一開始為智能手機(jī)設(shè)計(jì)的應(yīng)用程序處理器現(xiàn)在已經(jīng)被許多數(shù)字電視所采用，這已經(jīng)不是秘密；隨著智能手機(jī)應(yīng)用處理器越來越普遍且性能強(qiáng)大，原產(chǎn)自智能手機(jī)的技術(shù)也不可避免地涉足至其他不同市場領(lǐng)域，包括機(jī)器對機(jī)器通訊(M2M)、醫(yī)療甚至汽車。 隨著全世界開始把連網(wǎng)汽車形容為“四個(gè)輪子的智能手機(jī)”大肆吹捧(其實(shí)也就是為車用信息娛樂系統(tǒng)打造的調(diào)制解調(diào)器芯片與應(yīng)用處理器)，車用領(lǐng)域也為芯片供貨商帶來了獨(dú)特的挑戰(zhàn)。筆者先前有一篇文章提到了高通(Qualcomm)開始供應(yīng)車規(guī) LTE 芯片(參考原文)，指出因?yàn)樵絹碓蕉嗥噺S商開始關(guān)注消費(fèi)性電子技術(shù)，也讓各家芯片廠商競相將主要針對消費(fèi)性電子設(shè)備開發(fā)的方案，升級成車規(guī)。

P14esmc

高通身為全球最大智能手機(jī)芯片供貨商，應(yīng)該是最適合回答我稱之為“車規(guī)難題”的人選；為何說是“難題”？因?yàn)橐獙⒃緸橹悄苁謾C(jī)所設(shè)計(jì)的技術(shù)移植到汽車 應(yīng)用，確實(shí)是很大的挑戰(zhàn)，特別對那些沒有車規(guī)芯片設(shè)計(jì)經(jīng)驗(yàn)的公司(包括高通)來說。這家智能手機(jī)芯片大廠想進(jìn)軍汽車應(yīng)用，需要先克服三個(gè)挑戰(zhàn)： 1. 確保自家芯片的可靠性(符合汽車應(yīng)用需求)； 2. 采用先進(jìn)(且具備某種程度的成熟度)技術(shù)； 3. 為汽車業(yè)者尋找最經(jīng)濟(jì)的解決方案。 本文授權(quán)編譯自EE Times，版權(quán)所有，謝絕轉(zhuǎn)載 第2頁：高通的“雙管齊下”策略 第3頁：車規(guī)LTE芯片驗(yàn)證難題

相關(guān)閱讀：
• 汽車人機(jī)界面（HMI）迎來數(shù)字化浪潮
• 車載以太網(wǎng)推動下一代互聯(lián)汽車新浪潮
• 汽車電子2014進(jìn)入智能變革時(shí)代P14esmc

