陳迪平 張仁梓 曹倫武 陳卓俊 曾健平
摘? ?要:為了降低流水線模數(shù)轉(zhuǎn)換器功耗與提升輸入信號(hào)范圍,設(shè)計(jì)了一種無(wú)采樣保持運(yùn)放前端電路. 移除采樣保持運(yùn)放降低了功耗,并改進(jìn)開(kāi)關(guān)時(shí)序進(jìn)一步降低電路功耗;同時(shí)改進(jìn)傳統(tǒng)開(kāi)關(guān)電容比較器輸入,使得模數(shù)轉(zhuǎn)換器可達(dá)到0 ~ 3.3 V滿電源電壓的量化范圍. 將設(shè)計(jì)的無(wú)采樣保持運(yùn)放前端電路應(yīng)用在一款低功耗12位50 MS/s流水線模數(shù)轉(zhuǎn)換器進(jìn)行驗(yàn)證,采用0.18 μm 1P6M工藝進(jìn)行流片,芯片面積為1.95 mm2. 測(cè)試結(jié)果表明:3.3 V電壓下,采樣率為50 MS/s、輸入信號(hào)頻率為5.03 MHz時(shí),信噪失真比(SNDR)為64.67 dB,無(wú)雜散動(dòng)態(tài)范圍(SFDR)為72.9 dB,功耗為65 mW.
關(guān)鍵詞:流水線模數(shù)轉(zhuǎn)換器;無(wú)采樣保持運(yùn)放;孔徑誤差;開(kāi)關(guān)電容比較器
中圖分類(lèi)號(hào):TN495? ? ? ? ? ? ? ? ? ? ? ? ? ?文獻(xiàn)標(biāo)志碼:A
Front-end Circuit without Sample-and-hold
Amplifier for Pipelined ADC
CHEN Diping1,ZHANG Renzi1?,CAO Lunwu2,CHEN Zhuojun1,ZENG Jianping1
(1. College of Physics and Microelectronics Science,Hunan University,Changsha? 410082,China;
2. Hunan Advancechip Electronic Technology Company,Changsha 410205,China)
Abstract:A front-end circuit without Sample-and-Hold Amplifier(SHA)? is presented for reducing power consumption and increasing input range of pipelined ADC. Removing Sample-and-Hold Amplifier and improving switching timing reduce the circuit power consumption,while improving the traditional switched-capacitor comparator inputs to make ADC achieve a 0-3.3V full supply voltage quantization range. The front-end circuit without Sample-and Hold Amplifier is verified in a low power 12 bit 50 MS/s pipelined ADC. The circuit is implemented in a 0.18 μm 1P6M process,and occupies a chip area of 1.95 mm2. The test? results with a 5.03 MHz input wave under a sampling rate of 50 MS/s show that the ADC achieves a 64.67 dB signal-to-noise and distortion ratio(SNDR)? as well as a 72.9 dB spurious-free dynamic range(SFDR),while it consumes the power consumption of 65 mW.
Key words:pipelined Analog to Digital Converter(ADC);Sample-and-Hold Amplifier(SHA)-less;aperture error;switched-capacitor comparator
高速高精度模數(shù)轉(zhuǎn)換器(Analog to Digital Converter,ADC)是如今眾多高性能數(shù)?;旌想娐废到y(tǒng)中不可或缺的部分[1-4].隨著工藝水平的提高以及便攜式電子設(shè)備的興起,高速低功耗ADC[5-6]也正成為研究熱點(diǎn). 常見(jiàn)的ADC有流水線型(Pipelined)[7-8]、逐次比較型(Successive Approximation Register,SAR)[9]、混合型(Pipelined-SAR)[10]、過(guò)采樣型(Sigma Delta,Δ-Σ)[11]等,而Pipelined ADC因其較快的速度、較高的精度以及適中的功耗而被廣泛應(yīng)用.
