手机一天一充,它一个月不充:Kindle最伟大的设计,从来不是电子书,而是那块几乎不耗电的屏幕Phone charged daily, it lasts a month without charge: Kindle's greatest design is not the e-book, but the screen that consumes almost no power.


AI free token/AI免费积分

 

图片

屏幕不耗电,才是这台设备最重要的设计决定

上一篇我们聊了Pebble,讲到它用e-paper+FreeRTOS+Cortex-M的组合,把智能手表的续航做到了5到7天。

有个读者留言说:"这套架构我看懂了,但我没想到一个读者设备比一块手表早六年就把同样的事情想清楚了。"

他说的是Kindle。

我自己用过三代Kindle,前两代都是充一次电用三四个星期。有段时间出差频繁,Kindle和手机一起丢包里,手机要每天充,Kindle忘了充也不怕——因为我压根记不清楚上次充是什么时候。

这件事背后的工程逻辑,值得认真说一遍。


一、1993年,一个物理学家在斯坦福想到了一件事

Joseph Jacobson是MIT媒体实验室的教授,物理学博士,做过世界上最短脉冲激光的纪录。

但让他名留青史的,不是激光,是1993年在斯坦福做博士后时脑子里冒出来的一个问题:

能不能造一种屏幕,不需要通电就能显示内容?

他的出发点很朴素:书不需要电,报纸不需要电,白板写上字不需要电——为什么显示文字的电子设备一定要持续耗电才能保持画面?

1997年,他和MIT两名本科生——JD Albert和Barrett Comiskey——联合创立了E Ink Corporation。专利的核心思路是:把带电的黑色和白色颜料微粒,密封在几十微米大小的胶囊里,胶囊里充满透明液体;施加电场,黑粒子往上走、白粒子往下走,显示黑色;翻转电场,反过来,显示白色。

关键在于:颗粒到达之后,靠范德华力和表面吸附力自然停住,不需要继续通电维持。

这叫"双稳态"——两种状态都是稳态,切换需要能量,维持不需要。

这个特性,从物理层面把"显示设备必须持续耗电"这个假设直接推翻了。


二、2004年,Bezos对Steve Kessel说了一句话

2004年,亚马逊高管Steve Kessel被叫进Bezos的办公室,被委以一项新任务。

Kessel问:这个项目什么时候要完成?

Bezos说:"你已经迟了。"

这就是Kindle项目的起点。Bezos担心的不是有人比亚马逊更会卖书,而是有人会造出一台设备,让用户完全不需要亚马逊就能读书——就像iTunes让唱片公司集体哑火一样。

他的判断是:如果数字阅读这件事终究会发生,最好是亚马逊来做,而不是等别人来做完再后悔。

Kessel组建了一个代号LAB126的秘密项目组,总部在硅谷,对外不挂亚马逊的牌子。负责硬件的是Gregg Zehr,Palm Computing前硬件工程副总裁。整个项目代号叫"Fiona"——取自科幻小说《钻石年代》里的一本神奇互动书,作者Neal Stephenson后来和Bezos成了朋友。

LAB126研究了当时所有现存的电子书阅读器——1998年的Rocket eBook、索尼2004年的Librie——然后得出结论:这个市场是空的,还没有人做对过。

他们在显示方案上的选择只有一个:E Ink。


三、架构决定一切

2007年11月19日,Kindle第一代上市,399美元,5.5小时售罄,随后断货整整五个月。

这台设备的硬件规格放在今天看,普通得令人惊讶:

组件
规格
设计意图
处理器
532MHz ARM9
刚好够用,不追求性能
存储
256MB内部 + SD卡扩展
够装200本书
显示
6英寸E Ink四级灰度
双稳态,刷新耗电,保持不耗电
网络
Sprint EVDO 3G(Whispernet)
亚马逊包月费,用户零费用
续航
电池两周(3G常开约一周)
由E Ink双稳态决定