{pagination} 高通的“雙管齊下”策略 高通產(chǎn)品管理副總裁Nakul Duggal稍早前接受 EE Times美國版編輯電話采訪時(shí)強(qiáng)調(diào)，該公司正針對汽車應(yīng)用市場采取“雙管齊下”的策略，一邊進(jìn)行車規(guī)驍龍(Snapdragon)應(yīng)用處理器的設(shè)計(jì)，一邊與第三方合作伙伴連手打造采用高通消費(fèi)性調(diào)制解調(diào)器芯片的車規(guī)等級車用信息娛樂系統(tǒng)(telematic)模塊。 針對進(jìn)軍汽車中控臺(也就是車用信息娛樂系統(tǒng)應(yīng)用)的芯片，高通致力于將自己的地位由第三線汽車零組件供貨商，升級為第二線。在1月的國際消費(fèi)性電子展 (CES)上，高通宣布將在今年底推出該公司首款車規(guī) Snapdragon 應(yīng)用處理器，且將進(jìn)駐2016年上市的新款汽車。 Duggal表示，這些日子以來，市面上的應(yīng)用處理器在“面積僅10x10mm大小的單芯片上整合了大量技術(shù)”；為了成為第二線的汽車應(yīng)用處理器供貨商，高通的任務(wù)重點(diǎn)是控管每一個(gè)進(jìn)駐該單芯片的技術(shù)，確保各個(gè)功能區(qū)塊是可靠的、且符合車規(guī)。 應(yīng) 用處理器設(shè)計(jì)業(yè)者已經(jīng)采用先進(jìn)制程節(jié)點(diǎn)，因此芯片的功耗較低；此外這些業(yè)者也尋求采用先進(jìn)內(nèi)存技術(shù)，以支持更高帶寬、更高速度以及更高組件性能。但 Duggal指出：“以 LPDDR4 為例，這種最先進(jìn)的移動內(nèi)存技術(shù)在智能手機(jī)上也是剛剛才開始應(yīng)用?！彼訪PDDR4 能用在今日的車規(guī)應(yīng)用處理器嗎？他的回答是“還不可行”。 智能手機(jī)應(yīng)用處理器的生命周期比汽車短得多，并以非常先進(jìn)的技術(shù) 為基礎(chǔ)茁壯成長；要將之移植到汽車上，高通需要衡量可靠性因素。汽車產(chǎn)業(yè)希望盡可能跟上智能手機(jī)內(nèi)的先進(jìn)技術(shù)演進(jìn)步伐，但包括高通在內(nèi)的芯片業(yè)者要確認(rèn) 的是，應(yīng)用在智能手機(jī)內(nèi)的最先進(jìn)、最強(qiáng)大技術(shù)，是否在良率可靠度上“夠成熟”，可做為車規(guī)解決方案。 至于車用信息娛樂系統(tǒng)應(yīng)用的調(diào)制解調(diào)器芯片，Duggal指出，模塊化解決方案是一條可行之路；他坦承：“我們已經(jīng)在內(nèi)部廣泛討論過這個(gè)問題。”也就是到底高通該不該自己設(shè)計(jì)車規(guī)LTE芯片，而不是仰賴第三方合作伙伴將高通的芯片應(yīng)用包模塊里、再取得車規(guī)。 本文授權(quán)編譯自EE Times，版權(quán)所有，謝絕轉(zhuǎn)載 第3頁：車規(guī)LTE芯片驗(yàn)證難題

相關(guān)閱讀：
• 汽車人機(jī)界面（HMI）迎來數(shù)字化浪潮
• 車載以太網(wǎng)推動下一代互聯(lián)汽車新浪潮
• 汽車電子2014進(jìn)入智能變革時(shí)代P14esmc

