1234645 玫、發明說明： 【發明所屬技術領域]1 本發明係有關於在特定電路中之溫度感應裝置及用於 溫度感應之方法。 5 【先前技術】 就如用於音訊喇叭與線性電力供應調節器之功率放大 器的高功率電路而言，其有如外部短路電路所致之高的晶 片上電流的故障狀況之可能性。因這些電流所致的晶片上 功率消散會產生超額溫度之結果，此可能使在矽晶片上之 10電路特徵降級，且在極端情形甚至構成火災之危害。為此 之故，此類電力電路經常被提供熱關機功能，在晶片溫度 若超過例如為150°c之預設限度時電力輸出被失能。為實施 此功能，需有一晶片上電路以在此預設的溫度門檻被超出 時加以檢測及定旗標。其在某些微處理器系統中也需有一 15 溫度檢測器，例如在該微處理器以高速被定時鐘。在此種 糸統中，方 度限度已到，該時鐘可被減速以降低被該微 處理器所抽動的供應電流，與/或一輸出信號可被提供以打 開風扇。 在早期，Zener二極體電壓會被電阻式地分割且被施用 20至共同射極雙極電晶體之基極。要接通雙極電晶體之基極 射極電壓（Vbe)以每。C約2mV降低，使得溫度隨著以固定電 壓被施用被提高（或甚至Zener具有正溫度係數tempco時之 上升電壓），一溫度會被達成，此處該雙極電晶體被接通且 其集極電流便可被用作為一輸出。 1234645 隨著供應電壓已降低，由於在低於5至7V難以可靠地被 達成之典型的Zener電壓太大，故此方法已變得不務實的。 反而是，如在US 3,959,713，US 4,691，688，US 4,574,205 與US 5,099,381專利的例子所示地使用帶隙電壓取代Zener 5 電壓已變成慣常的作法。例如US，381專利描述一種電路， 此處來自Brokaw格之帶隙電壓與Vbe乘數電壓被比較。為避 免在門彳監溫度附近的電氣與/或熱所引發的不穩定，某些局 部的正回饋亦可被施用以提供具有一些磁滯之切換點。運 用帶隙電壓源與回饋以提供磁滯之溫度檢測電路在us 10 5，149，199專利中被描述。在溫度檢測領域中之一般背景習 知技藝可在US6，181，121，US 2002/0093325，US 6，188,270， US 6,366,071，US 5,327,028，US 4,789,819與US 5,095,227 專利中被找到。1234645 Description of invention: [Technical field to which the invention belongs] 1. The present invention relates to a temperature sensing device in a specific circuit and a method for temperature sensing. 5 [Prior art] As for high-power circuits such as power amplifiers for audio speakers and linear power supply regulators, it has the possibility of a fault condition of high on-chip current caused by an external short circuit. The dissipation of power on the chip due to these currents can result in excess temperature, which can degrade the 10 circuit characteristics on the silicon chip, and in extreme cases can even pose a fire hazard. For this reason, such power circuits are often provided with a thermal shutdown function, and the power output is disabled when the chip temperature exceeds a preset limit of, for example, 150 ° C. To implement this function, an on-chip circuit is required to detect and flag when the preset temperature threshold is exceeded. It also requires a 15 temperature detector in some microprocessor systems, such as when the microprocessor is clocked at high speed. In such a system, the limit is reached, the clock can be slowed down to reduce the supply current drawn by the microprocessor, and / or an output signal can be provided to turn on the fan. In the early days, the Zener diode voltage was resistively divided and applied 20 to the base of a common emitter bipolar transistor. To turn on the base of the bipolar transistor, the emitter voltage (Vbe) must be set to a minimum. C is reduced by about 2mV, so that the temperature is increased as it is applied at a fixed voltage (or even the rising voltage when Zener has a positive temperature coefficient tempco), a temperature will be reached, where the bipolar transistor is turned on and its set The pole current can then be used as an output. 1234645 As the supply voltage has been reduced, this method has become impractical because the typical Zener voltage, which is difficult to reliably achieve below 5 to 7V, is too large. Instead, as shown in the examples of US Pat. Nos. 3,959,713, 4,691,688, 4,574,205, and 5,099,381, the use of a band gap voltage instead of the Zener 5 voltage has become a common practice. For example, US Patent 381 describes a circuit where the band gap voltage from the Brokaw grid is compared to the Vbe multiplier voltage. To avoid instability caused by electrical and / or heat near the door monitor temperature, some local positive feedback can also be applied to provide a switching point with some hysteresis. A temperature detection circuit using a bandgap voltage source and feedback to provide hysteresis is described in US 10 5,149,199. General background knowledge in the field of temperature detection can be found in US 6,181,121, US 2002/0093325, US 6,188,270, US 6,366,071, US 5,327,028, US 4,789,819 and US 5,095,227 patents.
固態電路之IEEE期刊1996年7月第31卷第7期第933至 15 937 頁，A. Bakker與 J· H· Huijsing 的“Micropower CMOSIEEE Journal of Solid-State Circuits, July 1996, Vol. 31, No. 7, Nos. 933--15, 937, "Micropower CMOS by A. Bakker and J.H. Huijsing"
Temperature Sensor with Digital Output”描述一種CMOS溫 度感應為’其中與Vbe電壓成比例之電流與一基準電流被比 較，後者與藉由添增PTAT(與絕對溫度成比例）電流至基極 射極電壓基準電流源所形成的溫度獨立無關。此二電流之 20和大約是與溫度獨立無關的，因其具有相反的溫度係數： 對PTAT電流為正；對vbe電流為負。然而，Bakker與Huijsing 的電路相當複雜(例如見第4圖），且其敏感度可被改進。 另一種溫度檢測電路在US 5,980，106專利中被描述，其 也使用帶隙基準。由US，106專利取出之第1A與1B圖顯示此 1234645 電路之原理。廣泛而言，分別具有正與負溫度係數特徵12, 22之二電流源1G，2G被施用至第u圖中之反相㈣被耦合 於-輸出電路之-檢測節點A。如由第_1B圖之檢查可 看出者，该反相器輸出將在A點之電壓通過第_之門檻 5溫度TD的反相器之切換門才監處切換。υ§，ι〇6亦教習如 在，106之第3A圖中顯示之對檢剛節點A的回饋之施用。一詳 細的⑽度檢測電路（第4圖）亦被描述，其中以熱電壓(ντ)為 基礎之電流Ith與在節點Α由帶隙基準Ibg被導出之電流（在 t路消除Μ 被導k負溫度係數）被組合（被比 10較）。然而，US’106之電路也是相當複雜又包括如腸脈丁 之浮動雙極電晶體。 其欲於能更簡單、更便宜且更容易地組配溫度感應 器。帶隙電壓經常在如電壓調節器之電路中出現，但在如 擴音機放大器之應用中並非必要的，故不依賴外顯帶隙電 15壓產生杰之配置會是較佳的。進而言之，其已被了解僅藉 由以由溫度係數可被參照之某些基準溫度用不同溫度係數 與可預測的絕對值，或至少可預測的相對值比較二數量來 構建μ度^欢/則裔基本上為可能的。同時，越來越多的電 路用CMOS而非雙極技術被製造，就算是如揚聲器功率放大 20态之傳統性雙極領域亦然（例如見Fairchild FAN 7021)。 CMOS之使用排除很多習知技藝技術之應用。"Temperature Sensor with Digital Output" describes a type of CMOS temperature sensing in which the current proportional to the Vbe voltage is compared to a reference current, which is compared to the base-emitter voltage reference by adding a PTAT (proportional to absolute temperature) current The temperature formed by the current source is independent and independent. The 20 sum of these two currents is approximately independent of temperature, because they have opposite temperature coefficients: positive for PTAT current; negative for vbe current. However, the circuit of Bakker and Huijsing Quite complicated (see, for example, Figure 4), and its sensitivity can be improved. Another temperature detection circuit is described in the US 5,980,106 patent, which also uses a bandgap reference. The 1A and Figure 1B shows the principle of this 1234645 circuit. Broadly speaking, the current sources 1G, 2G, which have positive and negative temperature coefficient characteristics of 12, 22, respectively, are applied to the inverting phase in the u figure, and are coupled to the-output circuit- Detect node A. As can be seen from the inspection in Figure _1B, the inverter output will be switched by the switching gate of the inverter when the voltage at point A passes the threshold _threshold 5 temperature TD. Υ§ , Ι 6 also teaches the application of feedback to the rigid node A as shown in Figure 3A of 106. A detailed degree detection circuit (Figure 4) is also described, which is based on the thermal voltage (ντ) The current Ith is combined with the current derived at the node A from the bandgap reference Ibg (eliminated at the t channel by the negative temperature coefficient k). However, the circuit of US'106 is also quite complicated and includes Changmai Ding's floating bipolar transistor. It is intended to be simpler, cheaper and easier to assemble temperature sensors. Band gap voltage often appears in circuits such as voltage regulators, but in applications such as amplifiers It is not necessary, so it does not depend on the apparent band gap of 15 volts to generate a better configuration. In addition, it has been known that it is different only by using certain reference temperatures that can be referred to by the temperature coefficient. It is basically possible to compare the temperature coefficient with a predictable absolute value, or at least a predictable relative value, to build μ degrees, and at the same time, more and more circuits are being implemented using CMOS rather than bipolar technology. Manufacturing, even if it ’s like speaker power Large traditional state of the bipolar field 20 and vice versa (see e.g. Fairchild FAN 7021). CMOS exclude many applications to use the conventional art technologies.
