All you wanted to know about batteries but were afraid to ask: April 2015

Wednesday, April 29, 2015

62. YOUR WATCH BATTERY IS REALLY CHALLENGED

Apple just released its Apple Watch joining a growing list of manufacturers of new smart watches, including Samsung, LG, Motorola and others. These watches include, in addition to timekeeping, features such as messaging, basic email, navigation, in some cases voice calls, health & sports monitoring, as well as a slew of growing apps that seem dedicated for this tiny screen on the wrist. For the most part, they seem so far to be an extension of your mobile device, with the Apple Watch extending the reach of the iOS ecosystem, and the other watches performing the same for the Android ecosystem. Let's examine in today's blog the batteries used in these smart watches as well as batteries in "older and dumb" watches.

One of my favorite and practical watches is the Casio Pathfinder family. It was first introduced ca. 2001 and was the first to integrate a series of useful microsensors for the outdoorsman (or woman). Incorporating an altimeter and digital compass, it also was able to synchronize its time to NIST's universal clock broadcast radio signal out of Colorado. And it never ran out of battery juice -- it had a little lithium-ion battery made by Panasonic (CTL 1616) and was recharged by ambient sunlight. Made of a lithium cobalt titanate chemistry, this little cell had a terminal voltage of 2.3 V and a charge capacity of 18 mAh, equivalent to 0.041 Wh -- plenty to power the sensors and gray-scale LCD display.

Now, let's look at the new generation of smart watches in a family portrait of their respective batteries, courtesy of various teardowns from ifixit.

These batteries have a charge capacity that is about 15X larger than the one used in the Casio watch. This is clearly not surprising since these new smartwatches consume significantly more power to operate the AMOLED display and all the radios (e.g., WiFi, NFC, Bluetooth, and in some cases, LTE). Consequently, the batteries are physically larger and thicker, and use half or more of the smart watch volume. The CTL 1616 is a mere 1.6 mm thick whereas the cells in smart watches are twice as thick or more. Among these smart watches, the LG G Watch has the highest capacity at 400 mAh and the Apple Watch the least at 205 mAh.

But this 15X-increase in charge capacity does not yield a longer use time relative to my decade-old Casio. I never have to worry about charging my Pathfinder but these new smartwatches seem to have a run time between 3 hours (for the Apple Watch) and 4 hours (for the Android flavors) when operated constantly, with the display and radios on the entire time. Simple math gives us a quick estimate of the power usage: approximately 250 mW for the Apple Watch going up to about 350 mW for the Android watches. Naturally, aggressive power management, fancy parlance for frequently shutting down the display and the radios and CPU, preserves the battery and extends its life to an estimated full day of use. Additionally, there is probably room to bring these power consumption figures down with time as watch designs get optimized -- time will tell.

Apple does report a charge time of 2.5 hours corresponding to a rate of approximately 0.6 C, or about 120 mA of charge current into the battery. Measured charge times for the Samsung and LG watches appear to be in line, with the Moto 360 being a smidgen faster at about 2 hours (equivalent to 0.7 C or a charge current of 200 mA). This means that the charging power into the watch, whether it is wired or wireless, varies between 0.5 W and 1 W, considerably less than the charging power into a smartphone (which may reach up to 18 W).

So let's see if we can synthesize a coherent picture of the challenges that batteries in these smart watches face if this product category will become mainstream. First, it's imperative that the battery capacity is increased significantly past 300 mAh, the level that seems to be the norm for now. In other words, unless consumers want thick and large watches, the energy density of the cells will have to rise above the already-large figures in present batteries. This is not an easy task. Second, these batteries had better be thin. But thin and high energy density don't go well together. Third, I really don't like a 2-hour charge time. Consumers will want to see these watches off their wrists for no more than 15 - 30 minutes. So expect to see fast charging soon. Yes, that's doable. And lastly, time will tell how often consumers will replace their watches, but I bet that at $350 ea., they will need to last way more than 2 years...so has anyone yet screamed foul on cycle life.

Friday, April 24, 2015

61. LIAR, LIAR, PANTS ON ULTRAFIRE

Ok, ok, I should not be joking about the name. Safety is a serious matter. But let's talk honestly today about dishonesty.

Ultrafire is a China-based vendor of 18650 lithium-ion cells. One can buy their 18650 cells from eBay, Amazon and lots of other web-based stores. I was intrigued last week when I saw on eBay Ultrafire 18650 cells advertised with a charge capacity of 5,000 mAh. Wow, 5Ah! Is this real? The most I had seen ever were 3,400 mAh from Panasonic (what's inside the Tesla) and 4,000 mAh was the next promised land. But 5,000 mAh? I really wanted one.

You gotta love e-commerce. Within less than 48 hours, I was the proud owner of 4 Ultrafire 18650 cells each labeled 5,000 mAh -- exactly as shown in the photograph above. Now to put things in perspective, an 18650 cell has a volume of about 17 cc. These cells are rated at a maximum voltage of 4.2 V and an average voltage of 3.7 V. A quick calculation reveals that they would have an energy density in excess of 1,000 Wh/l. Now, that's either serious lying or some serious innovation. Let's find out.

My team was generous enough with their time to run a capacity test on these Ultrafire cells, and compare them relative to the Panasonic 3.4 Ah cells. For you tech geeks wanting the specifics, the Panasonic cells were charged and discharged at 1.7 A (0.5 C rate) whereas the Ultrafire cells had a much smaller charge/discharge current of 500 mA. The cells were all charged from a minimum of 3.0 V up to a maximum of 4.2 V with a termination current of C/20, then discharged back to 3.0 V. Temperature was maintained at 25 ºC. What did we measure?