这张表里最重要的,是显示屏那一行和续航那一行之间的因果关系。

E Ink屏在静止时的功耗,接近于零。用电的只有两件事:一是翻页时刷新屏幕的那一瞬间(典型值约100-150mW,持续约120-200毫秒);二是3G模块处于激活状态时的无线通信。

这意味着整台设备的能量管理策略可以非常简单:屏幕不刷新、网络不激活时,让处理器深度休眠,系统趋向零功耗。读者翻页的频率,大约每分钟不超过一两次,每次刷新持续不到0.2秒。剩下的99%以上时间,设备在深睡。

这和Pebble的策略完全一致,只是早了六年,用在了读书而不是看时间这件事上。


四、Whispernet——那个最被低估的工程决定

所有Kindle的技术选择里,最被低估的,不是E Ink屏,是Whispernet。

Whispernet是亚马逊给这套3G连接起的内部名字。Kindle内置Sprint EVDO模块,用户不需要购买数据套餐,不需要按月付费,下书是免费的——亚马逊把运营商费用打包进了硬件售价和书价里,完全吸收。

这个决定有多疯?Bezos最初不同意上3G,是Gregg Zehr一遍遍说服他的,理由是:如果用户每次要下书都得先找WiFi,买书这件事的摩擦力就还在;Whispernet让"想到就买"变成了现实。

从嵌入式角度看,这个选择带来了一个电源管理挑战:3G模块是设备上最大的功耗单体,激活时远超E Ink刷新的功耗。Kindle的解法是:3G默认处于低功耗待机态,只在用户主动打开商店或同步内容时才全速激活,完成后立即退回待机。

亚马逊最终承担了巨大的无线通信运营成本,换来了书城不断货、用户不想充值、买书不卡壳——这是商业模型和硬件架构深度绑定的一个罕见样本。


五、E Ink演化:从四级灰度到彩色

世代
年份
关键升级
刷新改善
第一代E Ink
2007
四级灰度,200ms全刷
翻页有明显白闪
E Ink Pearl
2010
对比度提升50%,灰阶更多
白闪有所减弱
E Ink Carta
2013
响应速度提升,分辨率至300ppi
局部刷新实用化
E Ink Carta 1200
2021
更高对比度,进一步提速
连续翻页白闪极小
E Ink Kaleido / Gallery
2020起
彩色方案,4096色
彩色刷新比灰度稍慢
E Ink Gallery 3
2023
彩色刷新速度大幅提升
支持低帧率动画

E Ink最顽固的用户体验问题从第一天就存在:翻页时那一下白闪。

这是电泳显示的物理本质——黑白颗粒换位需要时间,全刷时必须先让所有颗粒"清零"到白色,再重新排列。Carta之后的局部刷新(Partial Refresh)技术,允许只刷新内容发生变化的区域,大幅减少可见白闪——但代价是局部刷新积累的残影需要定期全刷清除,否则屏幕上会留下前几页的鬼影。

这个残影清除策略,是所有用E Ink做UI的嵌入式工程师都必须手动处理的问题。


六、从业者坑位:用E Ink做嵌入式显示的那些特有麻烦

上一篇Pebble写了e-paper选型的大方向,这里专门写E Ink驱动层的具体坑——做过的朋友大概对某几行有专属记忆:

现象
实际原因
应对思路
局部刷新几次后,屏幕上有前页残影
局部刷新不完全清除颗粒残留,累积到一定程度形成鬼影
计数局部刷新次数,每隔N次(典型值5-10次)强制全刷一次清除残影;N值要根据实测效果调,不能拍脑袋
低温环境下翻页明显变慢,有时刷新不完整
电泳液黏度随温度升高,低温下颗粒迁移速度下降,默认刷新波形不够驱动
E Ink控制器支持温度自适应波形(Temperature Compensation),必须读取温度传感器并动态切换波形表,不能用固定波形
相同内容连续显示数天后,出现永久残影(烧屏)
长期施加单向电场导致颗粒极化,形成不可逆物理变化
定时(如每24小时)执行全白→全黑→全白的"刷新操作";不能让E Ink长期保持同一静态画面不动
换E Ink模组供应商后,波形驱动代码失效
不同品牌/型号的E Ink模组使用不同的波形文件(Waveform),波形不兼容
E Ink波形文件是厂商私有的,换模组必须同步更换驱动波形,不能复用;立项时锁定供应商型号
快速翻页时屏幕发热明显
高频刷新时E Ink驱动IC(如IT8951)持续工作,加上频繁的TFT电极切换,整体功耗和热量远超静态显示
翻页动画或视频类应用不适合E Ink;如果必须做,要做帧率限制和主动散热评估