【明内J 依據本發明之一第一層面，其因而被提供一種溫度感 應器，包含：一電流鏡，具有一輸入與至少二輸出；一第 1234645 一基準電流產生器，具有一第一電流輸入與一第一電流輸 出且被組配以在回應該第一電流輸入下在該第一電流輸出 產生具有正溫度係數之一第一基準電流；一第二基準電流 產生器，具有一第二電流輸入與一第二電流輸出且被組配 5 以在回應該第二電流輸入下在該第二電流輸出產生具有負 溫度係數之一第二基準電流；且其中該等第一與第二基準 產生器之一具有各別的電流輸出被耦合至該電流鏡之輸 入；該第一基準產生器之第一電流輸入與該第二基準產生 器之第二電流輸入共用被耦合至該等電流鏡的第一個之一 10 輸入節點；以及該等第一與第二基準產生器之另一個具有 各別電流輸出被耦合至該等電流鏡的第二個以提供一電流 感應節點；以及其中該第一基準電流產生器包含一個雙極 電晶體熱電壓基準電流源，與該第二基準電流產生器包含 一溫度相依的半導體特徵基準電流源。 15 在本說明書中，該電流源一詞包括負電流源，其為其 中電流流入源極的來源（有時候替選地被稱為「換能器」）， 且電流因而可流入一電流源輸出。廣泛而言，二基準電流 源被提供，二者均與同一電流鏡相交，電流源之一被稱為 雙極電晶體基極射極電壓或實質上與之成比例（負電流係 20 數），該等電流源之另一被稱為雙極電晶體熱電壓或實質上 與之成比例（數學項為kT/q，其中k為Boltzman常數，T為 Kelvin絕對溫度，q為一電子上之電荷）。此熱電壓基準電流 源有時被稱為PTAT(與絕對温度成比例）源，雖然在實務上 若該輸出被外插回到絕對值0時可能有偏差。 1234645 此配置提供特別簡單且優雅的溫度感應電路，其績效 參數可相當直接地決定，且其在實務上相當一致地被做 成。在一較佳實施例中，該熱電壓基準電流源包含一對雙 極電晶體’這些電晶體之一亦提供一基極射極電壓’該第 5 二電流源可參用此而提供進一步的簡化及更緊密地一起鎖 定二電流源之爹數。 該溫度感應電路適用於以MOS，特別是CMOS技術來 組配，且在此情形中該電路為使得在該等電流源中被運用 之雙極電晶體可包含在CMOS技術中為固有的寄生（垂直或 10 側向）裝置，典型上為在P基體CMOS中之垂直PNP電晶體與 在N基體CMOS中之垂直NPN電晶體。該電路亦可以 BiCMOS被組配。 在其他實施例中，該第一（正溫度係數）源可運用MOS 而非雙極電晶體，例如使用AVgs而非AVbe式之配置，及該 15 第二（負溫度係數）源便可包含MOS VT為基準或低電流Vgs 為基準之源。 在較佳實施例中，該溫度感應器包括一正回饋且此可 藉由注入電流至共用的輸入節點而有利地被施用。此正回 饋易於在該電流感應節點輸出有切換式行為之結果，使得 20 當該輸出開始改變該正回饋時鼓勵此回饋。此正回饋在一 門檻切換温度附近亦提供磁滯。在一實施例中，該回饋可 被提供之形式為差別放大器，或其中一對電晶體之一具有 來自該電流感應節點之輸入且另一具有輸入被連接至一適 當偏壓的一對長尾電晶體。較佳地，該感應器亦包括一輸 1234645 出電路以在考慮磁滯下依該電路（更明確地為該等雙極電 晶體）之溫度是高於或低於該門檻而提供基本上為一個二 元之輸出。 在本發明之相關層面中提供一種溫度相依信號之方 5 法，該方法使用：一電流鏡，具有一輸入與至少二輸出； 一第一基準電流產生器，具有一第一電流輸入與一第一電 流輸出；一第二基準電流產生器，具有一第二電流輸入與 一第二電流輸出；且其中該等第一與第二基準產生器之一 具有各別的電流輸出被耦合至該電流鏡之輸入；該第一基 10 準產生器之第一電流輸入與該第二基準產生器之第二電流 輸入共用被耦合至該等電流鏡的第一個之一輸入節點；以 及該等第一與第二基準產生器之另一個具有各別電流輸出 被耦合至該等電流鏡的第二個而提供一電流感應節點；該 方法包含在回應於來自該共用第一節點之正溫度係數下使 15 用該第一電流產生器在該第一電流輸出產生一第一電晶體 熱電壓基準電流；在回應下來自該共用輸入節點之負溫度 係數下使用該第二電流產生器在該第二電流輸出產生一第 二電晶體電壓基準電流；以及在該感應節點組合與該等第 一及第二基準電流相依之信號以提供該溫度相依信號。 20 其將被了解，該等信號之組合包含彼此比較該等信號 或信號之彼此相減。該溫度相依輸出信號(在該感應節點處) 可包含一電流或一電壓信號。 在本發明之另一層面提供一種溫度檢測電路，包含： 一電流鏡具有一輸入與第一及第二照鏡後之電流輸出，該 10 1234645 輸入與該第一照鏡後輸出經由各別的第一與第二MOS電晶 體波道被耦合至各別的第一與第二電晶體以設定在該等第 一與第二電晶體中之電流密度的比值以提供來自該第二照 鏡後之電流輸出的一正溫度係數；一第三MOS電晶體具有 5 一閘極連接被耦合至該第一MOS電晶體之一閘極連接與一 對波道連接，該等波道連接之一經由一電阻器被耦合至該 等第一與第二電晶體之一共同連接以在該另一波道連接提 供一負溫度係數電流輸出，此處該電流輸出被作為該第一 電晶體之第一相依電壓的基準，該另一波道連接被耦合至 10 該照鏡後之電流輸出以提供一溫度相依的輸出。 在本發明之相關層面提供一種溫度檢測電路，包含： 一電流鏡具有一輸入與第一及第二照鏡後之電流輸出，該 第二與該第一鏡後輸出經由各別的第一與第二M0S電晶體 波道被耦合至各別的第一與第二電晶體；一第三M0S電晶 15 體具有一閘極連接被耦合至該第一M0S電晶體之一閘極連 接與一對波道連接，該等波道連接之一經由一電阻器被耦 合至該等第一與第二電晶體之一共同連接以在該另一波道 連接提供一負溫度係數電流輸出，此處該電流輸出被作為 該第一電晶體之第一相依電壓的基準，該另一波道連接被 20 耦合至該電流鏡輸入以提供來自該第二照鏡後之電流輸出 的負溫度係數電流；且其中該等第一與第二電晶體中之電 流密度的比值決定一正溫度係數電流，其與來自該第二照 鏡後之電流輸出的電流被組合以提供一溫度相依的輸出。 在一實施例中，該正溫度係數電流為在該第一M0S電 11 1234645 晶體信號中流動之電流。 在後者描述之特殊實施例中，該等第一與第二電晶體 為雙極電晶體，該第一MOS電晶體具有其排極與閘極被連 接在一起，及該第二MOS電晶體具有一電阻器在其源極與 5 該第二雙極電晶體間被連接。在CMOS技術中會是寄生性 的每一個雙極電晶體具有其基極及集極被連接在一起。一 回饋電路較佳地被運用，使得該溫度相依輸出在一門檻溫 度的任一側以某些磁滯大略地展現雙穩定之行為。設施亦 可被包括以例如用有效地調整該電阻器（用以將該第一個 10 雙極電晶體基極射極電壓轉換為電流）與/或藉由有效地注 入電流至該溫度相依輸出或由其抽出電流而調整該門檻溫 度。 在本發明之進一步層面中亦提供一種產生一溫度相依 信號之方法，該方法包含：使用在不同電流密度操作之一 15 對電晶體產生一熱電壓基準正溫度係數；使用該對電晶體 電壓產生一電晶體電壓負溫度係數信號；以及由該等信號 之其他者減除該等正與負溫度係數信號之一以產生該溫度 相依信號，此處該溫度相依信號之溫度相依性大於該等被 減除的信號之任一者。 20 較佳的是，該等電晶體為雙極電晶體且其電晶體電壓 為基極射極電壓。使用熱電壓基準與基極射極電壓基準信 號較佳地為電流信號，而非帶隙基準，此促成同一電晶體 就產生Vbe與PTAT電流二者均被使用。進而言之，利用彼此 減除該等正與負溫度係數信號，有效的溫度係數被提高， 12 1234645 且該溫度相依信號之溫度相依性因而被加強。較佳的是， 該減除包含施用該等正與負溫度係數信號至一檢測節點。 一正回饋亦可被施用至該共用雙極電晶體，此即被用以產 生該等正與負溫度係數信號之電晶體。 5 在本發明之相關層面中亦提供一種用於產生一溫度相 依信號之電路，該電路包含：設施使用在不同電流密度操 作之一對電晶體產生一熱電壓基準正溫度係數；設施使用 該對電晶體電壓產生一電晶體電壓負溫度係數信號；以及 設施由該等信號之其他者減除該等正與負溫度係數信號之 10 —以產生該溫度相依信號，此處該溫度相依信號之溫度相 依性大於該等被減除的信號之任一者。 圖式簡單說明 現在本發明之這些與其他層面將參照附圖，以僅為例 子之方式進一步被描述，其中： 15 第1A與1B圖分別顯示以電流源為基礎之溫度檢測電 路，及在第1A圖之電路中的電流源之熱特徵； 第2A至2C圖分別顯示一自我偏壓之基準電流源、一 Vbe 基準電流源、及一熱電壓基準電流源； 第3A至3D圖分別顯示依據本發明之不需磁滯的溫度 20 檢測電路之第一與第二實施例，及依據本發明之需磁滯的 溫度檢測電路之第一與第二實施例； 第4圖顯示依據本發明之溫度檢測電路的第三實施 例；以及 第5圖顯示依據本發明之溫度檢測電路的第四實施例。 13 1234645 L實施方式】 參照第2A圖，此顯示一所謂的自我偏壓基準電流源 200，包含一電流鏡202與一電流源204。對該電流鏡之一輸 入206在該電流鏡之輸出208設定一電流，且該基準電流源 5 204在輸出210提供一輸出電流，其依對輸入212之電·流而 定。該輸出210可為源頭或吸收電流，且在圖示之例中為吸 收電流。一般而言，該電流源之輸出在一段輸出電流範圍 大約為固定的但將以小的輸入電流減少。 該基準電流源200運用所謂的改進電路迴授技術，其中 10 該電流源輸出被連接至該電流鏡輸入，反之亦然。該電路 具有穩定的操作點，此處（就1 : 1電流鏡而言）I〇ut=Iin， 即對該電流源之輸入電流等於該電流源之輸出電流。此降 低該輸出電流之供應電壓相依性。 第2B與2C圖顯示第2A圖之基本技術的應用。第2B圖顯 15 示使用例如在John Wylie 4/E 2001 出版之PR Gray，P J Hurst， S H Lewis與 R G Meyer所著的 “Analysis and Design of Analogue Integrated Circuits” 第四章第 4·4·2 節中所描述的 CMOS技術的一雙極電晶體基極射極電壓基準電流源。 第2B圖之基極射極基準電流源220用正電源Vdd與接 20 地線222，224被供應。電晶體226與228包含一電流鏡等值 於第2A圖之電流鏡202、電晶體228提供該輸入、及電晶體 226提供該等輸出。電晶體232，234與236及電阻器238包含 一電流源等值於電流源204、電晶體232與234被配置以施用 電晶體236之基極射極電壓（實際上為一個二極體接合電壓） 14 1234645 通過電阻器238，使得lQut = Vbe/R238(因為電晶體232與234承 載相同的溫度檢測電路，且若相配便具有相同的閘極源極 電壓）。電晶體230僅提供來自電流鏡的額外輸出以提供在 線路231上等於Iout之一電流輸出。 5 第2C圖顯示一熱電壓(VT)基準電流源240。第2C圖之電 路與第2B圖者類似，且相同的元件用相同的元件編號表 示。明確地說，包含有電晶體226，228與230之一電流鏡再 次被提供，但不同的熱電壓基準電流源被運用。雙極電晶 體246，248例如藉由為其提供不同的射極面積而以不同的 10電流密度作業，但其承載相同的電流，使得(利用Ebers-Moll 公式）其乂以中之差等於(]<：丁/9)111(11/】2) = \^111(】1/】2)，其中 VT = kT/q為所謂的熱電壓(k，T與q之定義如上）、in代表以e 為底之log、J1與J2分別為電晶體Qpi與Qp2之(基極）電流密 度。在室溫時(27°C)，VT二25.9mV，在 150°C 時VT= 36.5mV。 15因而在來源240中，輸出電流1_=^丁/11250 111(：[1/12)，此大約 與絕對溫度成比例。（下面為了簡單起見，吾人假設電阻器 具有0溫度係數。在實務上，積體電路中可能具有約達 2000ppm/°C之溫度係數，但假若所有的電阻器均由相同的 材料被做成，其溫度係數都將上軌道，且後續的效應將至 20 少消除為第一階）。 現在參照第3A圖，此顯示依據本發明之溫度檢測電路 300的一第一實施例。此電路依據上述的基本原理而建立。 參照第3A圖，廣泛而言，MP1，2，3 , MN1，2，QP1， 2與R1包含類似第2C圖顯示之熱電壓電壓基準電流源。更 15 1234645 詳細地說，MOS電晶體MP1與MP2形成具有輸入302與輪出 304之電流鏡202，廣泛地對應於第2A圖之電流鏡2〇2。MOS - 電曰日體MN1與MN2、雙極電晶體Qp 1與Qp]、及電阻哭r 1 包含一vT基準電流源，實際上具有在線路3〇2上之一輸出與 5在線路304上之一輸入，因而廣泛地對應於第2A圖之電流源 … 204。MOS電晶體MP3在線路306上提供來自該電流鏡之額 外的輸出。 … MOS電晶體MN2與MN3、雙極電晶體QP2、及電阻器 R3—起包含一Vbe基準電流源以PNP雙極電晶體Qp2之基極 鲁 10射極電壓為基準。線路306亦有效地承載來自此電流源之一 輸出。其將被了解，此基極射極基準電流源與第2B圖顯示 者具有不同的組配，原因在於其對用熱電壓基準電流源被 驅動的電流鏡而非其本身的電流鏡之輸出被伺服。其亦 將被了解’在第3A圖之組配中，MOS電晶體MN2與雙極電 15晶體Qp2對熱電壓基準與Vbe基準電流源二者為共同的。 在第3A圖中，該等MOS電晶體之相對尺寸用變數μ表 示，其可被看出電流鏡電晶體ΜΡ1，ΜΡ2與MP3的尺寸比為 鲁 ΜΡ1 : ΜΡ2 : ΜΡ3=1 : 4 : 4以形成4 : 1之電流鏡，使得通 · 過ΜΡ1之電流為通過ΜΡ2之電流的1/4(及通過ΜΡ23電流的 奮 20 1/4)。MOS電晶體ΜΝ1，ΜΝ2與ΜΝ3為相同的比值，即 ΜΝ1 : ΜΝ2 : ΜΝ3 = 1 : 4 : 4。雙極電晶體QP1 與(^>2二者 具有其基極與集極接頭被連接，而尺寸比QPl : QP2 = 4 : 1， 即電晶體QP2之射極面積被設計為電晶體qP1者的1/4。 接著，第3A圖之作業將被描述。 16 1234645 假設線路306(此為接頭“0UT1”)起始地由外部被連接 至電壓源，其為高到足以使MOS電晶體MN3維持於其飽和 (固定電流）區，且低到足以使MOS電晶體MP3維持於其飽和 (固定電流）區。亦假設所有其他MOS電晶體為飽和的並承 5载電流。 電晶體MP1與MP2如先前提及地包含4 : 1電流鏡，使 件通過MP2之電流為通過MP1之電流的4倍。這些電流分別 通過電晶體MN1與MN2且因而分別通過雙極電晶體(^^與 Qp2。由於通過QP2之電流為通過QP1之電流的4倍，且由 10 於射極面積為電晶體QP1的1/4，電晶體QP2的電流密度為 電晶體QP1者之16倍。如先前地，具有J1/J2比值之電流密 度的一對雙極電晶體將具有(kT/q)ln(Jl/J2)之Vbe差，在些情 形為 25.9mVxln(16)，即在 T = 27。(：大約為 72mV，或 35.6mVxln(16)，在 150°C 時大約為 101mV。 15 現在考慮MOS電晶體MN1與MN2。電晶體MN2承載的 電流為電晶體MN1之電流的4倍，且尺寸為4倍，使得]VJN1 之閘極_源極電壓Vgs將與電晶體MN2之閘極一源極電壓 實質上相同。由於電晶體MN1之閘極被連接至電晶體MN2 之閘極，電晶體MN1之源極與電晶體MN2之源極將有相同 20 的電壓，此即在雙極電晶體QP2的基極射極電壓。此電壓被 施用至電阻器R1的上層端部，而電阻器R1的下層端部為在 雙極電晶體QP1的基極射極電壓。因此，通過R1之電壓等 於Vbe之差，△Vbe=l〇lmV，及通過R1且因而在線路3〇2之 電流為101mV/Rl。此電流再用電晶體MP3以4 : 1比值被形 17 1234645 成鏡作用’付到進入線路306(即進入或通過節點ουτί)之電 流在150 C等於404mV/Rl，而具有正溫度係數。由於此電 流與熱電壓成比例，VT = kT/q，故其實際為一PTAT電流。 現在考慮Vbe基準電流源。如先前提及者，在電晶體 5 MN2之源極的電壓為雙極電晶體QP2的基極射極電壓，且 再次如先前提及者，電晶體MN3被選擇為與電晶體MN2相 同的大小。現在假設MN2與MN3具有相同閘極一源極電 壓，則在電晶體MN3之源極的電壓亦將大約等於雙極電晶 體QP2的基極射極電壓。因此，通過r3且因而通過MN3至 春 10節點OUT1之電流將大約為（QP2Vbe)/R3。進而言之，由於 Vbe具有負溫度係數，典型上為_2mV/°C或等值之-3000ppm/ °C，所以電流將通過MN3至節點OUT1。 