发 帖  

[经验] 单个电容就足够了,不要加RC!

2018-10-11 14:12:43  163 安森美 电容
分享
0
我经常在日常工作中查看电气原理图。
有时是因为生产中出现故障,设计完整性可能会受到质疑,但大多数时候,我会在生产启动之前咨询电源转换器架构。
人们希望我报告我可以发现的任何特定缺陷,例如被忽视的重要参数。
通常,技术讨论始于权力(设计利润率,安全性等),并逐渐偏向不同的主体。
环路控制就是其中之一,尤其是TL431在隔离补偿路径中的使用。
TL431是一款有趣的三脚动物,它包含一个参考电压和一个运算放大器(运算放大器)。
当它的参考引脚超过2.5 V时,它基本上是阴极和阳极之间的集电极开路放大器吸收电流。与光耦合器相结合,它几乎出现在市场上销售的几乎所有AC-DC笔记本电脑适配器中。
尽管它很受欢迎,但许多设计人员仍然像补偿运算放大器一样配置该部件,如下所示。
图1:设计人员通常将传统的基于运算放大器的2型网络与TL431相关联
如果这种无源元件的关联本身并没有错,那么它反映了对电路操作的误解。
特别是,通过电阻器R20存在快速通道。
通过快速通道,我的意思是LED电流 - 关闭反馈环路 - 不仅取决于TL431操作,还直接取决于通过R20的Vout。
这就是困难:在高频时,当C9阻抗较低时,TL431交流电流不再依赖于R22(输出电压传感网络)带来的信息。
因此,您希望传感网络对LED施加零电流交流调制,从而很好地降低增益。
不幸的是,由于TL431的阴极电压保持在直流状态(如可编程齐纳二极管),如果Vout发生变化,LED中的交流电流会通过R20进行调制:这是快速通道效应,取消了高频时的增益滚降
正如您所期望的基于运算放大器的2型电路。
实际上,在实现图1草图时,设计人员希望计算元素值,就像围绕运算放大器构建的2型补偿器一样。
但是,由于快速通道贡献,传递函数完全不同,如下面简化的低Q定义所表示的:
(1)
如图所示,有3个极点和2个零,实际上并不是你想要的经典类型2(2极和1个零)。
在上面的表达式中,如果我们认为C9小于C10且R2小于R22那么极点和零点大致位于此处:
(2)
在该表达式中,C2是所添加的电容器C6和光耦合器寄生电容Copto的总和。
如果wz1和wp1位置很近,它们相互抵消,传递函数简化为另一种类型2表达式:
(3)
为了验证我们的结果,有趣的是在SPICE中实现电路,并将交流响应与数学求解器(例如Mathcad®)的交流响应相对应。
这是图2所示的内容,具有以下组件值:
Rpullup =20kΩ,R22 =38kΩ,R20 =1.8kΩ,C9 = 470pF,C2 = 2.3nF,CTR = 30%,C10 = 10nF,R2 =15kΩ。
使用这些值,极点和零点位于:
fz1 = 32.2 kHz,fz2 = 293 Hz,fp1 = 22.5 kHz,fp2 = 3.4 kHz
相位增强发生在fz2和fp2的几何平均值998Hz。
通过在该频率下发生的相位提升证实了这一点。
图2:SPICE仿真与基于方程的方法之间的完美匹配
所以,是的,这是一个类型2响应,但您可以在不使用C10R2网络的情况下获得完全相同的响应。
连接在TL431阴极和参考引脚之间的单个电容不仅会引入原点极点,而且由于快速通道动作也会产生零点。
更简单的原理图如图3所示。
图3:使用TL431构建2型补偿器的正确方法需要一个电容器
总而言之,不如像运算放大器那样补偿TL431,最好利用快速通道存在并理解其作用。
当你这样做时,你意识到单个电容器可以完成这项工作。
这是使用TL431和光耦合器构建2型补偿器的正确方法。
参考文献中给出了有关如何设计补偿器的更多细节。
1。
____________________________________________________________________________________________________
参考
Christophe Basso,“为线性和开关电源设计控制环 - 教程指南”,Artech House,Boston 2012,ISBN 978-1-60807-557-7