為達(dá)到低功耗Pipelined ADC設(shè)計(jì)要求,研究者們提出了許多降低功耗方法,如運(yùn)放共享技術(shù)[12]、無(wú)前端采樣保持運(yùn)放技術(shù)(Sample-and-Hold Amplifier-less,SHA-less)[13-14]等. 傳統(tǒng)Pipelined ADC設(shè)計(jì)中,為使子ADC和余量增益電路(Multiplying Digital-to-Analog Converter,MDAC)兩條路徑傳輸信號(hào)一致,信號(hào)通過(guò)采樣保持電路(Sample-and-Hold Amplifier,SHA)后才開(kāi)始逐級(jí)量化. 由于SHA電路位于ADC最前端,為達(dá)到高速高精度目的,系統(tǒng)對(duì)SHA中運(yùn)放的增益與帶寬具有很高要求,從而消耗巨大功耗. 因此,降低Pipelined ADC功耗的一種有效方法是移除采樣保持運(yùn)放. SHA電路具有穩(wěn)定兩條信號(hào)路徑一致性和電平轉(zhuǎn)移功能,移除采樣保持運(yùn)放會(huì)帶來(lái)孔徑誤差[14]及輸入信號(hào)范圍變小等問(wèn)題. 因此,采用SHA-less電路,需要進(jìn)一步優(yōu)化設(shè)計(jì).
本文設(shè)計(jì)了一種適用于低功耗高精度SHA-less Pipelined ADC的前端電路,改進(jìn)開(kāi)關(guān)時(shí)序減小首級(jí)ADC比較器導(dǎo)通時(shí)間來(lái)降低功耗和消除孔徑誤差,同時(shí)改進(jìn)傳統(tǒng)開(kāi)關(guān)電容比較器輸入,使得ADC可量化輸入信號(hào)電平達(dá)到0~3.3 V滿電源電壓. 所設(shè)計(jì)的電路應(yīng)用在一款中低端DSP芯片中的低功耗12位50MS_s_Pipelined ADC進(jìn)行驗(yàn)證,采用0.18 μm 1P6M工藝進(jìn)行電路設(shè)計(jì)仿真及流片,測(cè)試結(jié)果良好.
1? ?傳統(tǒng)前端SHA電路
圖1所示為傳統(tǒng)前端SHA電路原理圖. 采樣相時(shí),開(kāi)關(guān)S1閉合,S2斷開(kāi),采樣電容CS對(duì)輸入信號(hào)進(jìn)行采樣;保持相時(shí),開(kāi)關(guān)S1斷開(kāi),S2閉合,電容CS上的電荷往反饋電容CF上進(jìn)行轉(zhuǎn)移,最終得到輸出Vout = Vin CS /CF. 在下一次采樣時(shí)刻到來(lái)前,輸出將保持不變. 因此,在保持相,信號(hào)往首級(jí)子ADC與MDAC傳輸時(shí),兩條路徑不會(huì)存在偏差. 同時(shí),若輸入信號(hào)電平超出首級(jí)ADC量化范圍時(shí),可以通過(guò)設(shè)置合理的CS /CF比值,將輸入信號(hào)進(jìn)行電平變換后再進(jìn)行量化.
由于SHA電路位于Pipelined ADC最前端,因此系統(tǒng)對(duì)SHA電路中運(yùn)放的增益與帶寬具有很高要求,導(dǎo)致SHA電路消耗巨大功耗,通常占據(jù)整個(gè)系統(tǒng)功耗的30%[15]以上. 因此,在低功耗Pipelined ADC設(shè)計(jì)中,移除前端SHA電路中的運(yùn)放顯得很有必要. SHA-less的前端電路信號(hào)傳輸路徑如圖2所示,輸入信號(hào)傳輸存在兩條路徑,一條為信號(hào)直接送入首級(jí)ADC進(jìn)行量化,另一條為送入采樣電路. 若兩條通道時(shí)間常數(shù)不匹配,將會(huì)產(chǎn)生孔徑誤差,從而降低整個(gè)系統(tǒng)量化精度.
由式(1)和圖3可知,ADC可量化的最大信號(hào)輸入電平為ΔVREF = VREFP - VREFN. 增大ADC可量化的輸入信號(hào)電平,只能增大基準(zhǔn)電壓VREFP、VREFN差值,但不能達(dá)到滿電源電壓輸入. 考慮到Pipelined ADC由多級(jí)子ADC組成,若保持圖3中電阻串阻值不變,增大基準(zhǔn)電壓VREFP - VREFN差值將會(huì)明顯增加功耗,同時(shí)對(duì)基準(zhǔn)電壓VREFP、VREFN產(chǎn)生電路的電流驅(qū)動(dòng)能力要求也增高,從而導(dǎo)致基準(zhǔn)電壓產(chǎn)生電路面積大幅增加;若增大電阻串阻值以降低功耗和基準(zhǔn)電壓產(chǎn)生電路的電流驅(qū)動(dòng)能力,由圖4所示保持相比較器等效輸入可知,時(shí)間常數(shù)τ = Rep·VREFP、Co將會(huì)大幅增加,比較器工作速度降低,從而降低ADC量化速度. 因此,圖3所示比較器電路不適合高輸入信號(hào)電平的SHA-less Pipelined ADC.