在所圖示之電路中，R1被選擇為44kQ，以設定通過 MP3 之電流為 I(MP3) = 404mV/44k0hm = 9.20uA且通過 QP2 15 之電流為I(QP2) = 9.20uA/4 = 2.30uA。在一製造過程中此得 到 Vbe(QP2) = 462mV，及R3 因此被設定為462mV/9.20uA = 501Ω，所以在 150°C 時，I(MN3) = I(MP3)。 _ 然後若温度上升到高於150°C，通過電晶體MP1且因而 通過MP3之電流上升，及通過電晶體MN3之電流下降，得 · 2〇 到電流由節點OUT 1出來進入該外部電壓源之結果。若溫度 下降到低於150°c，通過電晶體MP1且因而通過MP3之電流 下降’及通過電晶體MN3之電流上升，得到電流由電壓源 進入OUT1節點之結果。若該電壓源與節點ουτί之連接鬆 開，此節點之電壓位準會分別上升或下降，最終分別使]^^ 18 1234645 與MN3不再飽和以平衡電流。其可被看出〇υτι節點大 應於在第1Α圖之基本配置中之節點a。 ' 電晶體尺寸之選擇可依任何特殊應用之需求而定。就 積體電路施作而言，主要的考慮包括元件所占用之晶片: 5積及使名義上相同裝置間不相配的影響最小。典型上，雔 極電晶體間及電阻器間之隨機偏差電壓將小於電路中 電晶體間之偏差電壓，且製造的擴展將受]^1^2與]^]^1間之 不相配所支配，由於此誤差基本上是被疊於通過R1之小的 靜止電壓上。 10 首先考慮MN2與MNl之比值的選擇。如上述的電路， 但其MN2與MN1間且MP2與MP1間之比值為1，則在Ri適當 調整下仍為可用的。然而，QP1與QP2間之電流密度便僅為 4而非16，故此在通過R1僅會得到電壓你T/q)ln4而非 (kT/q)lnl6)的一半，使得該電路對MN2與MN1間之不相配 15更敏感。為了恢復電流密度比，QP1可被做為QP2之16倍， 但此會占用大量的矽面積。另一方面，若MN2與MN1間且 MP2與MP1間之比值為8 : 1而非4 : ：1，此僅會以1η32/1η16 = 1.25之因子提高通過R1之電壓，但MOS電晶體已大到降 低製造容差且使該面積倍增。就所考慮之技術而言，4 : 1 20 被選擇，但其最適值將依特定製造技術的限制而定。 現在考慮MN3對MN1之比值。如上面指出者，通過R1 之電壓在150°C約為lOOmV，通過R3者約為450mV，而這些 電阻器被要求通過相同的電流。若MP3與MN3分別為與 MP1與MN1相同的大小，則R3的電阻將為R1的電阻之約4.5 19 1234645 倍。在CMOS中使用寄生垂直電晶體時為了最佳效能，Qp】 與QP2以少數微女培電流會最佳地運轉。同時报多應用具有 嚴格的電力預异，且在這類應用中，這些電阻器傾向於具 有數十個千歐姆及占用大量的面積。導入1^^^3對]^]^1之4 : 5 1的比值使R3與R1為類似值，此就總電阻器面積傾向為最 佳的。 電晶體MP2與MP3較佳地由多重單元被形成，每一個 類似於MP1之配置。其較佳地具有大的波道長度用於媒配 及高的輸出阻抗’但具有小的波道寬度對長度比值W/L以 10保持Vgs-Vt為大的而有良好的電流媒配。 電晶體MN2與MN3類似地以為MN1配置的倍數為較 佳’且較佳地Vgs-Vt為大的而有良好的電流媒配。然而，若 Vgs-Vt為大的，此將致使Vgs(MN3)之後續變異而衰減I(MN3) 之溫度係數（基本上將1/gm(MN3)之電阻置於與R3成串 15聯）’故這些電晶體一般應以夠大的W/L被設計以得到 vgs-vt<i〇〇mV(如以在臨界溫度而言）。然後丨/gm(MN3)約為 R3之10% ’且不使電路之溫度敏感度降級或不會導入因電 阻器之非相關及]\4〇8電氣特徵而致之製造敏感性。 回顧該電路操作之上面的描述，其可看出該熱電壓基 20準對該電流鏡被「伺服」，且此電流鏡亦驅動該檢測節點。 以基極射極為基礎之基準使用與熱電壓基準相同之電晶體 以提供一個第二、負溫度係數輸出，其在該檢測節點由該 以正溫度係數熱電壓為基礎之基準被減除。其將被了解， 此配置可能被交換，使得該以Vbe為基礎之基準以使用與該 20 1234645[Mingchi J. According to a first aspect of the present invention, it is therefore provided with a temperature sensor including: a current mirror with an input and at least two outputs; a 1234645 reference current generator with a first current The input and a first current output are configured to generate a first reference current with a positive temperature coefficient at the first current output in response to the first current input; a second reference current generator having a second The current input and a second current output are configured to generate a second reference current with a negative temperature coefficient at the second current output in response to the second current input; and wherein the first and second reference One of the generators has a respective current output coupled to the input of the current mirror; the first current input of the first reference generator and the second current input of the second reference generator are commonly coupled to the current mirrors One of the first 10 input nodes; and the other of the first and second reference generators has respective current outputs that are coupled to the second of the current mirrors to provide an electrical Sensing node; and wherein the first reference current generator comprises a bipolar transistor thermal voltage reference current source, the second reference current generator comprises a temperature-dependent reference current source semiconductor feature. 15 In this specification, the term current source includes a negative current source, which is the source in which current flows into the source (sometimes referred to as a "transducer"), and the current can therefore flow into a current source output . Broadly speaking, two reference current sources are provided, both of which intersect with the same current mirror. One of the current sources is called the bipolar transistor base emitter voltage or is substantially proportional to it (negative current is 20) The other of these current sources is called the bipolar transistor thermal voltage or is substantially proportional to it (the mathematical term is kT / q, where k is the Boltzman constant, T is the absolute temperature of Kelvin, and q is the electron Charge). This thermal voltage reference current source is sometimes referred to as a PTAT (proportional to absolute temperature) source, although in practice if the output is extrapolated back to an absolute value of 0, there may be deviations. 1234645 This configuration provides a particularly simple and elegant temperature sensing circuit, whose performance parameters can be determined fairly directly, and it is done fairly consistently in practice. In a preferred embodiment, the thermal voltage reference current source includes a pair of bipolar transistors. 'One of these transistors also provides a base emitter voltage.' The fifth second current source can be used to provide further information. Simplify and more closely lock the daddy of the two current sources together. The temperature sensing circuit is suitable for assembly with MOS, especially CMOS technology, and in this case the circuit is such that the bipolar transistor used in such current sources can be included in CMOS technology which is inherently parasitic ( Vertical or 10 lateral) devices, typically vertical PNP transistors in P-based CMOS and vertical NPN transistors in N-based CMOS. This circuit can also be configured in BiCMOS. In other embodiments, the first (positive temperature coefficient) source may use MOS instead of a bipolar transistor, such as an AVgs instead of AVbe configuration, and the 15th (negative temperature coefficient) source may include MOS. VT is the reference or low current Vgs is the source of the reference. In a preferred embodiment, the temperature sensor includes a positive feedback and this can be advantageously applied by injecting current to a common input node. This positive feedback is easy to have the result of switching behavior at the output of the current sensing node, so that it is encouraged when the output starts to change the positive feedback. This positive feedback also provides hysteresis near a threshold switching temperature. In an embodiment, the feedback may be provided in the form of a differential amplifier, or where one of a pair of transistors has an input from the current sensing node and the other has a pair of long-tailed inputs whose inputs are connected to a suitable bias. Crystal. Preferably, the inductor also includes an input 1234645 output circuit to provide substantially the same as the temperature of the circuit (more specifically, the bipolar transistors) is higher or lower than the threshold considering the hysteresis. A binary output. In a related aspect of the present invention, a method 5 for temperature-dependent signals is provided. The method uses: a current mirror with an input and at least two outputs; a first reference current generator with a first current input and a first A current output; a second reference current generator having a second current input and a second current output; and wherein one of the first and second reference generators has a respective current output coupled to the current Mirror input; the first current input of the first base 10 quasi-generator and the second current input of the second reference generator are shared to one of the first input nodes of the current mirrors; and the first The other of the first and second reference generators has respective current outputs coupled to the second of the current mirrors to provide a current sensing node; the method includes responding to a positive temperature coefficient from the shared first node Use 15 to use the first current generator to generate a first transistor thermal voltage reference current at the first current output; in response to the negative temperature coefficient from the common input node, use The second current generator generates a second reference current transistor voltage in the second current output; and the sense node in combination with these first and second reference signal to provide a current dependent of the temperature dependent signal. 