• 沒有找到結果。

適用於無線通訊之互補式金氧半積體電路-總計畫

N/A
N/A
Protected

Academic year: 2021

Share "適用於無線通訊之互補式金氧半積體電路-總計畫"

Copied!
5
0
0

加載中.... (立即查看全文)

全文

(1)

行政院國家科學委員會補助專題研究計畫成果報告

※※※※※※※※※※※※※※※※※※※※※※※※※※

※ ※

適用於無線通訊之互補式金氧半積體電路

※ ※

※※※※※※※※※※※※※※※※※※※※※※※※※※

計畫類別:□個別型計畫 ■整合型計畫

計畫編號:NSC 90-2215-E-009-115

執行期間:90 年 08 月 01 日至 91 年 07 月 31 日

計畫主持人: 吳重雨 教授

計畫參與人員:周忠昀、徐國鈞、詹歸娣

本成果報告包括以下應繳交之附件:

□赴國外出差或研習心得報告一份

□赴大陸地區出差或研習心得報告一份

□出席國際學術會議心得報告及發表之論文各一份

□國際合作研究計畫國外研究報告書一份

執行單位:

中 華 民 國 91 年 08 月 21 日

(2)

行政院國家科學委員會專題研究計畫成果報告

計畫編號:NSC 90-2215-E-009-115

執行期限:90 年 08 月 01 日至 91 年 07 月 31 日

主持人:吳重雨 教授 國立交通大學電子研究所

計畫參與人員:周忠昀、徐國鈞、詹歸娣

國立交通大學電子研究所

中文摘要 由於超大型積體電路技術的微小化,使得 電路的操作頻率可增高至射頻區域。而從已經 發表的論文中可以看出,在無線通訊的領域 中,2.4GHz 已經成為一個熱門而又確實能夠實 現的射頻頻段,但在可預見的將來,5.2 ~ 5.8 GHz 的免付費頻段也即將成為下一個重要的研 究里程碑。而在 RF IC 的應用上,靜電保護電 路不只需要有高的損壞臨界電壓以確保其良好 的性能,還必須擁有低寄生電容和電阻,以降 低 RC 延遲。所以一個 RFIC 的完成,實為整合 RFIC 電路設計、ESD 設計、以及 VLSI 製程三 個領域。 此計劃的目標主要是驗證以 0.18 ~ 0.13µm 場 效金氧半電晶體的製程來實現 5.2 ~ 5.8 GHz 高 頻段無線射頻金氧半電晶體收發機的可行性, 配合對半導體製程的改進,經由電路設計加上 靜電保護電路,完成一實際可用的 RFIC。 關鍵詞:射頻技術,次微米電晶體,射頻收發 機,靜電放電(ESD), 靜電放電防護電路 Abstract

As the VLSI technology continues scaling down, the circuit operation frequency improves and can be used in RF range. However, as various wireless standards continue to populate the 2.4GHz range, the next natural step is to extend the communications to the unlicensed 5.2 ~ 5.8 GHz band. Electrostatic discharge (ESD) phenomena continue to be a main reliability issue in CMOS IC’s because of technology scaling and high frequency requirements. For RF IC’s, the ESD protection design has some limitations: low parasitic capacitance, constant input capacitance, insensitive to substrate coupling noise, and high ESD robustness. So to implement RF IC’s is to integrate the knowledge of VLSI technology、RF IC’s circuit design and ESD protection design。

It is the aim of this project to evaluate the 0.18 ~ 0.13µm CMOS process for the development of RF transceiver front end for fixed wireless applications in the 5.2 ~ 5.8 GHz frequency band. Combine the improvement of VLSI technology, RF IC’s circuit design and ESD protection design to realize a useful RF IC.

Keywords: RF technology, sub- m MOSFET, RF transceiver, Electrostatic discharge (ESD), ESD Protection Circuit

一、緣由與目的

此三年整合計畫包括一個總計劃與三個子計 畫,分別是交通大學電子研究所荊鳳德教授的: Deep Sub-micro meter RF Devices 子計畫,和交 通大學電子研究所吳重雨教授的:無線應用的 CMOS Transceiver 前端電路設計(The design of CMOS Transceiver Front-end for Wireless

Applications)子計畫,與交通大學電子研究所柯 明道助理教授的:適用於 GHz 頻段之輸出入靜 電放電防護技術與銲墊設計子計畫。此三個子計 畫研究核心相同,包含了適用於現今無線通訊的 射頻元件製造、電路設計及保護電路。 二、研究成果 1. 子計畫一

