Xiaozhi Science Popularization丨The difference between accuracy and resolution is so big, don’t confuse them anymore!

NGI123

Xiaozhi Science Popularization丨The difference between accuracy and resolution is so big, don’t confuse them anymore! [Copy link]

Resolution and accuracy - that is, Resolution and Accuracy. These are two different parameters, but they are often used interchangeably. And the way ADC manufacturers define ADC performance in data sheets is also confusing, which may lead to wrong inferences in application development. But in fact, resolution does not represent accuracy, and vice versa.

In this article, Xiaozhi proposed and explained the difference between ADC "resolution" and "accuracy", hoping to be helpful to you.

Resolution and accuracy

First, let's take a look at the basic meanings of "resolution" and "precision".

1. Resolution

Resolution refers to the ability of ADC to distinguish and quantize the smallest signal, expressed in the number of binary bits.

For example, a 10-bit ADC can distinguish a minimum quantization level of 1/2 of the reference level (full scale). That is, the higher the resolution, the more parts the full scale level can be divided into, the more accurate the result, and the closer the digital signal is to the original input analog value after being converted back by DAC.

Therefore, for a given specific ADC device, its resolution value is fixed.

2. Accuracy

Precision refers to how close the actual digital output is to the theoretically expected digital output for a given analog input. In other words, the accuracy of the converter determines how many bits of the digital output code represent useful information about the input signal.

The accuracy value or accuracy range will be indicated in the datasheet of some ADC devices.

For a given specific ADC device, its accuracy value may vary due to the influence of the external environment (temperature, interference, etc.).

ADC dynamic range accuracy and resolution

Now that you have understood the basic meanings of "resolution" and "accuracy" , let me talk about their relationship with dynamic range and noise floor. Dynamic range is defined as the ratio of the minimum and maximum signals that can be measured by the system.

The maximum signal can be Peak-to-Peak, Zero -to-Peak, or RMS Full Scale. Any of these will give different values. For example, for a 1V sine wave: Peak-to-Peak (Full Scale) = 2V Zero-to-Peak = 1V

RMS full scale = 0.707 × peak amplitude = 0.707 × 1V = 0.707V

The minimum signal is usually the RMS noise, which is the root mean square value of the signal measured when no signal is applied. The measured RMS noise level will depend on the bandwidth used when making the measurement. Every time the bandwidth doubles, the recorded noise will increase by 1.41 or 3dB.

Therefore, it is important to note that dynamic range figures are always related to a certain bandwidth, which is usually not specified, making the recorded value meaningless. The signal-to-noise ratio (SNR) and dynamic range of a device are often defined as the same value, that is: Dynamic range = SNR = RMS full scale / RMS noise and often use dB as the unit, that is, dynamic range (dB) = SNR (dB) = 20*Log10 (RMS full scale / RMS noise)

As opposed to using RMS full scale, some manufacturers quote zero-to-peak or peak-to-peak values to make the graph look prettier. This increases the final dynamic range or SNR by 3dB or 9dB, so we need to study the specifications carefully to avoid misunderstandings.

ADC resolution is determined by the number of bits used to digitize the input signal. For a 16-bit device, the total voltage range is represented as ( 2 ¹⁶ =65536) separate digital values or output codes. Therefore, the absolute minimum level that the system can measure is represented as 1 bit, or 1/65536 of the ADC voltage range.

As mentioned earlier, for 16-bit ADC resolution, the actual accuracy may be much less than the resolution due to internal or external error sources. So, for example, a given 16-bit ADC may only provide 12 bits of accuracy. In this case, the 4LSb (least significant bit) represents the random noise generated in the ADC. ADC dynamic range and ADC accuracy usually refer to the same thing.

An ideal ADC generates a digital output code that is a function of the analog signal voltage and the voltage reference input, where output code = full-scale voltage × [VIN+ - VIN-] / [VREF+ - VREF-] = full-scale voltage × [VIN /VREF]

Each digital output code represents a fractional value of the reference voltage.

It is important to note that the ADC dynamic range should match the maximum amplitude of the signal to be converted in order to maximize the ADC conversion accuracy. Now assume that the signal to be converted varies between 0V and 2.5V, and VREF is equal to 3.3V.

A 16-bit ADC will consist of 2 ¹⁶ = 65536 steps or transitions, with the least significant bit ( LSB) = VREF/65536 = 3.3V/65536 = 50.35 uV . For an ideal ADC, all codes have the same width of 1LSB.

If the maximum signal value of the ADC is 2.5V, then this means that there are a total of 49652 conversions (2.5V/1LSB). For this case, there will be 15884 conversions that are not used (65536-49652=15884). This reflects the loss of signal accuracy or ENOB (effective number of bits) loss after the conversion (a loss of 0.4 bits). If the difference between the ADC reference (VREF) and the ADC maximum signal level increases, then the ENOB loss or accuracy loss will increase. For example, if the ADC maximum signal level is 1.2V and VREF=3.3V, then the ENOB loss will be 1.5 bits. Therefore, the ADC dynamic range must match the maximum signal amplitude to obtain the highest accuracy.

Analog-to-digital converters ( ADCs) claim to have "n" bits of resolution, which is often misunderstood as accuracy. Resolution and accuracy are two completely different concepts and should not be used interchangeably. It should be determined by the specific application whether missing codes are allowed and the required ADC accuracy

.

Resolution and accuracy should not be confused. "Accuracy" is used to describe the accuracy of physical quantities, while "resolution" is used to describe the scale division.

In fact, for ADC, these two parameters are very important and often determine the price of the chip. Obviously, we all know that in the same series, 16-bit AD is generally more expensive than 12-bit AD, but for the same 12-bit AD, what parameters do different manufacturers use to distinguish performance? Performance often determines price, so what parameters have a greater impact on price? At this time, we need to use accuracy to measure.

To quote someone else's analogy: there is a common plastic ruler with a measuring range of 10 cm and 100 scales, and the minimum effective value that can be read is 1 mm. Then we say that the resolution of this ruler is 1 mm, or 1% of the measuring range; and its actual accuracy is unknown (assuming it is 0.1 mm).

When we roast it with fire, stretch it for a while, and then examine it again, we can easily find that it still has 100 scales, and its "resolution" is still 1 mm, the same as before! However, do you still think that its accuracy is still the original 0.1 mm?

*Some of the information and pictures in this article are from the Internet. The copyright belongs to the original author. If there is any infringement, please contact me to delete it. Thank you!

se7ens

Nice, very practical analysis and sharing, the last metaphor is so appropriate

Full of useful information, it explains the concepts of accuracy, resolution, dynamic range, etc.

PS: One thing that is not very clear is the calculation principle of ENOB. The results given directly in the article are confusing.

vincentc543

The explanation is very detailed, thanks for sharing