最后这条让我想起一个项目。客户要在一块E Ink屏上做一个"流动水波"的背景动画,以为这会显得高端。我们在原型阶段演示了以后,他们的产品经理马上改主意了——不是因为技术做不到,是因为E Ink屏在高频刷新下,那个"科技感"直接变成了"显眼包"式的不停闪烁。

这件事让我后来养成了一个习惯:E Ink选型会议上,先放一段实机的高频刷新视频给所有决策者看,把预期对齐在动工之前。


七、2009年,亚马逊删掉了《一九八四》

Kindle历史上有个细节,不提不完整。

2009年7月,亚马逊从数千台Kindle上远程删除了两本书:乔治·奥威尔的《一九八四》和《动物农场》。

原因是:销售这两本书的出版商,其实没有合法的电子书发行权。亚马逊发现之后,直接向所有用户的设备发送了删除指令。

用户早上起来,发现购买的书从Kindle里消失了。

在一本讲极权政府操控现实的书上,亚马逊做了一个"远程删除"的操作。这个反讽被全球媒体反复引用,Bezos后来公开道歉,说这件事"是愚蠢的",并承诺修改系统策略。

这件事暴露了Kindle架构里一个从一开始就存在的设计决定——Whispernet不只是下行的,它也是双向的。亚马逊可以推送内容,也可以撤回内容。这是商业便利和数字所有权之间的根本张力,到今天也没有消失。

诚实地说,Kindle卖给你的不完全是"一本书",是"一个访问许可"——这件事在2009年第一次被所有人看清楚。


八、iPad来了,Kindle还活着

2010年,Steve Jobs发布iPad,发布会上说了一句话:"Amazon has done a great job pioneering the e-reader. We're going to stand on their shoulders and go further."

很多人以为Kindle要死。

Bezos的反应是:同年推出Kindle 3(Kindle Keyboard),价格从399美元降到139美元,然后79美元。同时开发了Kindle Fire——一台彩色触摸平板,主动参与iPad的战场。

但Kindle电子书阅读器这条线,没有死。

到2025年,Kindle仍然是全球销量第一的电子书阅读器,出货量遥遥领先。原因很简单——它做的事情,iPad做不好:连续读书两周不充电,阳光直射下清晰可见,没有微信消息弹出来打扰你。

这三件事,没有一件需要OLED屏幕,没有一件需要8核处理器。它们全部由E Ink的双稳态特性和深度睡眠架构免费提供。

Jobs那句"we'll go further"说的是功能的广度——他是对的。Bezos要的是阅读体验的深度——他也是对的。两种对的答案,能同时在市场上活着,是因为他们回答的是不同的问题。

这一点,和Pebble vs Apple Watch的故事惊人地相似。


写在最后

Kindle、Pebble,一台读书的设备,一块看时间的手表,相差六年,相差万里,但工程内核是同一件事:

先选定屏幕技术,再让整个系统架构围着它转。

E Ink的双稳态,免费给了Kindle两周续航,免费给了Pebble五天续航。那根ARM9在532MHz上跑着Linux,Cortex-M在Pebble里跑着FreeRTOS——处理器不一样,OS不一样,产品形态不一样,但在"屏幕不需要持续供电"这件事上,两个团队做出了完全一样的判断。

这个判断,来自MIT媒体实验室的一间实验室,来自1997年1月一个寒冷的早晨,Barrett Comiskey在显微镜下看到一颗颗料在电场里移动,然后停住——不需要电,就停在那里。