The measured frequency response of H21 and Gmax are summarized in Table 1.

measured/ calculated values mea. H21 (dB) 4GHz mea. fT (GHz) H21 =1 cal. fT (GHz) H21 =1 mea. fmax (GHz) MAG =1 mea. fmax (GHz) MSG =1 cal. fmax (GHz) MSG =1 mea. Gmax (dB) 4GHz cal. Gmax (dB) 4GHz 0.5-µm 14.7 25 23 20 82 80 13.0 13.9 0.25-µm 19.7 42 38 18 119 127 15.0 15.9 0.18-µm 22.2 58 56 17 161 171 16.3 18.0

H21, ft,Gmax, and fmax gradually saturate as device scaling down. The saturation rate is faster for Gmax

(3)

and a reducing fmax is even observed. Furthermore, the measured H21 and ft are about 50% lower than the calculated value from conventional equation of gm/2 Cgs or vsat/2 ?Lg-2Lov), where Lov is the gate-drain overlap length. Rnqs becomes more important as increasing frequency near ft. Good matching between measured and simulated ft in Table 1 can only be obtained by considering the NQS effect. Because of the additional term, ft increases slower than 1/ Lg scaling down.

To further analyze the frequency response, we have also derived Gmax and fmax by using the equivalent circuit modeling and including the NQS effect. From derived Gmax, Cgd related pole gives the 10dB/decade Gmax roll-off in MSG, while the large slope of ~30-40dB/decade in MAG is due to additional poles in K or the NQS effect on gm.

Good agreement between the measured and calculated fmax,MSG=1 is achieved and shown in Table 1. The primirary parameter for fmax,MSG=1 increase is due to the gm increase and Cgd decrease. In fact, Cgd is dominated by the Cgdo that is difficult to proportionally scale down with Lg.

W e h a v e a l s o u s e d numerical device simulation for further analysis. We have studied the NQS effect on Gmax and fmax. The MAG increases with decreasing Rnqs and eventually gives Gmax the same 10dB/decade roll-off as MSG when Rnqs equals 0. Therefore, the NQS effect is responsible for the transition from MSG to MAG. Because Rnqs is inversely related to Cgs, a higher dielectric or thinner gate thickness is required to improve the high frequency gain.

On the other hand, Gmax has a simple analytical solution in the most useful MSG region for amplifier design. Because the Rnqs(Cgs+Cgd) related zeros are effective only at high frequencies, Gmax in MSG can be further simplified and expressed by gm/ Cgd or vsat/ Lov. The numerical simulation results show that the reduction of Cgdo leads to a higher Gmax and fmax. However, the difference between the ideal 2CoxWtox and the measured data is larger as scaling down.

Here, a minimum Cgdo of CoxWLov (Lov=2tox) is required in order to develop a reproducible and manufacturable process, where Cox and tox are the gate capacitance and oxide thickness, respectively. Although down scaling gives a smaller Lg and a higher Cox, limited Gmax improvement in MSG is due to the slower scalable Lov. The reason for Lov

failing to follow tox scaling down in deep sub- m devices is due to the lateral diffusion from source and drain impurities. High temperature annealing after source and drain implantation is necessary to reduce the junction leakage but largely increases the lateral diffusion. The formation of silicide junction also requires high temperature RTA. Because of the combined small Gmax and K factor improvement, limited fmax improvement as device scaling down can be expected.

The smaller increase of measured Gmax than

calculated value in Table 1 as down scaling may be due to the parasitic effect neglected in our device model.

This paper has been submitted to IEEE MTT-S 2000. 2. 子計畫二 此計畫以 TSMC 018um 的製程來實現操作 於 5-GHz 的收發機前端電路。接收機採用雙正 交的架構(如下圖)。此架構利用多相位濾波器來 濾除鏡像雜訊,並能避免直接轉換架構的直流電 壓問題,具有高整合性的優點;除此之外,將輸 入射頻訊號轉成正交訊號後,可降低路徑不匹配 的缺點,提高訊雜比。 + + Q u a d ra tu re G e n e ra to r LNA P o ly p h a s e F ilt e r t cos?LO sin?L Ot I Q 雙正交接收機架構圖 在計畫執行期間完成的電路有低雜訊放 大器、正交相位調變器、多相位濾波器,其電路 及性能如下: a. 低雜訊放大器 L1 L2 M3 M4 M1 M2 L3 L4 Mb Cpad Lbw Cblk Lbias VB1 VB2 VIN+ VIN-RF+ RF-低雜訊放大器

