaks above 6 GHz before rolling off, and significant algorithms for phase correction. The scope with the higher ENOB isn’t necessarily the scope that will show the most accurate representation of the input signal.
How can I increase my scope’s ENOB?
The obvious answer is to purchase an oscilloscope with higher ENOB to begin with. If asked, scope vendors will share overall ENOB values for each scope model. Most high-end oscilloscopes come with user-selectable bandwidth limit filters. Turning on a filter limits the bandwidth of the oscilloscope. This limits the high frequency content, including the interleaving errors and noise, which will make a higher ENOB. Finally oscilloscopes can also use averaging or high res mode for repetitive signals to reduce broadband noise. Using these modes can be a very effective tool for greater measurement accuracy.
How important is ENOB for selecting the right scope?
It will greatly depend on what you are trying to measure, as to whether or not the ENOB will affect your measurement outcome. Certainly ENOB plots should be viewed collectively with noise floor measurements. High speed serial data has harmonics at very specific frequencies, which may pass through the measurement system mostly unaffected by a decrease in effective bits. For these, the scope’s noise floor may be a better indicator of measurement accuracy. If your signals are primarily fundamental sin waves, several defense applications come to mind, ENOB may be an excellent criteria. Ask your scope vendor for the ENOB plot of the specific scope model you are considering using. It is important that you know what the effective bits performance of the instrument you choose to measure with looks like across the full rated bandwidth of the instrument, as ENOB will vary with frequency.
About the authors:
Brig Asay is Product Manager High Performance Oscilloscopes in the Digital Test Division – Scopes, Electronic Measurements Group at Agilent Technologies. He manages product planning and strategic marketing for Agilent’s high performance oscilloscope business.
Joel Woodward is Senior Product Manager, Oscilloscopes in the Digital Test Division, Electronic Measurement Group at Agilent Technologies. During his 21 years with Agilent Technologies, his career has focused on bringing new and innovative system verification and validation solutions to research and development engineers in the electronics industry. Within Agilent’s electronic design automation and test and measurement businesses, He has held several individual contributor and management positions. Instrumental in the creation of three start-up businesses within Agilent, and holds a patent in the area of field-programmable gate array debug.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