他们看到的,是一种新的物理可能。

Bezos和Migicovsky,用各自的方式,把那个可能变成了产品。

时间线
事件
1993
Joseph Jacobson在斯坦福构想"不耗电的显示器"
1997年1月
Barrett Comiskey在显微镜下完成首次颗粒迁移演示
1997
E Ink Corporation成立,MIT Media Lab孵化
2004
Bezos启动"Fiona"项目,LAB126成立;Sony Librie发布(首款E Ink阅读器)
2007年11月19日
Kindle第一代上市,399美元,5.5小时售罄
2009
Kindle 2发布;同年亚马逊远程删除《一九八四》,舆论哗然
2010年4月
iPad发布;E Ink Pearl技术量产,对比度提升50%
2010年8月
Kindle 3(Kindle Keyboard)价格降至139美元
2012
Kindle Paperwhite第一代,带前光,300ppi高分屏
2013
E Ink Carta发布,局部刷新实用化
2021
Kindle Paperwhite 5,6.8英寸,USB-C充电
2023
Kindle Scribe发布,10.2英寸大屏+手写笔
2025
Kindle仍为全球销量第一电子书阅读器

致敬,Joseph Jacobson、JD Albert和Barrett Comiskey——三个人在MIT的实验室里问了一个简单的问题,花了四年把答案变成一家公司,然后又等了十年,等到亚马逊把这个答案变成了全球人手一台的阅读设备。从物理直觉到商业产品,中间这段路,比任何人预想的都更漫长。

也致敬所有在E Ink驱动层挣扎过的嵌入式工程师们——波形文件、温度补偿、残影清除、局部刷新计数——这些细节没有人会写在产品手册的显眼位置,但你们知道它们在哪里,也知道跳过它们会发生什么。正是这些默默处理掉的细节,撑起了那个"两周不充电"的承诺。

关注「奔跑的码仔」,我们下篇继续硬件产品系列。

 

 

 

图片

The screen consuming no power is the most important design decision of this device.

In the previous article, we talked about Pebble and how it used the combination of e-paper + FreeRTOS + Cortex-M to give a smartwatch a battery life of 5 to 7 days.

A reader commented: "I understand the architecture, but I never expected a reading device to figure out the same thing six years before a watch did."

They were talking about the Kindle.

I have used three generations of Kindles myself. With the first two, a single charge lasted three to four weeks. There was a period when I traveled frequently, and the Kindle was tossed into my bag along with my phone. The phone needed charging every day, but the Kindle was fine even if I forgot to charge it—because I honestly could not remember when I last charged it.

The engineering logic behind this deserves a proper explanation.

1. In 1993, a physicist at Stanford thought of something.

Joseph Jacobson is a professor at the MIT Media Lab, holds a Ph.D. in physics, and once held the record for the world's shortest pulsed laser.

But what made him famous was not the laser. It was a question that came into his head in 1993 while he was a postdoc at Stanford:

Could we build a screen that can display content without needing power?

His starting point was simple: books do not need electricity, newspapers do not need electricity, and a whiteboard with writing does not need electricity—so why must an electronic device that displays text constantly consume power just to hold an image?

In 1997, he co-founded E Ink Corporation with two MIT undergraduates, JD Albert and Barrett Comiskey. The core idea of the patent was to seal charged black and white pigment particles inside micrometer-sized capsules filled with transparent liquid. When an electric field is applied, the black particles move up and the white particles move down, displaying black. When the field is reversed, the opposite happens and white is displayed.

The key point: after the particles arrive, they naturally stay in place thanks to van der Waals forces and surface adhesion, so no continuous power is needed to hold them there.

This is called "bistability"—both states are stable states. Switching requires energy, but maintaining does not.

This property, at the physical level, directly overturned the assumption that a display device must continuously consume power.

2. In 2004, Bezos said something to Steve Kessel.

In 2004, Amazon executive Steve Kessel was called into Bezos's office and given a new assignment.

Kessel asked: When does this project need to be done?

Bezos said: "You are already late."

