【讨论】Tapping Mode 中力最小化问题

IG 和PG可以理解为增加针的高度？？怎么增加的？？

[以后有求助英文方面的问题，我就发到你这版，免得我想得头都大了！
我有些英文方面的资料，文件好大，不知道能不能分析。可能比较费时间啊！]


怎么回事，上面的话好像不是我说的

能说的详细一点吗?

我是说在本论坛某一个帖子里（我记不清是哪个）好像有人说过，我也不太明白，所以想求告人指点

对于这两个gain值，在Instruction Manual Software中是这样描述的：

2.3.1. Proportional and Integral Gain—An Analogy
To better understand gains and how they control SPM probes, consider the analogy of a hot air balloon carrying three balloonists. Each rider controls a separate valve on the balloon’s gas burner. The valves are mounted in parallel, such that if any one valve is open, gas flows to the burners, causing the balloon to rise. Similarly, each balloonist may turn their burner off to reduce altitude. Mounted beneath the balloon’s gondola is a camera, which automatically takes a photograph of the ground below. The balloon’s objective is to obtain detailed photographs of the surface. To obtain the highest resolution images, the balloon must track the surface as closely as possible without crashing into it. This poses a dilemma to the balloonists: how to tightly control the balloon’s position relative to the ground. Because the balloon will drift slightly up and down due to the effects of wind and temperature, the balloonists must establish some minimum altitude as a safety zone.
Let us call this the “setpoint” altitude, and let us assume that it is set at an altitude of 100 meters.1 When the terrain is flat, the problem is simplified. The balloonists need only ensure a constant supply of gas is supplied to the balloon’s burners to keep the balloon aloft. As the terrain becomes hilly, the task becomes more complex. If the terrain rises, the balloonists must respond by firing the burners to lift the balloon. As the balloon clears the hill and terrain drops away, the balloonists must turn the burners
off to reduce height and continue tracking the terrain. The type and intensity of the balloonists’ responses to terrain can be modeled in terms of three types of feedback: proportional, integral and LookAhead.
2.3.2. Proportional Gain
Proportional gain means that something is done proportionally in response to something else. In the case of our first balloonist, Peter, this means producing hot air in proportion to the balloon’s altitude above the terrain: where the terrain rises sharply, Peter uses large amounts of gas to lift the balloon; where the terrain is relatively flat, Peter supplies a small, steady amount of gas to maintain the setpoint altitude above the surface. A simple feedback loop emerges in this analogy: let us say Peter uses a range finder
every 30 seconds to determine the distance between the balloon and ground. If the balloon is below its setpoint altitude, he fires the burners. If the balloon is above its setpoint altitude, he turns off the burners to lower the balloon. The higher the proportional gain, the more Peter reacts to changes in altitude. For example, at a proportional gain of 1, if the balloon is 25 meters too low, he opens his valve at 10 liters per second; if the balloon is 50 meters too low, he opens his valve at 20 liters per second. The proportional gain value serves as a multiplier such that at a proportional gain of 2, the gas flow rates are doubled from a proportional gain of 1, and so on. Although this sort of feedback gain works well for simple, linear models, it does not function as well for nonlinear models. There remains always some residual
error which causes the system to approach, but not quite reach, the target state. Assuming that the balloonists wants to get as close as possible without crashing, the response will depend upon, among other things, the balloon’s speed over the terrain. When the balloon is being carried swiftly, it is necessary to apply feedback
earlier to compensate. (That is, more gas must be used earlier.) On the other hand, if there is little or no wind, the balloon may achieve a closer tracking of the terrain. There may also be sufficient knowledge of the terrain to anticipate its rises and
falls. In order to compensate for these effects, integral and LookAhead gain feedbacks may also be employed. These are discussed next.
2.3.3. Integral Gain
Integral gain is used to correct the cumulative error between a system and its target state. In the case of the balloon, it is not enough to use only proportional gain. As we have seen, the balloon will maintain a constant error around the setpoint altitude if it relies on proportional gain alone. It is also necessary to consider whether the total error between the balloon’s actual altitude and its setpoint altitude is increasing or decreasing over some interval of time. To correct for cumulative error, our second balloonist, Irene, utilizes integral gain. Let us assume that Peter announces the balloon’s altitude every 30 seconds from his range finder. Irene uses a stopwatch and clipboard to record the amount of error at
each measuring interval, averaging the error over a preceding interval of time (e.g., 3 minutes). Irene fires the burners based upon her observations: if she notices that the running average error puts the balloon below the setpoint altitude, she fires the
balloon’s burners, if she notices that the average error puts the balloon above the setpoint, she turns the burners off. The effect of integral gain feedback is to reduce total error by addressing error over a longer period of time. This tends to smooth out the short-term, fluctuating effects of proportional gain while narrowing the error closer to the setpoint value. Unfortunately, if the integral gain is set too high, there is a tendency to overshoot the setpoint. Therefore, integral gain is highly sensitive and must be used carefully.

[quote]原文由 zhjclock 发表:Tapping Mode 中减小力（增加setpoint）可以降低针尖对样品的损坏，特别是生物样，但力太小有不能很好跟踪样品表面；－－－这是正确的

增加I gain 和 P gain应该是相当于增加了接触力

[quote]原文由 jeasy 发表:呵呵，我和zhjclock斑竹有点不同意见，我的观点是增加I gain 和 P gain不能说是增加了接触力..建议你做个实验，把setpoint值设置到刚好不能正常成像，然后增加I G和P G,如果增加I gain 和 P gain应该是相当于增加了接触力的话,图像应该就能够正常成像了。如果不能正常成像，就是不能这样说....

关于setpoint可以参考帖子“关于setpoint，怎么理解 ”！http://www.instrument.com.cn/bbs/shtml/20060524/435104/
关于IG和PG可以参考帖子“如何理解增益的概念？”http://www.instrument.com.cn/bbs/shtml/20050926/244164/