Darlington’s Contributions to Transistor Circuit Design David A. Hodges, Fellow, IEEE (Invited Paper)

Darlingtons contributions to transistor circuit design
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—I: FUNDAMENTAL THEORY AND APPLICATIONS, VOL. 46, NO. 1, JANUARY 1999

达林顿与晶体管

01 达林顿

SIDNEY Darlington’s name is well known to electronic circuit designers. He is credited with the discovery and initial demonstration in the early 1950’s of what ever since has been known as the Darlington transistor pair, or simply the Darlington transistor or Darlington pair. This episode provides another example of Sidney’s wide-ranging technical interests and his creativity addressed to the solution of real engineering problems.

Early silicon transistors suffered from low values of common-emitter current gain β , and large variations of β from sample to sample.
β might range from 5 to 15 for good samples of silicon grown-junction transistors. Given these nonideal active elements, electronic circuit designers had difﬁculty designing circuits with reasonably stable and uniform overall gain. They badly wanted transistors with larger current gain so negative feedback could be employed, at the sacriﬁce of some circuit gain, to stabilize overall circuit performance against variations in operating conditions and transistor characteristics. Sid Darlington surely understood this engineering problem at the time.

The following account of the origins of the Darlington transistor pair is due to Professor Emeritus Filson Glanz at the University of New Hampshire.

“Just after the transistor was invented at Bell Labs, Sidney checked out for the weekend two of the few existing transistors from the head of Bell Labs. Transistors were not generally available and the head of the Labs kept the few that had been made in his desk. Sidney played with them at home on the weekend and discovered/invented the Darlington pair. He realized that they could be put in one package (“on one chip”), and that in fact any number of transistors could be put in one package. The next week he was encouraged to have the lawyers draw up the patent application. He said it should be written for any number in one package, but the lawyers only wanted to do it for two—which is what was applied for. As it turned out, if it had not been restricted to two transistors, Bell Labs and Dr. Darlington would receive a royalty on every IC chip made today! Anyway, that’s the story he tells.”

▲ 图1 达林顿晶体管对。其中两个基级电阻被用于加速达林顿管关断过程

▲ 图1 达林顿晶体管对。其中两个基级电阻被用于加速达林顿管关断过程

Fig. 1. Darlington transistor pair. Resistors as shown are usually included to reduce the switching delay when turning off a conducting pair.

U.S. Patent 2 663 806 titled “Semiconductor Signal Trans-ating Device” was issued on December 22, 1953 with Sidney Darlington as sole inventor. The drawings from the patent are reproduced here as an illustration on the following page. Drawings and claims are included for both two-transistor and three-transistor compound connections. (It seems that Sidney struck a compromise with the lawyers.) This patent often was cited as related art on patents issued subsequently. Online databases available today go back only to 1971; but from 1971 to 1991, 17 subsequent patents reference this Darlington invention. Titles of those patents indicate uses ranging from power supplies to security apparatus to television receivers.

▲ 图2 达林顿晶体管专利绘图

▲ 图2 达林顿晶体管专利绘图

Fig. 1 shows a schematic diagram with transistors $T_1$ and $T_2$ connected as a Darlington transistor pair.

The resistors shown are not essential, but are usually in¬cluded to permit independent design of bias currents and to reduce the time required to turn off a conducting pair. They reduce the current gain particularly at low currents. If we neglect the current ﬂowing in the resistors and deﬁne the common-emitter current gain for a single transistor as $\beta = I_c /I_b$ , simple analysis shows that the overall dc or low frequency current gain for the Darlington pair is

$I_{out} /I_{in} = \beta _1 + \beta _2 + \beta _1 \beta _2$

Thus at low frequencies the Darlington pair is approximately equivalent to a single transistor with a current gain greater than $\beta ^2$ . Electronic circuit designers welcomed the improvement in current gain. Unfortunately but not surprisingly, high frequency analysis shows that the Darlington pair has much more phase shift than a single transistor. While a single transistor ampliﬁer stage usually is unconditionally stable when negative i-feedback is applied, this is not true of a single-stage ampliﬁer using a Darlington pair. Conventional two-stage common-emitter transistor ampliﬁers are more easily rendered stable with negative feedback than single-stage amplifers using a Darlington pair. Therefore, the applications for Darlington pairs have been largely in noncritical circuits not requiring use of feedback. Another limitation is that the minimum voltage drop through the device when conducting must be greater than the base–emitter voltage (about 0.8 V) of the second transistor. When switching large currents, the power wasted by this voltage drop can become an issue. The corresponding voltage drop for a simple common-emitter transistor switch is about 0.2 V.