That was the starting point of the Kindle project. Bezos was not worried that someone else would sell books better than Amazon. He was worried that someone would build a device that would let users read without needing Amazon at all—just as iTunes had silenced the record labels.

His judgment was: if digital reading was going to happen anyway, it was better for Amazon to do it than to regret it after someone else did.

Kessel formed a secret project group codenamed LAB126, headquartered in Silicon Valley and not bearing the Amazon name publicly. Hardware was led by Gregg Zehr, former vice president of hardware engineering at Palm Computing. The project's overall codename was "Fiona"—taken from a magical interactive book in Neal Stephenson's novel The Diamond Age. Stephenson later became friends with Bezos.

LAB126 studied all the existing e-readers at the time—the Rocket eBook from 1998 and Sony's Librie from 2004—and concluded that the market was empty and nobody had gotten it right yet.

Their choice in display technology had only one option: E Ink.

3. Architecture decides everything.

On November 19, 2007, the first-generation Kindle launched at $399 and sold out in 5.5 hours, then stayed out of stock for five full months.

The hardware specs of that device, viewed today, are surprisingly ordinary:

Component

Spec

Design intent

Processor

532MHz ARM9

Just enough; no pursuit of performance

Storage

256MB internal + SD card expansion

Enough for 200 books

Display

6-inch E Ink, 4-level grayscale

Bistable; refresh consumes power, holding does not

Network

Sprint EVDO 3G (Whispernet)

Amazon paid the monthly fee; zero cost to users

Battery life

Two weeks on a battery (about one week with 3G always on)

Determined by E Ink bistability

The most important thing in this table is the causal relationship between the display row and the battery-life row.

The power consumption of an E Ink screen at rest is close to zero. The only two things that use power are: first, the brief moment of refreshing the screen when turning a page (typically around 100–150mW for about 120–200 milliseconds), and second, wireless communication while the 3G module is active.

This means the entire device's power management strategy could be very simple: when the screen is not refreshing and the network is not active, put the processor into deep sleep and let the system approach zero power. Readers turn pages at most once or twice per minute, and each refresh lasts less than 0.2 seconds. More than 99% of the remaining time, the device is in deep sleep.

This is exactly the same strategy Pebble used—just six years earlier and applied to reading rather than telling time.

4. Whispernet—the most underestimated engineering decision.

Among all of Kindle's technology choices, the most underestimated one is not the E Ink screen but Whispernet.

Whispernet was Amazon's internal name for its 3G connectivity system. The Kindle had a built-in Sprint EVDO module. Users did not need to buy a data plan or pay a monthly fee. Downloading books was free—Amazon folded the carrier costs into the hardware price and book prices and absorbed them completely.

How crazy was this decision? Bezos initially opposed adding 3G. Gregg Zehr had to convince him again and again. His argument was: if users had to find Wi-Fi every time they wanted to download a book, the friction in buying books would still exist. Whispernet turned "buy when you think of it" into reality.

From an embedded-systems perspective, this choice created a power-management challenge: the 3G module was the single largest power consumer in the device, far exceeding E Ink refresh power when active. Kindle's solution was: 3G stayed in a low-power standby state by default, activated at full speed only when the user actively opened the store or synced content, then immediately returned to standby.

Amazon ultimately absorbed enormous wireless communication operating costs in exchange for a bookstore that never went out of stock, users who never had to top up, and purchases that never stalled—a rare example of deep integration between a business model and hardware architecture.

5. E Ink evolution: from 4-level grayscale to color.

Generation

Year

Key upgrade

Refresh improvement

First-generation E Ink

2007

4-level grayscale, 200ms full refresh

Obvious white flash when turning pages

E Ink Pearl

2010

50% higher contrast, more grayscale levels

Reduced white flash

E Ink Carta

2013

Faster response, resolution up to 300ppi

Partial refresh became practical

E Ink Carta 1200

2021

Higher contrast and further speed increase

Minimal white flash during continuous page turns

E Ink Kaleido / Gallery

From 2020

Color solutions, 4096 colors

Color refresh is slightly slower than grayscale

E Ink Gallery 3

2023

Color refresh speed greatly improved

Supports low-frame-rate animation

E Ink's most stubborn user-experience problem has existed from day one: the white flash when turning a page.

