Ad resolution and precision differences

Recently made a board, of course, considering the selection of components, due to the high accuracy of the indicators, so the selection of ad is very cautious. Many people on the concept of precision and resolution is not clear, here I do a summary, I hope you do not confuse. We are engaged in electronic development, often with "precision" and "resolution" deal with, this problem is not few words can make clear, here only to give a talk.

In simple words, "precision" is used to describe the accuracy of physical quantities, and "Resolution" is used to describe the scale division. In terms of definition, these two volumes should be irrelevant. (is not a friend feel stunned ^_^). Many of the sensor-selling JS is using this to fool people. A simple analogy: there is such a common plastic ruler (the kind of students used), its range is 10 cm, which has 100 scales, the minimum can read out the effective value of 1 mm. So we're going to say that this ruler has a resolution of 1 millimeters, or 1% of the range, and its actual accuracy is unknown (0.1 millimeters). When we bake it with a fire and lengthen it for a while and then examine it again. It's not hard to see that it has 100 ticks, its "resolution" or 1 millimeters, as it was originally. However, you will also think that its accuracy is still the original 0.1 mm. (This example is quoted on the internet, the individual think the metaphor is very image.) ）

Returning to electronic technology, we examine a commonly used digital temperature sensor: AD7416. The supplier is just touting that it has 10-bit ad with a resolution of 1/1024. Well, a lot of people will be so happy: wow, if the measurement temperature of 0-100 degrees Celsius, 100/1024 ... Approximately equal to 0.098 degrees Celsius. So high precision, enough to use. But let's go through the AD7416 data sheet and find it impressively written: measuring accuracy 0.25 degrees Celsius. So the resolution and accuracy are completely different, in this temperature sensor, as long as you want, you can even use a 14-bit ad, get 1/16384 resolution, but the accuracy of the measured value is 0.25 degrees Celsius ^_^

So a lot of friends when it comes to precision, and the resolution is immediately linked, including some project leaders, will only say: This system is very high precision requirements Ah, you ad's number of bits at least how much ah ...

In fact, a closer look at the ad's data sheet will reveal two very important indicators related to accuracy: DNL and INL. It seems that there are not many friends who know these two indicators, so it is necessary to explain them here.

Differential nonlinearity of Dnl:differencial nonliner--

Inl:interger nonliner--Integral nonlinearity (precision is mainly represented by this value)

He represents the error value of the point where the ADC device corresponds to the analog value at all numeric points and the maximum error between the true values. That is, the output value deviates from the linear maximum distance. The unit is the LSB (the amount represented by the lowest bit).

Of course, like ad in AD-∑ series, it also uses linearity error to express accuracy.

Why some ad is expensive is because the INL is very low. With a resolution of two ADCs of 12bit, one inl=±3lsb, and one ±1.5lsb, their price may be one-fold.

So here to help you put these two concepts, in the future we can confidently say that the accuracy and resolution, rather than the accuracy to understand the resolution. Hehe, hope is useful to everyone. ^_^