The 12-bit “core” trend of oscilloscopes, how to achieve higher measurement accuracy?
INDUSTRIAL LCD DISPLAYS / IGBT MODULES DISTRIBUTOR

Infineon / Mitsubishi / Fuji / Semikron / Eupec / IXYS

The 12-bit “core” trend of oscilloscopes, how to achieve higher measurement accuracy?

Posted Date: 2024-02-02

Improving vertical resolution has long been a goal of oscilloscope designers as engineers need to measure finer signal details. However, obtaining higher vertical resolution cannot be achieved by simply increasing the number of bits in the oscilloscope's analog-to-digital converter (ADC). Tektronix 4, 5 and 6 series oscilloscopes use a new 12-bit ADC and two new low-noise amplifiers, which not only improves resolution in theory, but also greatly improves vertical resolution performance in practice. These game-changing products feature high-definition displays, fast waveform update rates, and higher vertical resolution to view signal details.

This article focuses on techniques used by designers of Tektronix 4, 5 and 6 Series MSOs to achieve higher resolution acquisition details. It also discusses the effective number of bits (ENOB) specification and the role and limitations of this important performance metric.

Let’s first look at a measured comparison of power switches.

In this example we want to observe periodic oscillations on a fairly large switching signal. Switching circuits produce oscillations after each cycle, and our goal is to examine these oscillations. But compared to the amplitude of the switching signal, the oscillation is relatively small. Figure 1 shows the results of the same test using oscilloscopes with different vertical resolutions. In order to see the entire switching cycle, the vertical scale must be set to approximately 1V/min to fit the signal into the 10 scales of the display.


Figure 1. Magnified display of a switching signal using 8-bit MDO4000C (left) and 12-bit 4 Series MSO (right) oscilloscopes

Test results: Figure 2 and Figure 3 show the test results of two oscilloscopes under the same conditions: (sampling rate 250MSa/s, number of sampling points 10k, vertical to scale 1V). Both instruments use the same IsoVu optically isolated voltage probe to eliminate noise that other probes may introduce. It can be seen that due to the small quantization level of the 8-bit oscilloscope, the results under high magnification are obviously jagged, making it difficult to analyze oscillations.


Figure 2. Shows the resonance phenomenon on an MDO4000C oscilloscope (8-bit resolution)

However, a 12-bit oscilloscope can still clearly show oscillation details at similar magnifications.


Figure 3. Shows the resonance phenomenon on the new 4 Series MSO oscilloscope (12-bit resolution)

The new 4, 5 and 6 Series MSO oscilloscopes offer 12-bit vertical resolution to see more signal detail and make more accurate measurements.

Need higher vertical resolution

When a digital oscilloscope samples a signal, the ADC divides the signal into vertical binary data (sometimes called analog-to-digital conversion levels or quantization levels or least significant bits (LSB)). Each binary data represents a discrete vertical voltage level, and the more binary data, the higher the resolution. These analog-to-digital conversion levels are represented in the ADC as 2Nwhere N represents the number of digits.

Generally, sine waves (Figure 4a) will show great differences depending on the vertical resolution. Figure 4b is a sine wave converted using a 2-bit ADC, 22 = 4 analog-to-digital conversion levels. Data can be stored in 4 different vertical binary data: 00, 01, 10 or 11. The 4-bit ADC has 16 analog-to-digital conversion levels, which are stored as 4-bit data (Figure 4c). Therefore, the more levels of analog-to-digital conversion, the higher the resolution, and the closer the signal represented by the digital oscilloscope to the original analog signal.

Higher vertical resolution provides two important advantages:

• See signals clearly and be able to zoom in to see signal details.
• Voltage can be measured more accurately, which is especially critical in power supply design verification.

Traditional digital oscilloscopes have always been based on 8-bit ADC technology, and most engineers improve horizontal resolution by increasing the sampling rate during design work. Over time, the 8-bit ADC has been optimized in terms of sampling rate, noise performance, and low distortion. But the ADC itself can only provide 28=256 vertical analog-to-digital conversion levels, which may be too coarse for applications requiring higher vertical resolution, such as power supply designs.

New oscilloscope ASIC enables higher vertical resolution


Due to the development of oscilloscope acquisition system technology, the vertical resolution achieved is greatly improved compared to the previous 8-bit ADC acquisition system. This is primarily accomplished by implementing a carefully planned ASIC design into the oscilloscope. In this article, we will explain how to significantly improve the resolution through ASIC:

• Higher performance ADC (12-bit)
• HD display processing technology
• Improved low-noise, high-gain analog front end
• Hardware filter to eliminate inherent noise
• Enables high-resolution triggering

Higher performance ADC (12-bit)

Because the new ASIC (TEK049) provides four 12-bit sequential proximity ADCs (Figure 5a), the TEK049 ADC operates at a rate of 25GS/s. Each 4, 5 or 6 series MSO can have one or two ASICs, depending on the Depends on the number of channels.