2? ?SHA-less Pipelined ADC前端電路
圖5為本文設(shè)計(jì)的SHA-less前端電路結(jié)構(gòu)圖,電路為全差分結(jié)構(gòu),為簡(jiǎn)化這里只給出單端電路圖. 電路共有16個(gè)比較器,可實(shí)現(xiàn)4 bit首級(jí)子ADC功能. 圖中給出了部分開(kāi)關(guān)時(shí)序,ADC采樣開(kāi)關(guān)φ1在采樣電容CS開(kāi)關(guān)φs后導(dǎo)通,降低了比較器導(dǎo)通工作時(shí)間,進(jìn)一步減小部分電路功耗. 因?yàn)椴蓸訒r(shí)刻不一致產(chǎn)生的誤差,可以通過(guò)兩條采樣路徑阻抗不匹配誤差進(jìn)行抵消以達(dá)到數(shù)字邏輯可校準(zhǔn)范圍. 比較器改進(jìn)輸入可實(shí)現(xiàn)滿電源電壓輸入,具體工作原理后面將做詳細(xì)討論.
2.1? ?孔徑誤差的消除
根據(jù)校正原理,總誤差需滿足式(2):
其中,總誤差包括比較器輸入失調(diào)、孔徑誤差、基準(zhǔn)源誤差等;VFS是ADC滿輸入量程;AV是MDAC增益. 本文設(shè)計(jì)中,ADC輸入為滿電源電壓量程,即0~3.3 V;首級(jí)MDAC增益為8. 假設(shè)因?yàn)閮蓷l信號(hào)路徑不匹配帶來(lái)的孔徑誤差Verror_mismatch占總誤差的一半,則可得:
Verror_mismatch < 51.56 mV? ? ? (3)
圖5給出的部分開(kāi)關(guān)時(shí)序圖,與傳統(tǒng)開(kāi)關(guān)時(shí)序不同,為了減少比較器導(dǎo)通時(shí)間以降低ADC功耗,比較器采樣開(kāi)關(guān)φ1在采樣電容開(kāi)關(guān)φS后導(dǎo)通. 此處引起的誤差可以通過(guò)采樣路徑不匹配進(jìn)行抵消. 假設(shè)輸入信號(hào)為一正弦波:
Vin = A0 sin ωt? ? ? (4)
采樣電容CS和ADC分別在t1、t2時(shí)刻對(duì)輸入信號(hào)進(jìn)行采樣,兩條路徑產(chǎn)生的延遲分別為τ1、τ2 . 因此產(chǎn)生的誤差為:
Verror_mismatch ≌ A0 [sin ω(t1+τ1)-sin ω(t2+τ2)]? (5)
輸入正弦波信號(hào)在過(guò)零處,式(5)所示的誤差將會(huì)呈現(xiàn)最大值:
Verror_mismatch_max ≌ A0 [sin ω(t1-t2)+sin ω(τ1-τ2)]? (6)
通過(guò)合理設(shè)置開(kāi)關(guān)φS和φ1尺寸,使得Δτ = τ1-τ2 = Rons CS - Ron1 CO與Δt = t1 - t2相抵消,以使誤差最小. 其中,Rons、Ron1分別為開(kāi)關(guān)φS和φ1的導(dǎo)通阻抗. 本設(shè)計(jì)通過(guò)不斷提取參數(shù)進(jìn)行后仿真來(lái)優(yōu)化電路設(shè)計(jì)與版圖匹配,使得在50 MS/s采樣率下,輸入正弦波頻率達(dá)10 MHz時(shí),孔徑誤差為11 mV,滿足式(3)要求且有較大余量.