{pagination} 車規(guī)LTE芯片驗(yàn)證難題 要打造車規(guī)LTE芯片(不是模塊)，會面臨兩個(gè)問題：第一，調(diào)制解調(diào)器芯片必須能支持復(fù)雜的品牌以及蜂巢式網(wǎng)絡(luò)的持續(xù)變動，以因應(yīng)全球電信營運(yùn)商的需 求；Duggal表示：“六或七年前，只有CDMA與GSM網(wǎng)絡(luò)，我們只需要支持兩種或三種不同頻段；當(dāng)全世界進(jìn)入 3G 時(shí)代，我們需要處理6~10個(gè)頻段。”而演進(jìn)至今日的4G LTE (還需要考慮向后兼容)網(wǎng)絡(luò)：“我們正在討論的是對51個(gè)頻段的支持。” 另一個(gè)問題是“網(wǎng)絡(luò)驗(yàn)證”，調(diào)制解調(diào)器芯片除了要通過車規(guī)認(rèn)證，還需要接受各家電信營運(yùn)商所設(shè)定的嚴(yán)苛驗(yàn)證程序。因此假設(shè)你已經(jīng)有一款車規(guī)調(diào)制解調(diào)器芯片，你該如何確保已經(jīng)內(nèi)建該款車規(guī)芯片的車輛通過某家電信營運(yùn)商的網(wǎng)絡(luò)驗(yàn)證測試？是把整輛車子都送交營運(yùn)商嗎？顯然這是一個(gè)比營運(yùn)商對驗(yàn)證手機(jī)的要求，更具困難度的后勤支持能力考驗(yàn)。 比較合乎邏輯的是打造一個(gè)調(diào)制解調(diào)器模塊──內(nèi)含基頻芯片、電源管理芯片、射頻芯片、內(nèi)存、天線等所有零件──然后讓該模塊取得車規(guī)并接受電信營運(yùn)商驗(yàn)證；不 過這樣高通仍是第三線汽車零組件供貨商，而且必須仰仗模塊供貨商如LG、Sierra Wireless與Gemalto等，接受繁重的蜂巢式網(wǎng)絡(luò)驗(yàn)證與車規(guī)認(rèn)證。Duggal表示，讓一個(gè)調(diào)制解調(diào)器模塊和營運(yùn)商網(wǎng)絡(luò)“與時(shí)俱進(jìn)”的最佳方式，仍是依賴模塊產(chǎn)業(yè)生態(tài)系統(tǒng)，才能確保長期可靠性。 為了成為二線車規(guī)芯片供貨商，高通一直很努力，其工作包括為車規(guī)芯片選擇合適 的制程技術(shù)；Duggal指出：“這是一個(gè)非常嚴(yán)謹(jǐn)?shù)男袆??！辈贿^高通并不是把同樣的Snapdragon應(yīng)用處理器做兩種設(shè)計(jì)，一種給智能手機(jī)用、另 一種給汽車用；“因?yàn)槲覀兊南M(fèi)性電子等級Snapdragon系列應(yīng)用處理器產(chǎn)品種類很多，遠(yuǎn)超出汽車廠商所需要的?！? Duggal進(jìn)一步表示：“我們的任務(wù)是確認(rèn)汽車產(chǎn)業(yè)需求，關(guān)注能符合車規(guī)的技術(shù)性能，并針對能運(yùn)用汽車市場解決方案做適當(dāng)?shù)木駬??！钡罴值牟糠秩允窃趯ο冗M(jìn)技術(shù)的強(qiáng)烈需求，以及汽車市場嚴(yán)苛的可靠性規(guī)格之間取得平衡點(diǎn)。 本文授權(quán)編譯自EE Times，版權(quán)所有，謝絕轉(zhuǎn)載 編譯：Judith Cheng 參考英文原文：Qualcomm Answers 'Auto-Grade Dilemma'，by Junko Yoshida

相關(guān)閱讀：
• 汽車人機(jī)界面（HMI）迎來數(shù)字化浪潮
• 車載以太網(wǎng)推動下一代互聯(lián)汽車新浪潮
• 汽車電子2014進(jìn)入智能變革時(shí)代P14esmc

