Internet Explorer is not a supported browser for TI.com. For the best experience, please use a different browser.
Video Player is loading.
Current Time 0:00
Duration 7:31
Loaded: 0.00%
Stream Type LIVE
Remaining Time 7:31
 
1x
  • Chapters
  • descriptions off, selected
  • en (Main), selected

Welcome to "Resolving the Signal, Module 1: Introduction to Noise in Analog to Digital Converters." In this video, we'll introduce some basic noise concepts. We'll discuss the difference between device-level noise specifications and system-level noise performance in order to help bridge this gap when designing a high-resolution data acquisition system.

So what is noise? Noise is any unwanted signal that can interfere with or hide another signal that we're interested in measuring. Noise is inherent to all electrical components. It can come from sources internal or external to our system.

Why is it important to us? Well, because every component in the system contributes noise and affects the overall noise performance, noise must be considered at a system level when choosing components.

Next, when we're trying to capture a small signal with our ADC, noise will limit the smallest signal or the smallest change in signal level that we can detect. And therefore, it directly impacts the precision of our system.

Finally, noise may also have an impact on our accuracy. Particularly during calibration where we need precise and repeatable results to ensure an accurate calibration.

As an example, here are two images of the same subject. The noisy image in the top looks grainy and has a significant loss of detail. So much so that it is very difficult to determine the subject of the image.

In the high-resolution image below, we can easily identify that we are looking at a pile of open books. And we could even zoom in and see the finer details, such as the text on the pages if we wanted to. Likewise, noise in our data acquisition system can make it very difficult to characterize and see the finer details of the signal that we are trying to measure.

Before we move on, let's define the terms "precision" and "accuracy." These terms are often used interchangeably, although there are a slight differences in their meanings.

Precision, or resolution as it is sometimes called, is the degree of the repeatability of our measurement results. Accuracy, on the other hand, refers to how closely our measurement results reflect the exact values of the signal we are measuring.

We can illustrate the difference between precision and accuracy using the classic target analogy. In the top image, we see that the results are not repeatable. They're scattered and far from the bullseye. This is an example of low precision and low accuracy.

In the second image, the results are still missing the bull's eye. However, this time they are at least tightly grouped or repeatable. This is an example of high precision but low accuracy. If you're an archer, this isn't such a bad result because you've been consistent and your accuracy can be corrected simply through calibration. That is, by adjusting your sights.

In the third image, we have achieved the ideal result of a very tight grouping directly in the center of the target. This would represent a system that's both accurate and repeatable. The kind of data acquisition system we want to help you design.

While both these metrics are important, in this presentation we're going to focus on precision. So we've shown an example of what noise looks like in an image. And we've compared it to target practice. But what does it look like in an actual data acquisition system? Well, let's look at two examples.

In the first example, let's say we are measuring a DC signal. Our ideal result would be a single repeatable ADC code as shown here. However, it's common to see a noisy output result, which only approximates the amplitude of the original DC signal that we were measuring, causing us to lose resolution.

In our second example, let's say we're measuring an AC signal. We would like to see a continuously changing plot of a sine wave. However, noise, again, distorts the measurement result. It obfuscates the shape of the sine wave and degrades the spectral purity of our signal. So where does this noise come from?

Noise can come from any part of the system or the outside world, but common noise contributors within the system are the analog to digital converter itself, which depending on the architecture, will contribute one of two types of noise, thermal or quantization noise. We'll soon explain these different types of noise. The amplifier, or signal conditioning circuitry connected to the ADC, will typically provide broadband or 1/f noise.

And if this amplifier provides a signal gain, then this noise will also get amplified and will be seen in the ADC's output. The reference source, whether it's integrated into the ADC or provided as an external IC, will have a certain level of noise. And because the ADC is comparing the input signal to the reference voltage, this noise also appears in the output result.

Power supplies, which are known for having a certain level of output ripple, may have several paths to coupling noise into the signal. Clocks, which control when the ADC samples the input signal, will have clock jitter that may translate into noise when sampling a time-varying signal.

The printed circuit board layout while not necessarily a source of noise on its own, may couple noise from other parts of the system or the outside world into our sensitive analog circuitry. And finally, the sensor itself can contribute noise on top of the signal that we're attempting to measure.

It's very important to notice here that when we look at an ADC datasheet and its specifications, the purpose of the datasheet is to characterize the performance of the ADC by itself. As a result, the way in which the ADC noise is tested and the system that it is tested in are designed specifically to show off the capabilities of the ADC and not the limitations of the testing system. So while the datasheet provides an accurate representation of the ADC's performance, it may not reflect the performance of our system or the system we're attempting to design.

We may have specific system-level design constraints that prevent us from achieving the same performance as the test system. Therefore, it is important for us to understand how each system element contributes to the noise and to focus on improving the weakest links in our signal chain when trying to achieve higher resolution in our data acquisition system.

To that end, we are going to focus on the impact of these three components-- the ADC, the reference, and the amplifier-- to see how we can estimate the overall system-level noise performance when we combine these components.

Why are we focusing on just these three? Well, these components are not always picked with each other in mind, or with the understanding of how they may affect one another. These components are typically the biggest sources of noise assuming that we've already designed a clean power supply and we've done a good job on our layout.

And finally, we'll see that we can fully quantify the noise from each of these sources and come up with an overall estimate of the combined system-level noise performance.

To reiterate, it is important to design our entire system with high-resolution, low-noise performance in mind. If we only choose the ADC for this task, we may find that a higher resolution ADC will not give us the performance we want. It may only quantize our systems noise into smaller increments.

So our goal is to see if we can bridge this performance gap by looking at the individual device specifications of the ADC, the amplifier, the reference, and to see how they combine and contribute to the overall system-level noise performance.