This is the physical nature of electrophoretic displays—black and white particles need time to switch positions, and a full refresh requires first "clearing" all particles to white before rearranging them. The partial refresh technology introduced after Carta allows only the region where content changes to be refreshed, greatly reducing visible white flash—but at the cost of accumulated ghosting from partial refreshes, which requires periodic full refreshes to clear, otherwise ghost images from previous pages remain on screen.

This ghost-clearing strategy is a problem that every embedded engineer using E Ink for UI must handle manually.

6. Engineer pitfalls: the specific troubles of using E Ink for embedded displays.

In the previous Pebble article, I covered the broad direction of e-paper selection. Here I will focus on the specific pitfalls at the E Ink driver layer—those who have done this work probably have private memories of certain rows:

Phenomenon

Actual cause

Countermeasure

After several partial refreshes, residue from previous pages appears on screen

Partial refresh does not completely clear pigment residue; it accumulates and forms ghosts after a certain threshold

Count partial refreshes and force a full refresh every N times (typically 5–10) to clear ghosts; N must be tuned based on measured results, not guessed

Page turns slow noticeably in low temperatures, sometimes with incomplete refreshes

Electrophoretic fluid viscosity changes with temperature; at low temperatures, particle migration slows and the default refresh waveform is insufficient

E Ink controllers support temperature-adaptive waveforms (temperature compensation); you must read the temperature sensor and dynamically switch waveform tables; fixed waveforms cannot be used

After displaying the same content continuously for days, permanent ghosting appears (burn-in)

Long-term application of a unidirectional electric field polarizes the particles, causing irreversible physical changes

Periodically (e.g., every 24 hours) perform a white→black→white "refresh operation"; never leave E Ink on the same static image for too long

After switching E Ink module suppliers, the waveform driver code stops working

Different brands/models of E Ink modules use different waveform files; waveforms are incompatible

E Ink waveform files are vendor-proprietary; when changing modules, the driver waveform must be replaced accordingly and cannot be reused; lock in the supplier and model at project kickoff

Screen heats up noticeably during rapid page turning

During high-frequency refresh, the E Ink driver IC (e.g., IT8951) works continuously, plus frequent TFT electrode switching, making overall power and heat far exceed static display

Page-turn animations or video-like applications are not suitable for E Ink; if you must do them, implement frame-rate limits and proactive thermal evaluation

This last one reminds me of a project. A client wanted to put a "flowing water ripple" background animation on an E Ink screen, thinking it would look premium. After we demonstrated it in the prototype stage, their product manager changed their mind immediately—not because the technology could not do it, but because under high-frequency refresh, the "tech feel" of the E Ink screen turned directly into an attention-grabbing, nonstop flicker.

That experience gave me a habit: at E Ink selection meetings, I first show a video of actual high-frequency refresh on a real unit to all decision-makers, aligning expectations before work begins.

7. In 2009, Amazon deleted Nineteen Eighty-Four.

There is a detail in Kindle history that cannot be omitted.

In July 2009, Amazon remotely deleted two books from thousands of Kindles: George Orwell's Nineteen Eighty-Four and Animal Farm.

The reason was that the publisher selling those two books did not actually hold legitimate electronic distribution rights. After Amazon discovered this, it sent deletion commands directly to all affected users' devices.

Users woke up in the morning and found that books they had purchased had disappeared from their Kindles.

On a book about a totalitarian government manipulating reality, Amazon performed a "remote deletion." The irony was cited repeatedly by global media. Bezos later publicly apologized, called the action "stupid," and promised to change the system's policies.

This incident exposed a design decision that had existed in Kindle's architecture from the start—Whispernet was not only downstream; it was bidirectional. Amazon could push content, and it could also retract content. This is a fundamental tension between commercial convenience and digital ownership, and it has not disappeared to this day.