The fact that the two transistors of a Darlington pair share a common collector region provokes the question of whether Sidney anticipated the development of the integrated circuit. Sidney’s patent in fact diagrams and claims a transistor pair sharing a single n-type semiconductor region forming the common collector for a Darlington pair. A jumper wire is shown making the connection between the emitter of one and the base of the other. At that time, the interconnecting jumper would have been applied by a manually controlled wire-bonding operation, one pair at a time. Modern planar integrated circuits readily provide batch-interconnected bipolar transistors sharing a common collector region. One concludes that Sidney conceived some but not all of the essential features of a modern integrated circuit.

An Internet search for “Darlington transistor” shows that even to this day there are dozens of commercial devices so identiﬁed, typically featuring a current gain of 1000 or more. They are used primarily for relay drivers and in other applications requiring simplicity and high gain. For example, they are widely used to actuate solenoid-driven ﬂippers and ﬂashing lights in electromechanical pinball machines. A logic signal of a few milliamperes from a microprocessor, ampliﬁed by a Darlington transistor, easily switches an ampere or more at 50 V on a time scale measured in milliseconds, as required for actuating a solenoid or a tungsten lamp.

Sidney Darlington made at least one other excursion into the world of transistor circuit design. According to Dr. Franklin Blecher, who joined Bell Laboratories around 1952, Sidney once designed, built, and demonstrated a complete three-stage direct-coupled transistor ampliﬁer of a sort that might be suitable for use in a hearing aid. This circuit used single transistors, not Darlington pairs. Properly designed negative feedback, stabilizing the gain and dc operating point, was included. While this was an original undertaking for Sid, it turned out that others had demonstrated such circuits earlier.

Sidney Darlington was a person of great curiosity and originality, armed with the best modern scientiﬁc and engineering knowledge. The scope of his creative accomplishments is an inspiration to all who knew him.

02 翻译版本

一、什么是达林顿对管？

任何电子工程师估计对 SIDNEY Darlington 的名字都耳熟能详。他在 20 世纪 50 年代初，突发奇想，发现并初步展示了达林顿晶体管对，或简称为达林顿晶体管或达林顿对。自此以后，达林顿晶体管对便成为人们设计电子线路中的常用器件。这个发明证明了西德尼广泛的技术兴趣，以及他为解决实际工程问题所表现出的创造力。

GM1693713315_1280_720.MPG|_-4

早期的硅晶体管共发射极电流增益 β 值很低，而且不同样品的 β 值变化很大。对于良好的硅生长结晶体管样品， β 的范围可能在 5 到 15 之间。鉴于这些有源元件的缺点，电子电路工程师在设计电路中犯了难，主要表现在电路不稳定，或者电路放大倍数偏差比较大。当时人们知道，可以通过在电路中引入负反馈来提高电路的稳定性和一致性，但这需要晶体管具有更大的电流增益。负反馈可以在工作条件和晶体管特性发生变化时稳定整体电路性能。西德-达林顿（Sid Darlington）当时对此心知肚明。

GM1693713787_1280_720.MPG|_-8

▲ 图1 达林顿晶体管对。其中两个基级电阻被用于加速达林顿管关断过程

▲ 图1 达林顿晶体管对。其中两个基级电阻被用于加速达林顿管关断过程

二、达林顿对管的发明

新罕布什尔大学的名誉教授费尔森-格兰兹，讲述了达林顿晶体管对起源的故事：

“就在贝尔实验室发明了晶体管之后，西德尼在周末从贝尔实验室负责人那里借走了两个为数不多的晶体管。当时晶体管还没有普及，实验室负责人把仅有的几个晶体管放在自己的办公桌上。西德尼周末在家摆弄这些晶体管的时候，灵光乍现，脑子里冒出了达林顿管设计方案。他意识到这些晶体管可以放在一个封装中，也就是在一个硅片上，形成单个器件。事实上，这个想法可以扩展到将任何数量的晶体管都集成在一个芯片上，这也就是未来集成芯片的雏形。第二周，他请律师将自己的想法起草专利申请。他说，专利申请内容应该包括任何数量晶体管集成在一起，但律师只想针对两个晶体管的级联。谁能想到，如果不是仅限于两个晶体管，贝尔实验室和达林顿博士就能从今天制造的每一块集成电路芯片上获得专利费！据达林顿讲述的这个故事来看，当时他的专利申请的确保守了”。