(4)

在低雜訊放大器中,對雜訊影響最大的為輸 入端的兩個電晶體 M1,M2,在固定功率消耗的 考量下(5mW),選擇最適當的閘極寬度,可得到 最佳的雜音指數。Spiral Inductor L1,L2 與位於輸 出端的寄生電容產生 5.05 GHz 的自振頻率。而 L3,L4 及 boundwire Inductors Lbw 則提供阻抗匹 配以利最大功率傳輸。性能如下: Gain 18.7 dB Noise Figure 3.37 dB IIP3 -5 dBm Power Dissipation 5 mW b.正交相位調變器 RF_I+ RF_I- RF_Q- RF_Q+

LO_I+ LO_I- LO_Q+

LO_Q-Mmx1 RF_Q- RF_Q+ RF_I+ RF_I-LO_I+ LO_I-LO_Q+ LO_Q-Rn

LO_I+ LO_I- LO_Q+

LO_Q-Mmx2 Mmx3 Mmx4 Vbias Vctrl Vctrl Mvco9 M v c o 1 3 M v c o 1 4 M v c o 1 5 M v c o 1 6 Lvco1 Mmx5 Mmx6 Mmx7 Mmx8 Rp Lvco2 Mvco10 Mvco11 Mvco12 Mvco5 Mvco6 Mvco7 Mvco8

Mvco1 Mvco2 Mvco3 Mvco4

正交混波器及正交電壓控制震盪器 在此電路中 Mvco1~Mvco16 為電壓控制震盪 器,Mvco9~Mvco12 產生負電阻抵消掉寄生電 阻提供震盪的條件,電感 Lvco1,Lvco2 及 Mvco13~Mvco16 產生的可變電容決定震盪頻 率,Mvco5~Mvco8 則產生正交的本地震盪訊 號。Mmx1~Mmx8 為混波器,LO 訊號由源極進 入,射頻訊號由閘極進入,混波器與 VCO 共用 同一電流以節省功率。性能如下: Power Dissipation 4.5 mW

Tuning Range 250 Mega Conversion Gain -0.35 dB c.多相位濾波器 複數濾波器的操作原理如下圖: ) ( jH ) ( H1ω + 2 ω ) ( jQ ) ( + ω )) ( H ) ( Q ) ( H ) ( I ( j ) ( H ) ( Q ) ( H ) ( I 1 2 2 1 ω ω ω ω ω ω ω ω + + − 複數濾波器的操作原理 適當的選擇實數濾波器的轉移函數即可以完成 對正負頻率作篩選的功能,電路如下: M1 M2 M3 M4 Mb M7 RH-RL- RH+ RL+ ML1 ML3 MLO1 C Vi+

Vi-Io- Qo- Qo+ Io+

2.1*C 2.1*C Vb_i Vb1 Vq+ Vq-Vb1 Vb_q Vb1 Vb1 M5 M6 M8 ML2 ML3 MLO2 2.1*C M1 M2 M3 M4 Mb M7 ML1 ML3 MLO1 C 2.1*C 2.1*C M5 M6 M8 ML2 ML3 MLO2 Mcm3 Mcm1 Mcm2 Mcmb Io+ Io-to Vb_i Mcm3 Mcm1 Mcm2 Mcmb qo+ qo-to Vb_q Common Mode Feedback 多相位濾波器電路圖 性能如下:

Channel Bandwidth 26 Mega Hz Gain 9.36 dB IRR > 60 dB 此部分的研究成果已撰寫成論文並發表[1]。 3. 子計畫三 本計畫的研究成果已經整理且發表了二篇 IEEE Conference 論文[2][3]以及三篇國際期刊論 文[4]-[6]。 在第一部份成果中[2][4],我們設計了一組 可供 CMOS 射頻電路所使用的 ESD 保護電路, 此保護電路主要是由堆疊的複晶二極體作為輸 入/輸出的 ESD 保護電路以達到減少寄生電容及 基板共擾雜訊干擾的影響,如下圖: 堆疊的複晶二極體作為 RF ICs 輸入端的 ESD 保護元件 另外搭配電源線間的 ESD 箝制電路使用,便可 以提高整體的靜電耐受度,電路如下:

(5)