To be honest, what Kindle sold you was not entirely "a book"; it was "an access license." That was first made clear to everyone in 2009.

8. The iPad arrived, and Kindle is still alive.

In 2010, Steve Jobs unveiled the iPad and said at the keynote: "Amazon has done a great job pioneering the e-reader. We're going to stand on their shoulders and go further."

Many people thought the Kindle would die.

Bezos's response was to release the Kindle 3 (Kindle Keyboard) the same year, cutting the price from $399 to $139, and then to $79. At the same time, he developed the Kindle Fire—a color touchscreen tablet—to actively join the fight against the iPad.

But the Kindle e-reader product line did not die.

As of 2025, the Kindle remains the world's best-selling e-reader, with shipments far ahead of competitors. The reason is simple: it does things the iPad cannot do well—read continuously for two weeks without charging, remain clearly visible in direct sunlight, and never interrupt you with WeChat messages popping up.

None of those three things requires an OLED screen, and none requires an 8-core processor. All of them are provided for free by E Ink's bistability and deep-sleep architecture.

Jobs's "we'll go further" was about breadth of features—he was right. Bezos wanted depth in the reading experience—he was also right. The reason two right answers can coexist in the market is that they were answering different questions.

This is strikingly similar to the story of Pebble vs. Apple Watch.

Final words

Kindle and Pebble—one is a device for reading, the other a watch for telling time. They are six years apart and a world apart, but their engineering core is the same thing:

First choose the screen technology, then build the entire system architecture around it.

E Ink's bistability gave Kindle two weeks of battery life for free and gave Pebble five days of battery life for free. The ARM9 running Linux at 532MHz and the Cortex-M running FreeRTOS inside Pebble—different processors, different operating systems, different product forms, but the two teams reached exactly the same judgment on the question of "the screen does not need continuous power."

That judgment came from a lab at the MIT Media Lab, from a cold January morning in 1997 when Barrett Comiskey watched under a microscope as pigment particles moved in an electric field and then stopped—without needing power, they simply stayed there.

What they saw was a new physical possibility.

Bezos and Migicovsky, each in their own way, turned that possibility into products.

Timeline

Event

1993

Joseph Jacobson conceives of a "power-free display" at Stanford

January 1997

Barrett Comiskey completes the first demonstration of particle migration under a microscope

1997

E Ink Corporation is founded, incubated by MIT Media Lab

2004

Bezos launches Project "Fiona," LAB126 is established; Sony Librie is released (first E Ink reader)

November 19, 2007

First-generation Kindle launches at $399, sells out in 5.5 hours

2009

Kindle 2 is released; Amazon remotely deletes Nineteen Eighty-Four, causing public outrage

April 2010

iPad is released; E Ink Pearl enters mass production, contrast improves 50%

August 2010

Kindle 3 (Kindle Keyboard) price drops to $139

2012

First-generation Kindle Paperwhite, with front light and 300ppi high-resolution screen

2013

E Ink Carta is released, making partial refresh practical

2021

Kindle Paperwhite 5, 6.8 inches, USB-C charging

2023

Kindle Scribe is released, 10.2-inch large screen with stylus

2025

Kindle remains the world's best-selling e-reader

A salute to Joseph Jacobson, JD Albert, and Barrett Comiskey—three people who asked a simple question in an MIT laboratory, spent four years turning the answer into a company, and then waited another decade until Amazon turned that answer into a reading device in the hands of people around the world. The road from physical intuition to commercial product was longer than anyone expected.

A salute as well to every embedded engineer who has struggled in the E Ink driver layer—waveform files, temperature compensation, ghost clearing, partial-refresh counting. These details are not written in prominent places in product manuals, but you know where they are, and you know what happens if you skip them. It is these silently handled details that uphold the promise of "two weeks without charging."

Follow "奔跑的码仔" (The Running Coder) and we will continue the hardware product series in the next article.

暂无评论,快来发表第一条评论吧!No comments yet. Be the first!

📮 需求咨询📮 Inquiry