{pagination} Qualcomm Answers 'Auto-Grade Dilemma' Junko Yoshida, Chief International Correspondent MADISION, Wis. — It's no secret that apps processors originally designed for smartphones are now driving many digital TV sets. As smartphone apps processors get more widespread and powerful, technologies native to smartphones are spilling inevitably into different segments of the market -- from Machine-to-Machine (M2M) to the medical and even automotive fields. As much as the whole world likes to tout the connected car as analogous to smartphones (a modem chip for telematics and an apps processor for infotainment), automotive presents unique challenges to chip suppliers. I recently wrote a story entitled Qualcomm LTE IC Makes (Auto-) Grade, But Only in Module. The story was about a growing trend among car OEMs embracing consumer technologies. Its essence was to investigate how various chip vendors are scrambling to get their predominantly consumer technology solutions "automotive qualified." Qualcomm, the world's largest supplier of smartphone chips, is ideally positioned to answer what I call the "auto-grade dilemma." It's a dilemma because moving technologies originally designed for smartphones to automotive use is a big challenge -- especially for a company (Qualcomm) without background designing auto-grade chips from the ground up. The shift requires the smartphone chip giant to juggle three competing needs: 1) ensuring the reliability of its chips (for automotive use); 2) using advanced (and yet somewhat mature) technology; 3) finding the most economical solutions for car OEMs. Two-pronged approach In a phone interview with EE Times last week, Nakul Duggal, vice president of product management for Qualcomm Technologies, Inc., stressed that the company is taking a two-pronged approach to automotive: designing automotive-qualified chips for Snapdragon applications processors for cars; and working with third-party vendors to build auto-grade telematics modules that incorporate Qualcomm's consumer modem chips. For chips that go inside the center stack of a car (for infotainment applications), Qualcomm is committed to upgrading the company's status from a Tier-3 to Tier-2 supplier of auto-qualified parts. As announced at the International CES last month, Qualcomm is rolling out the company's first auto-grade Snapdragon apps processor by the end of this year, which will go inside 2016 model cars. Apps processors these days integrate "a lot of technologies on the same die as small as 10x10 mm," explained Duggal. To be a Tier-2 company for apps processor is important so that Qualcomm can control every technology crammed into the SoC, making sure each block is reliable and automotive qualified. Apps processor designers leverage the advanced process node so that their chips consume less power. They also look to use advanced memory technologies, to enable higher bandwidth, faster speed and higher performance for their devices. "But take an example of LPDDR4," said Duggal. The low-power double-data rate 4 (LPDDR4) is the cutting-edge mobile DRAM technology about to surface for smartphones. Could LPDDR4 be used in auto-qualified apps processors today? Duggal answered, "It's not feasible yet." Apps processors for smartphone, whose life cycle is much shorter than cars, thrive on building on a very advanced technology. But to spin that to cars, Qualcomm needs to weigh the reliability factor. The automotive industry hopes to stay as close as possible to the pace of advanced technologies used inside smartphones. But chip vendors like Qualcomm must figure out if the latest and greatest technologies used in smartphones are "mature enough" to yield reliability for automotive qualified solutions. In contrast, for modem chip sets used in telematics, the module-based approach is the way to go, according to Duggal. He acknowledged, "We've had extensive discussions on this internally." The question was whether Qualcomm should design an auto-grade LTE chip set, instead of relying on third-party vendors to package it in a module and get that module auto-grade certified. Submit a car for network certification? There are two problems with making an auto-grade LTE chip set (rather than modules). One, a modem chip set needs to support the complexity of bands and constant changes in cellular networks, in order to respond to operators' requirements -- on the global market. "Six or seven years ago, there were only CDMA and GSM. We only had to support two or three different frequency bands," said Duggal. "When the entire world moved to 3G, we need to handle six to 10 bands." But today with 4G LTE (which needs to be backward compatible), "We are talking about supporting 51 bands." Another problem is "network certification." The chip set needs to be qualified not just to be "auto-grade," but it needs to go through a rigorous certification process set by a cellular network operator. Assume you already have an auto-grade modem chip set. But how would you make sure a car integrated with that particular auto-grade chip set will pass a given cellular operator's network certification test? Submit a whole car to network operators? Obviously, this is an approach that presents far more logistical challenges than what’s demanded by operators for handsets. More logical is to build a modem module -- which includes everything from baseband, power management ICs to RF, memories, and antenna -- and make that module auto-grade. Then, have the module certified by network operators. Under this scenario, Qualcomm remains a Tier-3, and the company must rely on such module vendors as LG, Sierra Wireless and Gemalto to do the heavy lifting required for cellular network and automotive certification. Still, the best way to keep a modem module "very current with an operators' network is to depend on "the module eco-system for the long-term reliability," said Duggal. In its ambition to become a Tier-2 supplier for auto-grade chip sets, Qualcomm has been working hard. The task includes choosing the right process technology for automotive qualified chips. "This is a very serious initiative," said Duggal. Qualcomm, however, is not designing the same Snapdragon apps processor twice -- one for smartphones and another for cars. "Because we have a lot of Snapdragon apps processors in our CE portfolio -- more than what automotive companies need," Duggal said, “our job is to identify the automotive industry's needs, look at technology capabilities that can meet auto-grade, and make appropriate decisions on solutions that can be taken to the automotive market." The tricky part, however, remains the balancing act between the urge to use advanced technology and the discipline to go after the rigorous reliability demanded by automotive market.