GM1693714292_1280_720.MPG|_-9

题为 "半导体信号传输装置 "的美国专利 2 663 806 于 1953 年 12 月 22 日颁发，西德尼-达林顿为唯一发明人。下面给出了专利中的配图。大家可以看到，专利申请中的确包含有两个晶体管和三个晶体管的方案，也许这反映了达林顿和律师之间的妥协结果。这一专利经常被后来颁发的专利作为相关技术引用。现在的在线数据库只能追溯到 1971 年；但从 1971 年到 1991 年，有 17 项后续专利引用了这项达林顿发明。这些专利的名称显示了从电源到安全设备再到电视接收器的各种用途。
GM1693714512_1280_720.MPG|_-5

▲ 图2 达林顿晶体管专利绘图

▲ 图2 达林顿晶体管专利绘图

三、达林顿对管的特性

1、优点

通常情况下，达林顿管内部的两个基级电阻并不是必需的，虽然通常包括在内，只是用来方便设计偏置电流，并减少关闭达林顿管所需的时间。当然，这些基级电阻的存在会降低电流增益，尤其是在低电流时。如果我们忽略电阻器中流过的电流，并将单个晶体管的共发射极电流增益定义为 β=Ic/Ib，那么简单的分析表明，达林顿晶体管对的整体直流或低频电流增益为
GM1693714771_1280_720.MPG|_-10

2、缺点

因此，在低频下，达林顿对管近似等同于单个晶体管，其电流增益大于β的平方。电子电路工程师对于电流增益的提高表示欢迎。但遗憾的是，高频分析表明达林顿对管的相移比单晶体管大得多。单晶体管放大器级在施加负反馈时通常是无条件稳定的，但使用达林顿对管的单级放大器却并非如此。与使用达林顿对管的单级放大器相比，传统的两级共发射晶体管放大器更容易通过负反馈实现稳定。因此，达林顿对管主要应用于不需要使用反馈的非关键电路。另一个限制是，导通时通过器件的最小压降必须大于第二个晶体管的基极-发射极电压（约 0.8 V）。在开关大电流时，该压降造成的功率浪费会成为一个问题。而普通的共发射晶体管开关的相应压降只有 0.2 V。

GM1693715280_1280_720.MPG|_-16

四、集成电路的先驱

达林顿晶体管对中的两个晶体管共享一个公共集电极区，这是否说明西德尼是否预见到集成电路发展的问题。事实上，西德尼的专利图示并声称一对晶体管共用一个 n 型半导体区域，该区域构成达林顿晶体管对管的公共集电极。专利图中的跳线连接了一个晶体管的发射极和另一个晶体管的基极。当时，互联跳线是通过手动操作进行连接的，一次接一对。现代平面集成电路可以方便地提供批量互连的双极晶体管，它们共享一个共同的集电极区域。由此可见，西德尼构想出了现代集成电路的部分而非全部基本特征。

GM1693715499_1280_720.MPG|_-8

五、达林顿管的应用

在互联网上搜索 "达林顿晶体管 "就会发现，时至今日，被认定为达林顿晶体管的商用器件已经多达几十种。它们通常具有 1000 或更高的电流增益，主要用于继电器驱动器和其他要求简单和高增益的应用。例如，应用于电动机械弹球机中的电磁铁驱动触发器和闪光灯。来自微处理器的几毫安逻辑信号经达林顿晶体管放大后，可以轻松地以 50 V 的电压开关一个安培或更大的电流，动作时间只有数个毫秒，这正是驱动电磁线圈或钨丝灯所需要的。

GM1693715793_1280_720.MPG|_-5

六、达林顿其他工作

西德尼-达林顿在晶体管电路设计领域至少还有过一次探索。据 1952 年左右加入贝尔实验室的富兰克林-布莱切尔博士（Dr. Franklin Blecher）称，西德尼曾经设计、制造并演示了一个完整的三级直接耦合晶体管放大器，该放大器可能适合用于助听器。该电路使用的是单晶体管，而不是达林顿晶体管对。电路中还包括适当设计的负反馈，以稳定增益和直流工作点。虽然这是西德尼的一项创举，但事实证明，早在之前就有人展示过这种电路。

西德尼-达林顿具有强烈的好奇心和独创性，并掌握了最先进的现代科学和工程知识。他的创造性成就激励着所有认识他的人。

GM1693716195_1280_720.MPG|_-6