After more than 10 years of development and promotion, magnetic angle encoder chips have been widely used in consumer, industrial and automotive industries. At present, the domestic demand for magnetic encoder chips has exceeded 50 million per year, and it is still increasing every year. 20%~30% increase.

As a sensor chip, the magnetic encoder chip needs to have a matching external magnetic circuit, such as magnets, magnetic rings, etc., and in most applications, the magnetic encoder chip participates in the closed-loop control of the entire system. The performance will significantly affect the performance of the entire closed-loop system.

Noise is a basic problem that cannot be bypassed in the design of all electronic systems. The noise of the chip mainly comes from the noise of various internal active and passive devices, such as the thermal noise of resistors and MOS transistors, and the 1/f noise of resistors and MOS transistors. In addition, if the signal is subjected to some interference during transmission, some of it will also appear in the final output in the form of irregular noise. Fortunately, although there are many noise sources, thanks to the continuous development of circuits and signal processing technology, for users, the noise of magnetic encoder chips is mainly the thermal noise of resistance and MOS, and other forms of noise Basically, they were "processed" by special circuits. Therefore, when some high-end users build a set of magnetic encoder system (usually a 17-bit magnetic encoder system) using separate magnetic sensitive components and discrete operational amplifiers, AD converters, and MCUs, how to confirm each link Noise and suppressing them will be the first problem to face.

The size of the noise will directly affect the minimum angle that we can distinguish, that is, the resolution. Let’s consider a simple example, an ordinary ruler with a minimum scale of 1mm. Assuming that each 1mm scale can be clearly distinguished, we think that the resolution of this ruler is 1mm (but please note that here When talking about resolution, we don’t pay attention to whether the length represented by the scale is accurate, such as whether the scale of 13mm is really 13mm); if the processing accuracy is not enough, some adjacent 1mm scales overlap, then obviously this The effective resolution of the ruler is less than 1mm. This means that a ruler with a nominal resolution of 1mm, due to poor processing, may have an effective resolution of only 2mm, which is higher than the nominal resolution. The resolution is twice as bad.