Steering gear PWM signal | Three key steps to say goodbye to jitter

In winter, there is a biting wind in the laboratory, blowing outside the window. The mechanical arm in front of me was originally as precise as a virgin, but now it seemed to be suffering from madness. Its head kept shaking and shaking, and it also made a tooth-piercing crunching sound. Just think about it, do you think you have seen such a scene somewhere? That servo, it, was in a state of rage. The reason for its anger is that the signals I gave were ambiguous and unclear. It's not that it doesn't want to move, it's that I haven't learned the correct way to communicate with it. This confrontation in a silent state, in the dead of night, has tortured many students who have just entered this field. If you want to explore the truth, you should ask: What is the language of the steering gear? The answer is: PWM signal, which is pulse width modulation. However, knowing its name is easy, but understanding its meaning is difficult. Only those who know its method can make it bow its head and obey orders.

The PWM signal is indeed a secret command for the steering gear. No matter what great ambition the secret command contains, it only interprets two elements: the duration of the high level, There is also the cycle that goes back and forth. In the past, in the laboratory, students could often be seen using rough pulses to drive it. The response of the servo was either like a frightened bird, twitching violently, or like an old man, slow and sluggish. Looking at the experimental diaries of predecessors, there are such frustrations everywhere in the words. The root cause is because the one-to-one correspondence between "pulse" and "position" is not understood. The period of a standard servo is always twenty milliseconds, just like a fixed rope.从绳索范围之内来看，高电平所经历的时间处于零点五毫秒至二点五毫秒这个区间内来回变动。就是这仅仅两毫秒的差异数值，然而却划分出了舵机从零度到达一百八十度的整个范围区域。 0.5 milliseconds, it is the silent command at the left extreme position; 1.5 milliseconds, it is the still waiting and watching at the midpoint; 2.5 milliseconds, it is the passionate call at the right extreme position. I once saw a case where a student was debugging a mechanical claw, and the servo always twitched and jittered near the midpoint. After repeated thinking and research, he still could not figure out the reason.Later, I used an oscilloscope to observe it, and I found out that its pulse width was erratically changing between 1.49 and 1.51 milliseconds.. This is not the fault of the servo itself, but caused by the noise interference generated by the signal source. A pure power supply and a well-shielded signal line, their importance is obvious at this moment. Let me ask, if there is no order to be cleansed, how can a loyal servant do accurate things?

The stable period is the skeleton of the signal. Twenty milliseconds, no mess at all.If the cycle is erratic, either shrinking to ten milliseconds or extending to thirty milliseconds, the decoder inside the servo will feel like it is in a fog.. It cannot recognize this kind of garbled code and can only forcibly drive the motor based on the incomplete information. As a result, it will jitter, heat, howl, and even burn out. I once dealt with an example where the servos on a robot's legs moved very strangely. Sometimes they were powerful, and sometimes they were weak and weak. It was initially determined that the gear was worn, but replacement had no effect. Later, I checked its control program and found that the developer had dynamically changed the signal period in the loop in order to pursue speed. Such a design is like inflicting torture on the steering gear. The cycle is constant, just like the rhythm of the heartbeat. Once the heartbeat is irregular, the whole body's functions will be disordered. Therefore, the first thing for anyone who wants the servo to operate smoothly is to lock the cycle limit of twenty milliseconds. It cannot be exceeded even half a step. Only in this rock-solid rhythm On the basis of playing, only by changing the high-level time can we talk about precise control. A simple example is that the PWM signal is like a lone wolf howling at the moon. Its tone (pulse width) conveys the intention, and its interval (period) follows the eternal rhythm of heaven and earth!

In addition to technology, there are also changes in the environment that can also confuse the heart of the steering gear. Especially when the seasons change and the temperature suddenly drops. Inside the servo, the resistance of the potentiometer used to feedback the position will drift with temperature. In the bitter cold of winter, the originally calibrated 1.5 millisecond midpoint signal may actually be pointing a few degrees to the left or right.When the servo senses autumn and vibrates, it is not its own will, but is actually caused by physical characteristics.. Just think about it, why would a machine that was perfectly adjusted in a greenhouse lose its accuracy once it is moved outdoors in a world of ice and snow? This is not an error in the software logic, but a natural property of the hardware material. If we face such a situation, what should we do? On the one hand, it is necessary to choose components with better low-temperature performance. However, due to cost constraints, not everyone can obtain them. On the other hand, a temperature compensation algorithm is introduced into the program. It is necessary to monitor the ambient temperature in real time and dynamically adjust the mapping relationship between pulse width and angle. Such a move may seem complicated, but it is actually done once and for all. beforekpowerServo has an internal training handout, which says at the beginning: "Only by knowing the characteristics of the rudder can you control it; only by understanding the changes in the environment can you maintain stability." Although this sentence is simple, it explains the secret of stable control. There is another method, which is to perform zero-point self-calibration when the system starts. Put the servo in the no-load state, send the theoretical left and right limits and midpoint signals one by one, read the actual feedback position, and then generate a mapping curve that specifically belongs to this servo and this moment. This method is a secret that high-level players don't easily spread.

Frequently asked questions, answers are as follows:

Why is the servo shaking for no reason? The reason is that the shaking is caused by unstable PWM signals or mismatched cycles. Then what needs to be done is to check the signal source and power supply to ensure that the cycle is locked at 20 milliseconds.。

Question: How to judge whether the pulse width is accurate? Answer: Use a logic analyzer or oscilloscope for actual measurement. Values ​​are measured in microseconds, and the standard range is from 500 microseconds to 2500 microseconds.

Q: What will happen if the power supply is insufficient?Answer: When the instantaneous current is insufficient, the machine controlling the rudder will show insufficient power, vibrate, or even stop rotating.. In this case, a separate power supply is required and filtering is performed with the help of a capacitor.

Question: Is it normal for the servo to whine? Answer: Not normal. The phenomenon of steering gear whistling is mostly caused by the signal frequency entering the range that can be sensed by the human ear or mechanical jamming. The power must be cut off immediately for inspection.

Q: How to synchronize multiple servos? Answer: Send signals one by one, according to the time slice rotation method. All steering gear command updates must be completed within the cycle.

Question: How to implement loss-of-control protection? Answer: When the signal is lost, the stop pulse width should be preset in the program. In addition, the hardware can add a pull-down resistor to return the input to a zero state.

The controlled steering gear seems to be just the ups and downs of electrical signals, but in fact it is a mutual communication and dialogue between humans and machines. Each turn was precise and precise, and behind it all was a reverent respect for the pulse, which lasted only two milliseconds. We should remember that the jitter that occurs is not a malfunction, but a misunderstood cry; the sluggishness of existence is not stupidity, but silence while waiting for the correct command. Now that you understand the truth, why not take practical action immediately? First of all, the first step is to pick up the oscilloscope at your hand and "see" the shapeless signal; then the second step is to stick to the cycle length of twenty milliseconds, just like guarding the basic bottom line of the law; then the third step is to start with a single steering gear and divide it into three stages: slow speed, medium speed, and fast speed to feel the angle changes caused by changes in pulse width. Don’t expect a complete success all at once, but hope for steady and solid progress in everything. When you successfully subdue the first arrogant and impetuous steering gear and make it docilely and softly outline a smooth and smooth arc under your control, the quiet and silent joy is the most appropriate reward for this time of exploration. Go ahead and take action. With pious and reverent heart, write down the first line of secret words that you and the steering gear formed.