Make a highly stable constant temperature crystal oscillator for the microcontroller clock

　　MCU clock has always been one of the required courses for MCU enthusiasts. From all kinds of screens, LCD, digital tube, dot matrix screen, VFD, OLED and even Nixie tube, edge light display, there is a wide variety of them. All those display modules with "inherent potential" are used in their own MCU designs. But many friends have encountered such a problem: the time is not accurate. Sometimes the new work will be 3-8 seconds faster in a day or even a few hours; the clock adjusted two months ago is not accurate after a few months...

　　I think a large part of the reason is caused by the crystal (some friends have found that it may also be related to the data reading and writing speed, chip quality and even the wiring method, but we will not explore it for the time being). In order to achieve an accurate display effect, many friends have bought real-time clock chips with temperature compensation or built-in crystals. The cumulative method is also used to adjust the travel speed, and the algorithm is optimized in the program for fast or slow clocks. However, it is difficult to make real-time compensation for temperature changes through algorithm optimization, which leads to differences in the travel accuracy of alarm clocks in different seasons. There are many ways to solve this problem. The following two different methods are used to explain the optimization of the temperature-frequency characteristics of crystals.

　　1. Optimize crystal cutting.

　　As an important member of the crystal oscillator family, the tuning fork crystal oscillator is widely used in electronic products due to its light, thin and compact features, and therefore plays an important role in the electronics industry. However, due to its frequency characteristics, its frequency-temperature characteristics are not very good.

　　In the operating temperature range of -10~60℃, Δf0 reaches about 50PPM (Note 1PPM is one part per million)

　　The crystal frequency of the improved crystal cut

　　It can be seen that the frequency-temperature characteristic of the sound difference crystal oscillator is a downward-opening parabola, and the frequency accuracy is the highest at 25°C. The frequency-temperature characteristic diagram of an AT-cut crystal oscillator with a cutting angle of 35°15' is a set of curves with 25 degrees Celsius as the symmetric center, which can roughly compensate for the frequency error in winter and summer.

　　2. Set the constant temperature crystal.

　　If we set a certain temperature point to keep the crystal constant, we can stabilize the frequency near that point to form a constant temperature crystal. After measuring the temperature of a constant temperature crystal in my hand, I found that the crystal temperature was maintained at 42°C.