20 It will be understood that the combination of these signals involves comparing them to each other or subtracting each other from each other. The temperature-dependent output signal (at the sensing node) may include a current or a voltage signal. In another aspect of the present invention, a temperature detection circuit is provided, including: a current mirror having an input and a current output after the first and second mirrors, the 10 1234645 input and the output after the first mirror are passed through separate The first and second MOS transistor channels are coupled to the respective first and second transistors to set a ratio of the current density in the first and second transistors to provide a signal from the second mirror. A positive temperature coefficient of current output; a third MOS transistor has 5 a gate connection coupled to a gate connection of the first MOS transistor and a pair of channel connections, and one of the channel connections passes through A resistor is coupled to one of the first and second transistors in common to provide a negative temperature coefficient current output at the other channel connection, where the current output is used as the first of the first transistor Based on a voltage-dependent reference, the other channel connection is coupled to 10 current outputs behind the lens to provide a temperature-dependent output. In a related aspect of the present invention, a temperature detection circuit is provided, including: a current mirror having an input and a current output after the first and second mirrors, the second and the first mirror output passing through respective first and A second M0S transistor channel is coupled to each of the first and second transistors; a third M0S transistor 15 has a gate connection coupled to a gate connection of a first M0S transistor and a For a channel connection, one of the channel connections is coupled to one of the first and second transistors via a resistor to jointly connect to provide a negative temperature coefficient current output at the other channel connection, here The current output is used as a reference for the first dependent voltage of the first transistor, and the other channel connection is coupled to the current mirror input to provide a negative temperature coefficient current from the current output after the second mirror; And the ratio of the current densities in the first and second transistors determines a positive temperature coefficient current, which is combined with the current output from the second lens to provide a temperature-dependent output. In one embodiment, the positive temperature coefficient current is a current flowing in the first MOS signal 11 1234645 crystal signal. In the special embodiment described by the latter, the first and second transistors are bipolar transistors, the first MOS transistor has its drain and gate electrodes connected together, and the second MOS transistor has A resistor is connected between its source and the second bipolar transistor. Each bipolar transistor, which would be parasitic in CMOS technology, has its base and collector connected together. A feedback circuit is preferably used so that the temperature-dependent output roughly exhibits bi-stable behavior with some hysteresis on either side of a threshold temperature. Facilities may also be included, for example, by effectively adjusting the resistor (to convert the first 10 bipolar transistor base-emitter voltage to current) and / or by effectively injecting current to the temperature-dependent output Or draw the current to adjust the threshold temperature. In a further aspect of the present invention, a method for generating a temperature-dependent signal is also provided. The method includes: using one of 15 pairs of transistors operating at different current densities to generate a thermal voltage reference positive temperature coefficient; A transistor voltage negative temperature coefficient signal; and subtracting one of the positive and negative temperature coefficient signals from the other of the signals to generate the temperature dependent signal, where the temperature dependency of the temperature dependent signal is greater than the temperature dependent Any of the subtracted signals. 20 Preferably, the transistors are bipolar transistors and the transistor voltage is a base emitter voltage. The use of a thermal voltage reference and a base-emitter voltage reference signal is preferably a current signal rather than a bandgap reference. This promotes the use of the same transistor to generate both Vbe and PTAT currents. Both are used. Furthermore, by subtracting the positive and negative temperature coefficient signals from each other, the effective temperature coefficient is increased, and the temperature dependence of the temperature-dependent signal is enhanced. Preferably, the subtracting comprises applying the positive and negative temperature coefficient signals to a detection node. A positive feedback can also be applied to the common bipolar transistor, which is the transistor used to generate the positive and negative temperature coefficient signals. 5 In a related aspect of the present invention, a circuit for generating a temperature-dependent signal is also provided. The circuit includes: the facility uses one of the transistors operating at different current densities to generate a thermal voltage reference positive temperature coefficient of the transistor; the facility uses the pair The transistor voltage produces a transistor voltage negative temperature coefficient signal; and the facility subtracts 10 of the positive and negative temperature coefficient signals from the other of the signals to generate the temperature-dependent signal, where the temperature of the temperature-dependent signal is Dependency is greater than any of these subtracted signals. These and other aspects of the present invention will now be briefly described with reference to the drawings, which are further described by way of example only, where: Figures 1A and 1B respectively show a current detection circuit based on a current source, and Thermal characteristics of the current source in the circuit of Figure 1A; Figures 2A to 2C show a self-biased reference current source, a Vbe reference current source, and a thermal voltage reference current source; Figures 3A to 3D respectively show the basis The first and second embodiments of the hysteresis-free temperature 20 detection circuit of the present invention, and the first and second embodiments of the hysteresis-required temperature detection circuit according to the present invention; FIG. A third embodiment of the temperature detection circuit; and FIG. 5 shows a fourth embodiment of the temperature detection circuit according to the present invention. 