Since the TEK049 ASIC has a built-in 12-bit ADC, they provide 4,096 vertical analog-to-digital conversion levels, and the vertical resolution is 16 times higher than the previous 8-bit ADC. In the 4 and 5 Series MSOs, they provide completed 12-bit samples at 3.125GS/s. At 6.25GS/s, data is acquired via a 12-bit ADC but stored in 8-bit storage memory to accommodate the maximum transfer rate between the ASIC and memory. In the 6 Series MSO, they provide completed 12-bit samples at 12.5GS/s. At 25GS/s, data is acquired via a 12-bit ADC but stored in 8-bit storage memory to accommodate the maximum transfer rate between the ASIC and memory.

Hardware filter technology improves vertical resolution

For many years, Tektronix has provided noise reduction technology and vertical resolution enhancement capabilities to achieve 8+ bits of vertical resolution on instruments equipped with 8-bit ADCs. In this article, we focus on techniques that can be used with single acquisitions rather than waveform averaging or equivalent time sampling.

Generally speaking, oscilloscope ADCs always run at their maximum sample rate, regardless of the settings used. The user can then set a lower sample rate and compress (discard) samples to store the desired record length/sample rate combination. This mode is called "sampling mode", which discards excess samples. Tektronix has been using something called high-resolution or "HiRes" mode to make more efficient use of "excess" samples. The samples are averaged to create the desired sample rate, a process often called "signal group averaging." Each sample point consists of more information, providing better accuracy, effectively increasing vertical resolution. Figure 6 compares sampling mode to HiRes (signal group averaging) mode. This technology is still widely used.


Figure 6: Comparison of sampling mode and HiRes (signal string averaging) mode.

By using signal group averaging techniques, the number of bits of vertical resolution can be increased:

0.5 log D

Where: D is the compression ratio, or the ratio of the maximum sampling rate to the actual sampling rate.

It can be predicted that the ability to improve vertical resolution is limited by the inherent noise of the system. For example, if the ADC acquires samples after passing through an amplifier/attenuator with a high noise floor, the accuracy of those points will degrade, negating the resolution enhancement achieved by signal group averaging or traditional "HiRes" mode. It should be noted that high-resolution mode achieves the best results when analog signal conditioning and ADC sampling are combined to optimize real-time signal characteristics.

The 4, 5 and 6 Series MSOs further improve upon the signal group averaging or “HiRes” method. In traditional approaches, high-frequency noise is limited by relatively high-bandwidth anti-distortion filters. The new High Resolution mode (also called High Res) leverages hardware in the TEK049 ASIC to not only perform the averaging function, but also implements anti-distortion filters and a uniquely designed set of finite impulse response (FIR) filters for each sample rate. Assuring the user the highest resolution representation of the original signal being measured. FIR filters maintain maximum bandwidth for the selected sample rate, preventing distortion and eliminating noise energy when the available bandwidth is exceeded. Figure 8 illustrates the differences in how filters are used.


Figure 7: The filter capabilities of Series 4, 5 and 6 MSOs are significantly improved compared to the MSO/DPO5000. 5th and 17th order filters are adjustable depending on the oscilloscope settings; FFT (after trigger) on the 6 Series provides probe correction to ensure measurement system accuracy.

On 4, 5 and 6 Series MSOs, each filter's low-pass response is designed for a comprehensive balance of noise rejection and transient step response. Brickwall filters provide maximum noise suppression, but do not provide optimal transient response.

Gibbs phenomenon describes an effect where large frequency response discontinuities (such as brickwall filters) cause ringing and overshoot/undershoot in the step response of the system, as shown in Figure 9. Therefore, the equalization method must consider limiting the noise without causing a poor step response. If not carefully equalized, the oscilloscope may result in poor noise floor specifications but fail to accurately reproduce the signal in the waveform display.


Figure 8: Step response of rectangular signal in High Res mode

High Res mode in Series 4, 5 and 6 MSOs has always provided a minimum of 12-bit vertical resolution, and up to 16-bit at sample rates of 125MS/s or below.

ASIC can trigger and quickly display high-resolution samples

In addition to viewing higher resolution signals, users must be able to capture events with confidence. Therefore, the oscilloscope's triggering system must be able to handle higher resolutions to capture the display's behavior in a consistent manner. Because the TEK049 ASIC performs DSP filtering in real time, using hardware modules rather than a trigger system, triggering can be based on processed high-resolution samples. In contrast, traditional HiRes (signal group averaging) methods target stored samples rather than trigger signals, so high-frequency transients or glitches may trigger falsely and not be visible on the display screen.