2.2? ?首級(jí)子ADC可量化輸入信號(hào)電平的提升
由第1節(jié)的分析可知,采用SHA-less結(jié)構(gòu)后,首級(jí)ADC能量化的最大輸入信號(hào)電平將會(huì)變小. 如圖5所示,本設(shè)計(jì)對(duì)傳統(tǒng)電容開(kāi)關(guān)比較器輸入進(jìn)行改進(jìn),通過(guò)設(shè)置合理的VREFP、VREFN值,可以使得ADC達(dá)到0 ~ 3.3 V的滿電源電壓輸入.
比較器輸出如式(7)所示.
由式(7)可知,輸入信號(hào)ΔVin = Vinp - Vinn,減去一個(gè)固定直流量ΔVREF = VREFP - VREFN后再與參考電壓做比較,實(shí)現(xiàn)了輸入信號(hào)電平變換的功能. 因此,合理設(shè)置VREFP、VREFN的值,可以實(shí)現(xiàn)ADC滿電源電壓的輸入. 本設(shè)計(jì)中,Vrefp - Vrefn的取值為:
因此可得VREFP - VREFN? = 1.65 V,結(jié)合電路設(shè)計(jì)的VREFP + VREFN共模值,便可確定VREFP、VREFN的值. 圖6為0 ~ 3.3 V滿幅輸入時(shí),運(yùn)放輸出的曲線圖. 由圖可以看出,ADC可以實(shí)現(xiàn)0 ~ 3.3 V滿電源電壓量化范圍.
3? ?SHA-less前端電路應(yīng)用及測(cè)試結(jié)果
本文設(shè)計(jì)的SHA-less前端電路已成功應(yīng)用于一款12位50 MS/s的pipelined ADC芯片中,結(jié)構(gòu)圖如圖7所示. 12位50 MS/s的pipelined ADC采用4+3+3+3的四級(jí)流水結(jié)構(gòu),為降低ADC整體功耗,首級(jí)采用了本文設(shè)計(jì)的4bit SHA-less結(jié)構(gòu),電路0.18 μm 1P6M工藝進(jìn)行設(shè)計(jì)及流片.
圖8為芯片照片圖,面積為1.95 mm2.
圖9為ADC測(cè)試的線性度曲線,微分非線性(Differential NonLinearity,DNL)誤差為-0.61/+0.61LSB,積分非線性(Integral NonLinearity INL)誤差為-0.82/+0.7LSB.
圖10顯示了在3.3 V電源電壓下,采樣率為50 MS/s、輸入信號(hào)為5.03 MHz時(shí)的測(cè)試FFT頻譜圖,信噪比(Signal to Noise Ratio,SNR)為65.03 dB,信噪失真比(Signal to Noise and Distoration Ratio,SNDR)為64.67 dB,無(wú)雜散動(dòng)態(tài)范圍(Spurious Free Dynamic Range,SFDR)為72.9 dB,整體功耗為65 mW. 功耗與具有采樣保持運(yùn)放的相似架構(gòu)流水線ADC相比[3],降低了40%.
表1給出了文中所設(shè)計(jì)ADC與部分參考文獻(xiàn)提出的ADC性能對(duì)比,可以看出本文設(shè)計(jì)具有較小的功耗與精度、面積以及最大的輸入范圍.
4? ?結(jié)? ?論
本文提供了一種適用于低功耗Pipelined ADC設(shè)計(jì)的無(wú)采樣保持運(yùn)放的前端電路. 改進(jìn)采樣開(kāi)關(guān)時(shí)序減小了電路功耗,同時(shí)改進(jìn)傳統(tǒng)開(kāi)關(guān)電容比較器輸入,使得ADC可量化0~3.3 V滿電源電壓輸入信號(hào)電平. 將所設(shè)計(jì)的SHA-less前端電路應(yīng)用在一款12位50MS/s Pipelined ADC進(jìn)行驗(yàn)證,采用0.18 μm 1P6M工藝進(jìn)行流片,整體ADC面積為1.95 mm2,測(cè)試結(jié)果良好.
參考文獻(xiàn)
[1]? ? MOHAMED M Y,HEGAZI E,EI-NOZAHI M. A 0.35 mW 8-bit 140MS/s asynchronous hybrid ADC for low energy radio applications[C]// 2019 36th National Radio Science Conference. Piscataway: IEEE,2019:315—323.