13 1234645 L Embodiment] Referring to FIG. 2A, this shows a so-called self-biased reference current source 200, which includes a current mirror 202 and a current source 204. A current is set to an input 206 of the current mirror at an output 208 of the current mirror, and the reference current source 5 204 provides an output current at an output 210, which depends on the current and current to the input 212. The output 210 may be a source or a sinking current, and is a sinking current in the illustrated example. In general, the output of this current source is approximately fixed over a range of output current but will be reduced with a small input current. The reference current source 200 utilizes a so-called modified circuit feedback technique in which 10 the current source output is connected to the current mirror input and vice versa. This circuit has a stable operating point, where (for a 1: 1 current mirror) Iout = Iin, that is, the input current to the current source is equal to the output current of the current source. This reduces the supply voltage dependency of the output current. Figures 2B and 2C show the application of the basic techniques of Figure 2A. Figure 2B shows the use of, for example, "Analysis and Design of Analogue Integrated Circuits" by PR Gray, PJ Hurst, SH Lewis, and RG Meyer, published in John Wylie 4 / E 2001. Chapter 4, Section 4 · 4 · 2 A bipolar transistor-based base-emitter voltage reference current source as described in CMOS technology. The base-emitter reference current source 220 in FIG. 2B is supplied with a positive power source Vdd and ground wires 222, 224. Transistors 226 and 228 include a current mirror equivalent to current mirror 202 in Figure 2A, transistor 228 provides the input, and transistor 226 provides the outputs. Transistors 232, 234, and 236 and resistor 238 include a current source equivalent to current source 204, and transistors 232 and 234 are configured to apply the base emitter voltage of transistor 236 (actually a diode junction voltage ) 14 1234645 Through the resistor 238, make lQut = Vbe / R238 (because the transistors 232 and 234 carry the same temperature detection circuit, and if they match, they have the same gate-source voltage). Transistor 230 provides only an additional output from the current mirror to provide a current output on line 231 equal to Iout. 5 Figure 2C shows a thermal voltage (VT) reference current source 240. The circuit in Figure 2C is similar to that in Figure 2B, and the same components are indicated by the same component numbers. Specifically, a current mirror containing transistors 226, 228, and 230 is again provided, but a different thermal voltage reference current source is used. The bipolar transistors 246, 248 operate at different current densities, for example, by providing different emitter areas to them, but they carry the same current, so that (using the Ebers-Moll formula) the difference between them is equal to ( ] <: Ding / 9) 111 (11 /】 2) = \ ^ 111 (] 1 /] 2), where VT = kT / q is the so-called thermal voltage (k, T and q are as defined above), in Representing the log with e as the base, J1 and J2 are the (base) current densities of the transistors Qpi and Qp2, respectively. At room temperature (27 ° C), VT is 25.9mV, and at 150 ° C, VT = 36.5mV. 15 Therefore, in the source 240, the output current 1 _ = ^ 丁 / 11250 111 (: [1/12), which is approximately proportional to the absolute temperature. (For the sake of simplicity, I assume that the resistor has a temperature coefficient of 0. In practice, the integrated circuit may have a temperature coefficient of about 2000 ppm / ° C, but if all resistors are made of the same material , Its temperature coefficient will be on orbit, and subsequent effects will be eliminated to at least 20 as the first order). Referring now to FIG. 3A, this shows a first embodiment of a temperature detection circuit 300 according to the present invention. This circuit is built on the basic principles described above. Referring to Fig. 3A, in broad terms, MP1, 2, 3, MN1, 2, QP1, 2 and R1 include a thermal voltage voltage reference current source similar to that shown in Fig. 2C. More 15 1234645 In detail, the MOS transistors MP1 and MP2 form a current mirror 202 having an input 302 and a wheel-out 304, which broadly corresponds to the current mirror 202 of FIG. 2A. MOS-electric body MN1 and MN2, bipolar transistors Qp 1 and Qp], and resistance cry r 1 contains a vT reference current source, which actually has one output on line 302 and 5 on line 304 One input, and therefore corresponds broadly to the current source in Fig. 2A ... 204. The MOS transistor MP3 provides an additional output from the current mirror on line 306. … MOS transistors MN2 and MN3, bipolar transistor QP2, and resistor R3 together contain a Vbe reference current source based on the base of the PNP bipolar transistor Qp2 and the 10-emitter voltage reference. Line 306 also effectively carries one output from this current source. It will be understood that this base-emitter reference current source has a different combination from the one shown in Figure 2B, because the output of the current mirror driven by the thermal voltage reference current source is not its own. Servo. It will also be understood that in the assembly of FIG. 3A, the MOS transistor MN2 and the bipolar transistor 15 Qp2 are common to both the thermal voltage reference and the Vbe reference current source. In Figure 3A, the relative size of these MOS transistors is expressed by the variable μ. It can be seen that the size ratio of the current mirror transistor MP1, MP2 and MP3 is LuMP1: MP2: MP3 = 1: 4: 4. A 4: 1 current mirror is formed so that the current passing through MP1 is 1/4 of the current passing through MP2 (and 20 1/4 of the current passing through MP23). The MOS transistors MN1, MN2 and MN3 have the same ratio, that is, MN1: MN2: MN3 = 1: 4: 4. The bipolar transistors QP1 and (^ > 2 both have their base and collector joints connected, and the size ratio QPl: QP2 = 4: 1, that is, the emitter area of the transistor QP2 is designed as the transistor qP1 1/4. Next, the operation of Figure 3A will be described. 