New and improved High Res averaging and filtering are tightly integrated with triggering, as well as improved display modes such as FastAcq® waveform capture. In this mode, the instrument can capture more than 500,000 waveforms per second and can be used in conjunction with High Res to better view and identify signal details critical to performance, such as power supply design verification. The left side of Figure 9 shows a sine wave false trigger with noise on both edges in FastAcq mode, and the right side shows the FastAcq signal when High Res is turned on. The filtered rising edge is being triggered on the right.


Figure 9: FastAcq applies the signal group averaging function independently of the triggering system, resulting in false triggers (left). FastAcq uses the new High Res method, filtered triggering (right).

Resolution is meaningless without accuracy

There is no point in having a higher resolution if the front end of the instrument is noisy or prone to distortion, or if its sampling rate suffers from time interval errors. To quantify meaningful resolution, distortion and jitter must be considered in addition to the number of bits in the ADC. To achieve this goal, the electronics industry invented the concept of "effective number of bits" (ENOB) to account for errors due to noise, distortion, interpolation errors, and sampling jitter.

What is Effective Number of Bits (ENOB)?

ENOB represents the equivalent practical number of bits provided by an analog-to-digital converter or oscilloscope, taking into account instrument noise, harmonic distortion, linearity, and sampling jitter. It does this by taking in a very high-quality signal and then comparing the output of an analog-to-digital converter to that input. Tektronix uses the method specified by the IEEE Analog-to-Digital Waveform Recorder Standard (IEEE std. 1057). Because of the noise and distortion mentioned above, ENOB is always lower than the number of bits in the ADC. Generally speaking, the ENOB of a good 8-bit ADC oscilloscope is between 4 and 6 bits, depending on the bandwidth and vertical scale chosen. High-resolution oscilloscopes with 10- or 12-bit ADCs typically have an ENOB between 7 and 9 bits. Because ENOB takes into account more than just the theoretical ADC resolution, it is a better measure of the actual resolution of the analog-to-digital conversion system.

Although ENOB is an important factor in determining the accuracy of analog-to-digital conversion systems, it is not a be-all and end-all for comparing measurement quality. It does not include DC offset, gain, phase and frequency errors. These errors must be considered individually, for example, if the measurements made affect the accuracy of the frequency performance, then a better metric may be the error vector magnitude (EVM). ENOB may hide problems with poor frequency response or flatness on the oscilloscope.

To achieve higher ENOB, 4, 5, and 6 Series MSO oscilloscopes incorporate the enhancements highlighted earlier in this white paper:

• Higher performance ADC (12-bit)
• HD display processing technology
• Improved low-noise, high-gain analog front end
• Hardware filter to eliminate inherent noise
• Enables high-resolution triggering

To achieve higher ENOB, Figure 10 compares measurement screenshots from a 1.5V DDR3 power supply. On the left is a screenshot of a DDR3 power measurement captured with a traditional 8-bit oscilloscope with 6-bit ENOB. The power supply seems to have noticeable noise and some significant periodic voltage spikes. Shown on the right is a measurement screenshot of the same power supply, but captured using a high-resolution oscilloscope with lower noise and better ENOB of 7+ bits. Note that the reference noise is greatly reduced compared to the previous oscilloscope measurement. The consistency in amplitude of the pronounced periodic tips is also greatly improved. Using an oscilloscope with a higher ENOB helps identify problems faster and easier. In this case, 1MHz switching noise from the 1.5V buck regulator is the source of the problem.


Figure 11: DDR3 power supply comparison between 8-bit oscilloscope (left) and 12-bit oscilloscope (right).

Summarize

Higher vertical resolution in an oscilloscope allows viewing of important signal details. However, providing this resolution cannot simply rely on increasing the number of ADC bits. The 4, 5 and 6 Series MSOs use a multi-angle approach to not only achieve higher ADC resolution, but also use digital signal processing, trigger system integration, higher ENOB and a low-noise analog front end to effectively increase resolution.

4, 5 and 6 Series MSO performance compared to previous generation instruments is as follows:

Learn more about the new 4 Series B MSO at https://www.tek.com.cn/products/oscilloscopes/4-series-mso.

About Tektronix

Headquartered in Beaverton, Oregon, USA, Tektronix is ​​committed to providing innovative, precise, easy-to-operate test, measurement and monitoring solutions to solve various problems, unlock insights and promote innovation capabilities. For more than 70 years, Tektronix has been at the forefront of the digital age. Welcome to join our innovation journey, please log in: tek.com.cn


#12bit #core #trend #oscilloscopes #achieve #higher #measurement #accuracy