電源線間的 ESD 防護電路 在第二部份成果中[3][5][6],隨著半導體當 電晶體元件尺寸被縮小,其操作電壓也隨之降低 以維持元件的可靠性,可是外界傳輸的電路訊號 之電壓準位大都依然維持在 5V,這 5V 的電路訊 號不能直接傳入深次微米的積體電路內部,因此 在積體電路的輸出入端必需要有混合電壓輸出 入界面電路來隔離輸入訊號的高電壓準位,但又 必需把輸入訊號的電路訊號送入該積體電路內 部。因此我們針對 5V/3.3V 混合式電壓輸出入界 面電路的需求,提出了一組使用 SNTSCR 元件的 新穎性 ESD 保護電路,如下圖: SNTSCR 元件的等效電路圖 將堆疊的 NMOS 嵌入 SCR 元件中,因此 SNTSCR 元件在混合式電壓輸出入界面電路中僅需使用 薄的閘極氧化層就能避免氧化層可靠度的問 題,而且可以減少生產成本。 為了讓 SNTSCR 元件更快導通將靜電的能 量排掉,在本篇論文中也提出了新的閘極耦合 (gate-coupled)觸發電路,如下圖: 混合式電壓輸出入界面電路的 ESD 防護電路 Vg1 偏壓在 VDD, Vg2 偏壓在 VSS,當 I/O 端 的訊號為 5V 時,此保護電路也不會有氧化層可 靠度的問題,而且 SNTSCR 元件處於關閉狀態, 不會影響內部電路運作。在靜電放電的情況下, 透過此閘極耦合觸發電路對堆疊 NMOS 的閘極 提供大於其 threshold voltage 的偏壓,降低 SNTSCR 元件的第一次崩潰電壓,使其更快導通 以提供內部電路更有效的保護。 三、參考文獻

[1] Chung-Yun Chou, Chung-Yu Wu, “The Design of a New Wideband and Low-Power CMOS Active Polyphase Filter for Low-IF Receiver Applications”, accepted by APCCAS 2002.

[2] M.-D. Ker, T.-Y. Chen, and C.-Y. Chang, “ESD protection design for CMOS RF integrated circuits,” in Proc. of EOS/ESD Symp., 2001, pp. 346-354. [3] M.-D. Ker, C.-H. Chuamg, and H.-C. Jiang, “ESD

protection design for mixed-voltage I/O buffer by using stacked-NMOS triggered SCR device,” in

Proc. of EOS/ESD Symp., 2001, pp. 32-43.

[4] M.-D. Ker and C.-Y Chang, “ESD protection design for CMOS RF integrated circuits using polysilicon diodes,” Journal of Microelectronics

Reliability, vol. 42, pp. 863-872, 2002.

[5] M.-D. Ker and C.-H. Chuang, “Stacked- NMOS triggered silicon-controlled rectifier for ESD protection in high/low-voltage- tolerant I/O interface,” IEEE Electron Device Letters, vol. 23, pp. 363-365, 2002.

[6] M.-D. Ker and C.-H. Chuang, “ESD protection design for mixed-voltage CMOS I/O buffers,” IEEE Journal of Solid-State Circuits, vol. 37, no. 8, pp.1046-1055, 2002.

參考文獻

相關文件

(C)John’s love for graffiti began in junior high s chool when his school held a graffiti contest.. (D)John’s love for crows start s in junior high school when he joined

maintenance and repair works should be carried out by school and her maintenance agent(s) to rectify defect(s) as identified in routine and regular inspections. Examples of works

ADSL(A symmetric D igital S ubscriber L ine ,非對稱數位

傳統的 RF front-end 常定義在高頻無線通訊的接收端電路,例如類比 式 AM/FM 通訊、微波通訊等。射頻(Radio frequency,

Sugii, “Junction profile engineering with a novel multiple laser spike annealing scheme for 45-nm node high performance and low leakage CMOS technology,” in IEDM

IEEE 1394 Controller IC、無線週邊控制晶片 Wireless Peripheral Controller IC、滑鼠控制晶片 Mouse Controller IC、鍵盤控制晶片 Keyboard Controller IC、掃描器控制晶片

for compound B, Rf = b/S.. Solvent is allowed to move 10 cm from the origin up a TLC plate. The time taken to develop the plate for this distance is 15 min..  The more polar

The main purpose of this study is to explore the status quo of the food quality and service quality for the quantity foodservice of the high-tech industry in Taiwan;