16 1234645 Assume that the line 306 (this is the connector "0UT1") is initially connected to the voltage source from the outside, which is high enough for the MOS transistor MN3 Maintained in its saturation (fixed current) region, and low enough to keep the MOS transistor MP3 in its saturated (fixed current) region. It is also assumed that all other MOS transistors are saturated and carry 5 currents. The transistors MP1 and MP2 As mentioned previously, a 4: 1 current mirror is included, so that the current through the MP2 is 4 times the current through MP1. These currents pass through the transistors MN1 and MN2, respectively, and thus through the bipolar transistors (^^ and Qp2, respectively). Since the current through QP2 is 4 times the current through QP1, and the emitter area is 10 times that of transistor QP1, the current density of transistor QP2 is 16 times that of transistor QP1. As previously, having A pair of bipolar transistors with a current density of J1 / J2 ratio will have a Vbe of (kT / q) ln (Jl / J2) The difference is 25.9mVxln (16) in some cases, ie at T = 27. (: about 72mV, or 35.6mVxln (16), about 101mV at 150 ° C. 15 Now consider the MOS transistors MN1 and MN2. Transistor MN2 carries 4 times the current of transistor MN1, and its size is 4 times, so that] the gate-source voltage Vgs of VJN1 will be substantially the same as the gate-source voltage of transistor MN2. The gate of transistor MN1 is connected to the gate of transistor MN2. The source of transistor MN1 and the source of transistor MN2 will have the same voltage. This is the base-emitter voltage of the bipolar transistor QP2. This voltage is applied to the upper end of resistor R1, and the lower end of resistor R1 is the base-emitter voltage at the bipolar transistor QP1. Therefore, the voltage across R1 is equal to the difference between Vbe, △ Vbe = lOlmV, and the current through R1 and thus the line 302 is 101mV / Rl. This current is then shaped by the transistor MP3 at a ratio of 4: 1 17 1234645 into a mirror effect 'paid into the line 306 (that is, into or The current through the node ουτί) is equal to 404mV / Rl at 150 C, and has a positive temperature coefficient. Because this current and thermoelectricity It is proportional, VT = kT / q, so it is actually a PTAT current. Now consider the Vbe reference current source. As mentioned earlier, the voltage at the source of transistor 5 MN2 is the base of bipolar transistor QP2 The emitter voltage, and again as previously mentioned, transistor MN3 is selected to be the same size as transistor MN2. Now assuming that MN2 and MN3 have the same gate-source voltage, the voltage at the source of transistor MN3 will also be approximately equal to the base-emitter voltage of bipolar transistor QP2. Therefore, the current through r3 and thus through MN3 to spring 10 node OUT1 will be approximately (QP2Vbe) / R3. Furthermore, since Vbe has a negative temperature coefficient, typically _2mV / ° C or equivalent -3000ppm / ° C, the current will pass through MN3 to node OUT1. In the circuit shown, R1 is selected as 44kQ to set the current through MP3 to I (MP3) = 404mV / 44k0hm = 9.20uA and the current through QP2 15 to I (QP2) = 9.20uA / 4 = 2.30 uA. In a manufacturing process, this results in Vbe (QP2) = 462mV, and R3 is therefore set to 462mV / 9.20uA = 501Ω, so at 150 ° C, I (MN3) = I (MP3). _ Then if the temperature rises above 150 ° C, the current through transistor MP1 and thus through MP3 rises, and the current through transistor MN3 decreases, resulting in a current of 20 to the external voltage source from node OUT 1 result. If the temperature drops below 150 ° c, the current through transistor MP1 and thus MP3 decreases' and the current through transistor MN3 rises, resulting in the current flowing from the voltage source into the OUT1 node. If the connection between the voltage source and the node ουτί is loosened, the voltage level of this node will rise or fall respectively, and eventually make [^^ 18 1234645 and MN3 no longer saturated to balance the current. It can be seen that the υυτι node is larger than the node a in the basic configuration of Fig. 1A. '' The choice of transistor size can depend on the needs of any particular application. In terms of integrated circuit implementation, the main considerations include the chip occupied by the components: 5 products and minimize the impact of nominal mismatch between the same devices. Typically, the random deviation voltage between the 雔 polar transistors and the resistors will be less than the deviation voltage between the transistors in the circuit, and the expansion of manufacturing will be governed by the mismatch between] ^ 1 ^ 2 and] ^] ^ 1 Because this error is basically superimposed on the small quiescent voltage through R1. 10 First consider the choice of the ratio of MN2 to MN1. As the above circuit, but the ratio between MN2 and MN1 and between MP2 and MP1 is 1, it will still be available with proper adjustment of Ri. However, the current density between QP1 and QP2 is only 4 instead of 16, so you will only get half the voltage T (q) ln4 instead of (kT / q) lnl6) through R1, making the circuit for MN2 and MN1 Mismatches between 15 are more sensitive. In order to restore the current density ratio, QP1 can be 16 times of QP2, but this will take up a lot of silicon area. On the other hand, if the ratio between MN2 and MN1 and between MP2 and MP1 is 8: 1 instead of 4 :: 1, this will only increase the voltage through R1 by a factor of 1η32 / 1η16 = 1.25, but the MOS transistor is already large To reduce manufacturing tolerances and double the area. As far as the technology under consideration, 4: 1 20 is selected, but its optimum value will depend on the limitations of the particular manufacturing technology. Now consider the ratio of MN3 to MN1. As noted above, the voltage through R1 is approximately 100mV at 150 ° C and approximately 450mV through R3, and these resistors are required to pass the same current. If MP3 and MN3 are the same size as MP1 and MN1, respectively, the resistance of R3 will be approximately 4.5 19 1234645 times the resistance of R1. For best performance when using parasitic vertical transistors in CMOS, Qp] and QP2 operate optimally with a small amount of micro-current. At the same time, multiple applications have strict power pre-differentiation, and in such applications, these resistors tend to have tens of thousands of ohms and occupy a large area. Introducing 1 ^^^ 3 pairs] ^] ^ 1 of 4: 5 The ratio of 1 to R3 and R1 is similar, so the total resistor area tends to be the best. The transistors MP2 and MP3 are preferably formed of multiple units, each having a configuration similar to that of MP1. It preferably has a large channel length for medium matching and high output impedance 'but has a small channel width to length ratio W / L at 10 to maintain Vgs-Vt as large and good current medium matching. Transistor MN2 is similar to MN3 in that multiples configured for MN1 are better 'and preferably Vgs-Vt is large and has a good current medium. However, if Vgs-Vt is large, this will cause subsequent variation of Vgs (MN3) and attenuate the temperature coefficient of I (MN3) (basically put the resistance of 1 / gm (MN3) in series with 15 connected to R3) 'Therefore, these transistors should generally be designed with a sufficiently large W / L to obtain vgs-vt < 100mV (for example, at a critical temperature). Then 丨 / gm (MN3) is about 10% of R3 ′ and does not degrade the temperature sensitivity of the circuit or introduce manufacturing sensitivity due to the non-relevant and electrical characteristics of the resistor. Reviewing the above description of the operation of the circuit, it can be seen that the thermal voltage base 20 is "servo" the current mirror, and the current mirror also drives the detection node. The reference based on the base emitter uses the same transistor as the thermal voltage reference to provide a second, negative temperature coefficient output, which is subtracted from the positive temperature coefficient based voltage reference at the detection node. It will be understood that this configuration may be exchanged such that the Vbe-based benchmark is used in conjunction with the 20 1234645