[2]? ? JEON B K,HONG S K,KWON O K. A low-power 12-Bit extended counting ADC without calibration for CMOS image sensors[J]. IEEE Transactions on Circuits and Systems II:Express Briefs,2018,65(7):824—828.
[3]? ? MATHUR S,DAS M,TADEPARTHY,et al. A 115mW 12-bit 50 MSPS pipelined ADC[C]// 2002 IEEE International Symposium on Circuits and Systems. Piscataway:IEEE,2002:913—916.
[4]? ? CHEN T,PENG L L,LI H B,et al. A 2-V 40-MS/s 14-bit pipelined ADC for CMOS image sensor[C]// 2015 IEEE 11th International Conference on ASIC. Piscataway:IEEE,2015:1—4.
[5]? ? OHHATA K,HAYAKAWA D,SEWAKI K,et al. A 900-MHz,3.5-mW,8-bit pipelined subranging ADC combining flash ADC and TDC[J]. IEEE Transactions on Very Large Scale Integration(VLSI) Systems,2018,26(9):1777—1787.
[6]? ? ZHOU Y,XU B W,CHIU Y. A 12-b 1-GS/s 31.5-mW time-interleaved SAR ADC with analog HPF-Assisted skew calibration and randomly sampling reference ADC[J]. IEEE Journal of Solid-State Circuits,2019,54(8):2207—2218.
[7]? ? 陽(yáng)輝,何怡剛,黃姣英. 0.25μm CMOS工藝10位150MHz流水線型ADC設(shè)計(jì)[J]. 湖南大學(xué)學(xué)報(bào)(自然科學(xué)版),2010,37(8):45—48.
YANG H,HE Y G,HUANG J Y.A 10-b 150MHz pipeline ADC in 0.25μm CMOS[J]. Journal of Hunan University(Natural Sciences),2010,37(8):45—48.(In Chinese)
[8]? ? WU H J,CHEN Z H,YU Z G,et al. A 12-bit 250MSPS pipeline ADC with 4Gbps serial output interface[C]//2016 13th IEEE International Conference on Solid-State and Integrated Circuit Technology. Piscataway:IEEE,2016:1464—1466.
[9]? ? OKUNO K,OBATA K,KATO T,et al. An 800-MHz 8-bit high speed SAR ADC in 16nm FinFET process[C]//2017 IEEE International Meeting for Future of Electron Devices,Kansai. Piscataway:IEEE,2017:24—25.
[10]? LIU H Z,LIU M L,ZHU Z M. A 12-bit 200MS/s pipelined-SAR ADC in 65-nm CMOS with 61.9 dB SNDR[C]//2019 IEEE International Conference on Electron Devices and Solid-State Circuits. Piscataway:IEEE,2019:1—2.
[11]? 于健海,尹亮. 用于微機(jī)械加速度計(jì)的亞微米工藝ADC設(shè)計(jì)[J]. 西安電子科技大學(xué)學(xué)報(bào),2019,46(3):140—147.
YU J H,YIN L. Design of an ADC with the submicron process for micromechanical accelerometers[J]. Journal of Xidian University,2019,46(3):140—147. (In Chinese)
[12]? CHO J K. A 2.24-mW,61.8-dB SNDR,20-MS/s pipelined ADC with charge-pump-based dynamic biasing for power reduction in Op AMP sharing[J]. IEEE Transactions on Circuits and Systems I:Regular Papers,2017,64(6):1368—1379.
[13]? XU W G,GAO Y F,LIU X D,et al. A 18 mW 12 bit 50MS/s SHA-less pipelined ADC[C]//2017 IEEE 12th International Conference on ASIC. Piscataway:IEEE,2017:776—779.
[14]? WANG C W,WANG X,DING Y,et al. A 14-bit 250MS/s low-power pipelined ADC with aperture error eliminating technique[C]//2018 IEEE International Symposium on Circuits and Systems. Piscataway:IEEE,2018:1—5.
[15]? 張凌東,魏敬和,陳珍海,等. 用于16位125MS_s ADC的無(wú)采樣保持運(yùn)放前端電路[J]. 中國(guó)電子科學(xué)研究院學(xué)報(bào),2018,13(3):331—334.
ZHANG L D,WEI J H,CHEN Z H,et al. Front-end circuit without sample-and-hold amplifier for 16bit 125MS/s ADC[J]. Journal of CAEIT,2018,13(3):331—334. (In Chinese)