以上来自于谷歌翻译


以下为原文

I often review electrical schematics in my daily job. Sometimes it is because a failure occurred in production and design integrity may be questioned, but most of the time, I am consulted on the power converter architecture, before production is launched. People expect me to report any particular flaw I can spot such as an overlooked important parameter for instance. Usually, technical discussions start around power (design margins, safety and so on) and slowly deviate towards different subjects. Loop control is one of them and, in particular, the TL431 usage in the isolated compensation path.
The TL431 is an interesting 3-leg animal, packing a reference voltage together with an operational amplifier (op amp). It is basically an open-collector op amp sinking current between cathode and anode when its reference pin exceeds 2.5 V. Coupled with an optocoupler, it is found in almost every ac-dc notebook adapter sold on the market these days. Despite its popularity, numerous designers still configure the part like a compensated op amp as shown below.



























Figure 1: Designers often associate a traditional op amp-based type 2 network with a TL431



If this association of passive elements is not wrong per se, it reflects a misconception of the circuit operation. In particular, the presence of a fast lane through resistor R20. By fast lane, I mean the fact that the LED current – which closes the feedback loop – does not only depend on the TL431 operation but also directly on Vout via R20. And that is the difficulty here: at high frequency, when C9 impedance is low, the TL431 ac current no longer depends on information brought by R22, the output voltage sensing network. So you expect a zero-current ac modulation in the LED imposed by the sensing network, nicely rolling off the gain. Unfortunately, as the TL431 cathode voltage is maintained in dc (like a programmable Zener diode), if Vout changes, the ac current in the LED undergoes modulation through R20: this is the fast lane effect, canceling the gain roll-off at high frequency as you would expect from an op amp-based type 2 circuit.
Actually, when implementing the Figure 1 sketch, the designer expects to calculate elements values as with a type 2 compensator built around an op amp. However, because of the fast lane contribution, the transfer function totally differs as expressed by the simplified below low-Q definition:





(1)

As indicated, there are 3 poles and 2 zeros, not really the classical type 2 you want (2 poles and 1 zero). In the above expression, if we consider C9 smaller than C10 and R2 smaller than R22 then the poles and zeros are approximately located here:
(2)
In this expression, C2 is the sum of the added capacitor C6 and the optocoupler parasitic capacitance Copto. If wz1 and wp1 are closely located, they cancel each other and the transfer function simplifies to another type 2 expression:
(3)
To verify our results, it is interesting to implement the circuit in SPICE and confront the ac response with that of a mathematical solver such as Mathcad® for instance. This is what Figure 2 shows you, with the following component values:

Rpullup = 20 kΩ, R22 = 38 kΩ, R20 = 1.8 kΩ, C9 = 470 pF, C2 = 2.3 nF, CTR = 30%, C10 = 10 nF and R2 = 15 kΩ.

With these values, the poles and zeros are located at:

fz1 = 32.2 kHz, fz2 = 293 Hz, fp1 = 22.5 kHz and fp2 = 3.4 kHz

The boost in phase occurs at the geometric means of fz2 and fp2 which is 998 Hz. It is confirmed by the phase boost taking place at this frequency.

Figure 2: Perfect matching between the SPICE simulation and the equation-based approach

So yes, this is a type 2 response but you can obtain exactly the same response without the usage of the C10R2 network. A single capacitor connected between the TL431 cathode and the reference pin will not only introduce an origin pole, it will also produce a zero thanks to the fast lane action. The simpler schematic appears in Figure 3.

Figure 3: The right way to build a type 2 compensator with a TL431 requires a single capacitor

In conclusion, rather than compensating the TL431 as you would do with an op amp, it is best to capitalize on the fast lane presence and understand its role. When you do that, you realize that a single capacitor can do the job. This is the correct way to build a type 2 compensator with a TL431 and optocoupler. More details on how to design the compensator are given in Ref. 1.
____________________________________________________________________________________________________

Reference

  • Christophe Basso, “Designing Control Loops for Linear and Switching Power Supplies – A Tutorial Guide”, Artech House, Boston 2012, ISBN 978-1-60807-557-7

相关经验

只有小组成员才能发言,加入小组>>

3个成员聚集在这个小组

加入小组

创建小组步骤

关闭

站长推荐 上一条 /7 下一条

快速回复 返回顶部 返回列表