Vbe為基礎之基準相同電晶體的熱電壓基準對該電流鏡（此 鏡再次驅動該檢測節點）被伺服且亦驅動該檢測節點。此替 選的配置在第3B圖中被顯示，此處電晶體MP1上之閘極― 排極連結已被移至電晶體MP3，且其輸出由QUT2,線路302 5被取得，此為電晶體MP1與MN1之接合。該分析與成份值 維持相同，至少為第一階。主要的差異在於該電路所耗用 的電流現在具有負的而非正溫度係數。 至目前所描述之不需回饋的電路會易在亞穩定狀態附 近振盪，且正回饋因而為所欲的以提供磁帶。第3C圖顯示 10第3A圖之電路的擴充以實施此點。MOS電晶體MP4與MP9 提供來自該電流鏡之進一步輸出，其被用作為固定的電流 源。線路306被連接至與電晶體MP6不同組配之一輸出電晶 體MP5，如下面更詳細描述地被連接至提供正回饋之電晶 體MP4 ’電晶體MP6所提供的一共同電流源。電晶體MP6 15之閘極以類似於對節點306先前討論的電壓源之電壓被連 接至一偏壓線路，使得當MP5與MP6之閘極在相同電壓 日守’ MN3與MP3一者均為飽和，以避免該等溫度相依電流 在該門檻溫度或附近時之惡化。電晶體MN10與MN11包含 一進一步的電流鏡，且配合電晶體“㈧包含一輸出電路用 20於實質地驅動電源軌VDD與Vss(或接地）間之一輸出線路 310以便驅動邏輯電路。 在第3C圖之電路中，正回饋係被電晶體mp4，5與6提 供。在冷溫度時，節點〇1711將為低的，因此電晶體膽5將 為接通的，且在注意到通過電晶體MP5與MP6之波道的電 21 1234645 流(被MP4決定）之固定和下，電晶體MP6被關閉。隨著溫度 上升，電晶體MP5開始關閉且電晶體MP6開始接通，因而 導動某些電流（來自MP4)至電晶體MN2與QP2内。此使電晶 體MN2，MN1與MN3之閘極接頭的電壓上升Δν。目前忽 5 略ΜΝ1與ΜΝ3之Vgs中的任何變異及Vbe(QPl)中的任何變 異，此將以△V/(I(Rl).Rl)=AV/(AVbe)=Z\V/101nA^〇 比例提高通過MOS電晶體之電流，而提高通過電晶體MP1及 因此通過MP3之電流，而進一步鼓勵在節點0UT1中之上升。 其亦將提高通過R3之電流，但以△ V/(I(R3) · R3)= △ V/Δ Vbe 10 = △ V/462mV之較小比例。在產生I(R1)中之上升並非恰為 △ V/R1 ’原因在於該額外回饋電流提局4 : 1比值之MN2與 MN1中的電流，使得這些電晶體現在具有稍微不同的閘極 —源極電壓，且QP1與QP2之Vbe亦將不同，但整體的效應 仍為I(MP3)比起I(MN3)被提高很多。 5 此過程持續至電晶體MP 5貫質地完全被關閉且電晶體 MP6實質地承載通過電晶體MP4之全部電流為止。在此 點，MP4有效地出現成與電晶體MP2成並聯，而改變該電 流鏡之比值。因而當該溫度最終降低時，該熱解扣點 (thermal trip point)為比先前當溫度正提高時為較低的溫 度’而k供所欲的磁滯效果。其將被看出，正回饋不直接 設定正或負温度基準電流而是取代地藉由添增至來自電晶 體MP2之輸出電流而變更在電流鏡中之電流比。此變更 基準與熱電壓基準電流，但變更熱電壓基準電流較多，因 而實際上改變通過電晶體MN1與MN3及因而通過電晶體 22 1234645 MP3與MN3之電流的平衡。因而，該回饋並非直接至該以The Vbe-based reference is the same as the thermal voltage reference of the transistor, which is servoed to the current mirror (which again drives the detection node) and also drives the detection node. This alternative configuration is shown in Figure 3B, where the gate-to-row connection on transistor MP1 has been moved to transistor MP3, and its output is taken from QUT2, line 3025, which is a transistor The junction of MP1 and MN1. The analysis and component values remain the same, at least for the first order. The main difference is that the current consumed by the circuit now has a negative rather than a positive temperature coefficient. The circuits described so far that do not require feedback tend to oscillate near the metastable state, and positive feedback is thus provided to provide the magnetic tape as desired. Figure 3C shows the expansion of the circuit of Figure 3A to implement this. The MOS transistors MP4 and MP9 provide further output from the current mirror, which is used as a fixed current source. The line 306 is connected to an output transistor MP5, which is a different combination from the transistor MP6, and is connected to a common current source provided by the transistor MP4 'transistor MP6 which provides positive feedback, as described in more detail below. The gate of transistor MP6 15 is connected to a bias line with a voltage similar to the voltage source previously discussed for node 306, so that when the gates of MP5 and MP6 are at the same voltage, the MN3 and MP3 are saturated. To avoid the deterioration of these temperature-dependent currents at or near the threshold temperature. Transistors MN10 and MN11 include a further current mirror, and in conjunction with the transistor "㈧ includes an output circuit 20 to substantially drive one of the output lines 310 between the power supply rail VDD and Vss (or ground) in order to drive the logic circuit. In the circuit of Figure 3C, the positive feedback system is provided by the transistors mp4, 5 and 6. At cold temperature, the node 01711 will be low, so the transistor 5 will be turned on, and it is noted that the transistor 5 When the electricity of the channel of MP5 and MP6 is 21 1234645 (determined by MP4), the transistor MP6 is turned off. As the temperature rises, the transistor MP5 starts to turn off and the transistor MP6 starts to turn on, thus driving some Current (from MP4) into transistors MN2 and QP2. This increases the voltage at the gate junctions of transistors MN2, MN1 and MN3 by Δν. At present, any changes in Vgs of MN1 and MN3 and Vbe (QPl) are omitted. Any variation of this will increase the current through the MOS transistor by the ratio △ V / (I (Rl) .Rl) = AV / (AVbe) = Z \ V / 101nA ^ 〇, and increase the current through the transistor MP1 and therefore through The current of MP3 will further encourage the rise in node OUT1. It will also increase the pass The current of R3, but with a small proportion of △ V / (I (R3) · R3) = △ V / Δ Vbe 10 = △ V / 462mV. The rise in generating I (R1) is not exactly △ V / R1 'The reason is that the extra feedback current provides a 4: 1 ratio of the current in MN2 and MN1, so that these transistors now have slightly different gate-source voltages, and the Vbe of QP1 and QP2 will also be different, but the overall The effect is still that I (MP3) is much improved compared to I (MN3). 5 This process continues until transistor MP 5 is completely turned off and transistor MP6 essentially carries the entire current through transistor MP4. At this point MP4 effectively appears in parallel with transistor MP2 and changes the ratio of the current mirror. Therefore, when the temperature finally decreases, the thermal trip point is lower than when the temperature was increasing previously Temperature, and k provides the desired hysteresis effect. It will be seen that the positive feedback does not directly set the positive or negative temperature reference current but instead changes the current by adding to the output current from transistor MP2 Current ratio in the mirror. This changes the reference and thermovoltage reference currents, but changes the thermoelectric The reference current is more, because in fact changed through transistor MN1 and MN3 and thus balance through transistor 22 1234645 MP3 and MN3 of the current. Therefore, the feedback is not directly to the order
Vbe為基礎之基準源或直接回到輸出節點ουτι，而是取代 ~ 地回到一共用節點（線路304)與電晶體(雙極電晶體QP2)。來 · 自QP5之排極電流與通過MP9之固定電流被鏡MN10，MN11 5 比較以給予執對執邏輯信號在線路HOT搖擺。 … 第3D圖顯示類似於被施用至第3B圖之電路的回饋做 法。注意’由於在比較節點〇UT2之信號走低而高於溫度門 檻’故MP5之排極電流為現在被饋入節點3〇4者以提供正回 饋。 · 10 現在參照第4圖，此顯示與第3C圖顯示之相同基本型式 之溫度檢測器400，且其中相同的元件編號表示相同的元 件。在第4圖之電路中，第一4〇2與第二4〇4溫度調整線路被 提供以允許該電路之門檻溫度的外部調整。 溫度调整線路402控制電晶體MNX以由電晶體MP10所 15提供的電流鏡的額外輸出注入一部分的正溫度係數電流至 電阻鏈R3A，B，C内。此額外的上拉電流降低該門檻溫 度。 · 溫度調整線路404控制電晶體MN9以降低電阻器鏈尺3 Γ 之電阻或使較低部分R3 Α短路，此提高Vbe/R3電流且因而提 20 高該門檻溫度。 被線路402與404提供之溫度調整功能可被用以變更或 調變該溫度門檻，以例如提供一「早期警告」功能或在功 能性地測試製造零件時允許熱解扣電路在室溫被施行。 洋細地說，在第4圖中，電晶體Mp6之閘極被綁至在電 23 1234645 流鏡中電晶體的閘極。如上述者，MP6之閘極應被偏壓為 適當的電壓以在MP5與MP6為平衡時MP3與MN3二者為飽 和的。此處，如在所圖示的實施例中，此處理技術製作可 取得的備選「低Vt」或降低門檻電壓之PM〇S電晶體，線路 5 406上之電壓被用以供應此偏壓，而不須迫使Mp4離開此飽 和區。在不需此選配的處理中，Mp6之閘極可被連接至某 一其他適當的點。 其將被了解，如第2B與2C圖之電路具有一第二穩定狀 悲，其中所有電晶體被關閉。只要小的開始電流（例如通過 10電晶體236)足以使該電路脫離此狀態。此可經常被接合洩 漏電流或被供電時的電容性電流供應，但一「啟動」電路 可被用以確保該電路可靠地離開其零電流狀態。 第5圖顯示根據第3D圖之配置且納有此啟動電路的一 溫度檢測電路的實施例5〇〇。在第5圖中，與第3〇圖相同的 15元件用相同的元件編號被表示。在第5圖之電路中，MN5 提供小電流至PMOS鏡閘極，以其閘極電壓起始地被Mp7 上拉至Vdd。MN5僅在每一次MN4接通時被關閉，其僅在 MN3與因而之MP3已開始通過電流時發生。類似的技術用 第3C與4圖之電路被運用。其他的解決做法對熟習電路設計 2〇 工程師為易於明白的。 無疑地报多有效的變化將對熟習的人發生。例如，雖 然特定的實施例已參照PNP雙極電晶體被描述，熟習的人 將易於了解該電路可被逆轉且NPN雙極電晶體可被運用。 典型上CMOS處理的垂直寄生電晶體將被使用，但寄生側 24 1234645 式電晶體(如以排極、塊體與源極分別作用成集極、基極與 射極之MOS電晶體）或寄生二極體（由於雙極電晶體基本上 被用以提供二極體接合)在原理上可被運用，原因在於該電 路對低貝它型式的此類電晶體為不敏感的。 5 在其他實施例中，雙極電晶體QP1與QP2可用尺寸比值 後之MOS電晶體取代。較佳地是這些MOS電晶體在次門檻 區内被操作，此處其展現雙極似的指數I-V特徵，但就算在 次門檻區外時，而其將提供較小但仍為正溫度係數電流。 其將被了解本發明不受限於所描述的實施例且包容對 10 位於此處所附之申請專利範圍的精神與領域内之技藝熟習 者為明白之修改。 L圖式簡單說明】 第1A與1B圖分別顯示以電流源為基礎之溫度檢測電 路，及在第1A圖之電路中的電流源之熱特徵； 15 第2A至2C圖分別顯示一自我偏壓之基準電流源、一 Vbe 基準電流源、及一熱電壓基準電流源； 第3A至3D圖分別顯示依據本發明之不需磁滯的溫度 檢測電路之第一與第二實施例，及依據本發明之需磁滯的 溫度檢測電路之第一與第二實施例； 20 第4圖顯示依據本發明之溫度檢測電路的第三實施 例；以及 第5圖顯示依據本發明之溫度檢測電路的第四實施例。 25 1234645 【圖式之主要元件代表符號表】 10...電流源 234...電晶體 12...正溫度係數特徵 236…電晶體 20...電流源 238...電阻器 22...負溫度係數特徵 240...熱電壓基準電流源 30...反相器 242...電晶體 200...自我偏壓基準電流源 244…電晶體 202...電流鏡 246...電晶體 204...電流源 248...電晶體 206···輸入 250...電阻器 208...輸出 300...溫度檢測電路 210...輸出 302. · ·輸入 212···輸入 304...輸出 220...基極射極基準電流源 306...線路 222...正電源 308…偏壓線路 224…接地線 310…輸出線路 226...電晶體 400…溫度檢測器 228...電晶體 402...第一溫度調整線路 230...電晶體 404...第二溫度調整線路 232...電晶體 500...溫度檢測電路 26The Vbe-based reference source may return directly to the output node ουτι, but instead of ~ return to a common node (line 304) and transistor (bipolar transistor QP2). Coming · The pole current from QP5 is compared with the fixed current through MP9 by mirrors MN10, MN11 5 to give the logic signal for swinging on the line HOT. … Figure 3D shows a feedback approach similar to the circuit applied to Figure 3B. Note that ‘the signal at the comparison node OUT2 goes lower than the temperature threshold’, so the MP5 current is now fed into node 304 to provide positive feedback. · 10 Referring now to Figure 4, this shows the same basic type of temperature detector 400 as shown in Figure 3C, and the same component numbers represent the same components. In the circuit of Fig. 4, a first 402 and a second 404 temperature adjustment circuit are provided to allow external adjustment of the threshold temperature of the circuit. The temperature adjustment circuit 402 controls the transistor MNX to inject a part of the positive temperature coefficient current into the resistor chains R3A, B, and C with the extra output of the current mirror provided by the transistor MP10. This additional pull-up current reduces the threshold temperature. The temperature adjustment circuit 404 controls the transistor MN9 to reduce the resistance of the resistor chain ruler 3 Γ or short the lower portion R3 A, which increases the Vbe / R3 current and thus raises the threshold temperature by 20 °. The temperature adjustment function provided by lines 402 and 404 can be used to change or adjust the temperature threshold, for example to provide an "early warning" function or to allow the thermal trip circuit to be implemented at room temperature when functionally testing manufactured parts . In detail, in Figure 4, the gate of the transistor Mp6 is tied to the gate of the transistor in the current mirror. As mentioned above, the gate of MP6 should be biased to an appropriate voltage so that both MP3 and MN3 are saturated when MP5 and MP6 are balanced. Here, as in the illustrated embodiment, this processing technique produces an alternative "low Vt" or PMMOS transistor with a reduced threshold voltage, and the voltage on line 5 406 is used to supply this bias Without having to force Mp4 out of this saturation region. In a process that does not require this option, the gate of Mp6 can be connected to some other suitable point. It will be understood that the circuit shown in Figures 2B and 2C has a second stable state in which all transistors are turned off. As long as a small starting current (for example, through 10 transistors 236) is sufficient to take the circuit out of this state. This can often be coupled with leakage current or capacitive current supply when powered, but a "start-up" circuit can be used to ensure that the circuit reliably leaves its zero current state. Fig. 5 shows an example 500 of a temperature detection circuit configured in accordance with Fig. 3D and containing the start-up circuit. In Fig. 5, the same 15 elements as those in Fig. 30 are indicated by the same element numbers. In the circuit in Figure 5, MN5 provides a small current to the gate of the PMOS mirror, and its gate voltage is initially pulled up to Vdd by Mp7. MN5 is turned off only every time MN4 is turned on, and it only occurs when MN3 and thus MP3 have begun to pass current. Similar techniques are applied using the circuits of Figures 3C and 4. Other solutions are familiar to circuit designers. 2 Engineers are easy to understand. No doubt how effective the change will be for those who are familiar with it. For example, although specific embodiments have been described with reference to PNP bipolar transistors, those skilled in the art will readily understand that the circuit can be reversed and NPN bipolar transistors can be used. CMOS-treated vertical parasitic transistors will typically be used, but parasitic side 24 1234645 type transistors (such as MOS transistors with collector, base, and emitter acting as the collector, base, and emitter respectively) or parasitic Diodes (because bipolar transistors are basically used to provide diode junctions) can be used in principle because the circuit is insensitive to low-beta types of such transistors. 5 In other embodiments, the bipolar transistors QP1 and QP2 can be replaced by MOS transistors with a size ratio. Preferably, these MOS transistors are operated in the subthreshold region, where they exhibit bipolar exponential IV characteristics, but even outside the subthreshold region, they will provide smaller but still positive temperature coefficient currents. . It will be understood that the invention is not limited to the described embodiments and encompasses modifications apparent to those skilled in the art that are within the spirit and scope of the patentable scope attached hereto. Simple illustration of L diagram] Figures 1A and 1B respectively show the current source-based temperature detection circuit and the thermal characteristics of the current source in the circuit of Figure 1A; 15 Figures 2A to 2C each show a self-bias voltage Reference current source, a Vbe reference current source, and a thermal voltage reference current source; Figures 3A to 3D show the first and second embodiments of the temperature detection circuit without hysteresis according to the present invention, and according to this The first and second embodiments of the temperature detection circuit requiring hysteresis of the invention; 20 FIG. 4 shows a third embodiment of the temperature detection circuit according to the present invention; and FIG. 5 shows the first and second embodiments of the temperature detection circuit according to the present invention. Four embodiments. 25 1234645 [Representative symbols for main components of the drawing] 10 ... current source 234 ... transistor 12 ... positive temperature coefficient characteristic 236 ... transistor 20 ... current source 238 ... resistor 22. .. Negative temperature coefficient characteristic 240 ... Thermal voltage reference current source 30 ... Inverter 242 ... Transistor 200 ... Self-biased reference current source 244 ... Transistor 202 ... Current mirror 246. ..Transistor 204 ... current source 248 ... transistor 206 ... input 250 ... resistor 208 ... output 300 ... temperature detection circuit 210 ... output 302 ... input 212 ··· Input 304 ... Output 220 ... Base emitter reference current source 306 ... Line 222 ... Positive power supply 308 ... Bias line 224 ... Ground line 310 ... Output line 226 ... Transistor 400 ... temperature detector 228 ... transistor 402 ... first temperature adjustment circuit 230 ... transistor 404 ... second temperature adjustment circuit 232 ... transistor 500 ... temperature detection circuit 26