The graph shows the standard discharge curves for both Ultrafire and Panasonic cells. Panasonic provides a capacity of 3,000 mAh vs. the advertised 3,400 mAh. In actuality, one would probably get very very close to 3,400 mAh had we gone down to 2.7 V and charged at a much slower rate of 0.2 C (instead of 0.5 C). The Panasonic cell is made with an NCA cathode which provides additional energy down to 2.7 V. So Panasonic seems quite honest with their capacity claim.

But Ultrafire is not even close...shame on you, Ultrafire! The advertised 5,000 mAh has a charge capacity of barely above 1,000 mAh. I have known that crooks are everywhere attempting everything under the sun, but for some naive reason, I had thought that lithium-ion batteries might be, just might be immune to this degree of cheating. But making a claim that is 5X reality, well, I'd better reset my expectations.

The lesson du jour: if you see Ultrafire cells, run, and run fast!

Tuesday, April 14, 2015

60. WHAT IS CELL BALANCING ?

We have covered in prior blogs the operation of batteries in smartphones. The vast majority of such devices use single-cell batteries. In other words, there is one physical cell that is the battery. As such, it has a given charge capacity measured in mAh or Coulombs, and it has a voltage range that is between 3.0 and 4.35V. If we stack multiple cells in an electrical configuration, then in principle, one can obtain a multi-cell battery configuration, called a pack, that can deliver more charge capacity.

The electrical configuration of such cells defines the nomenclature - see figure below. If the cells are electrical tied in series, then the pack is called s-configuration. If they are tied in parallel, then they are in a p-configuration. The former serves to raise the maximum voltage of the pack in multiples of 4.35V, whereas the latter serves to increase the maximum current through the pack without increasing the voltage.

Now let's examine what happens if the cells in a multi-cell pack are not identical. For example, they could be slightly different from the onset, or perhaps aged at different rates. In a parallel configuration, the voltage is always equal for both cells. Any difference in charge capacity between the cells will manifest itself as a difference in current in the two branches. In particular, this parallel configuration always guarantees that the cells do not exceed their maximum safe voltage, often 4.35V.

But a series configuration creates a different and more challenging situation. The current is shared and equal to both cells, and hence, each cell will manifest a different voltage. Let's first examine the charging of two cells in series. If the two cells are truly identical, then they will reach their maximum capacity and their maximum voltage at the same moment. But if there is a difference in capacity between them, then the cell with a smaller capacity will reach 4.35V before the other cell does. At this point time, one cell is 100% full while the other one is not. If the charging is not disconnected immediately, one cell will certainly get overcharged and cause a hazard.

To remedy this situation, electrical circuits called cell balancing are used. In principle they are simple They add a little switch and a small resistor across each cell in series. This added circuitry provides the ability to "bleed off" additional charge from the "strong" cell, so that its voltage stays about equal to that of the weak sister. This type is called "passive balancing." Naturally, this is not a very energy-efficient nor cost-efficient method, but at least it guarantees that the weak cell will not be overcharged. As we covered in prior blogs, lithium-ion cells, unlike lead-acid batteries, risk catching fire or exploding when they are overcharged above their maximum voltage, typically 4.35V for one individual cell.

Let's now examine discharging two cells in series. The figure below shows the voltage vs. charge curve for two similar but slightly different cells. They are both nominally 7,000 Coulombs (or about 1,900 mAh) but in reality, one cell is 7,200 Coulombs and the other one is 6,800 Coulombs. This is about 5% difference in capacity, and can readily happen in a pack without the proper precaution.

Let's now assume that both cells are charged to an identical voltage. For the blue cell, this will correspond to a stored electrical charge of 3,600 Coulombs, or about 100 Coulombs more than its sister cell. Let's now start discharging the cells in series; in other words, the exact same discharge current flows through both of them for exactly the same duration of time. This means that both cells will lose the same amount of charge; for the purpose of this discussion, we assume it is 3,000 Coulombs. We notice from the figure above that the blue cell will have a terminal voltage across its cells that is higher than the red cell (the more aged cell). Any further discharge will cause the red cell to drop precipitously and cause it further degradation, effectively over discharging the cell. This is not an unsafe event but it is a phenomenon where the weak cell (the red cell) will actually degrade at a faster rate in a series configuration. This is why it is always said that a "pack is only as good as its weakest cell." In other words, without the use of clever algorithms and balancing, the cycle life of the entire pack will be equal to the cycle life of its weakest cell.

Battery-pack manufacturers try to minimize this problem by matching the cells in a pack as much as possible. It is very common for pack manufacturers (including makers of electric vehicles) to measure the capacity of each and every cell in a pack, and matching the cells to within less than 1% in charge capacity. But as one will immediately observe, this gets very expensive especially for large packs as the yield of useable cells can be quite low.

In some extreme cases, some packs can utilize "active balancing." This includes more sophisticated electronic circuits that will actually shuffle charge from the strong cell to the weak cell. The effect is to increase the cycle life of the pack by shoring up this weak cell and ensuring that it does not get overcharged nor over-discharged.

It is important to close here by saying that the vast majority of mobile devices use single cell configuration, and hence do not implement cell balancing. Most laptop computers and some tablets use 2S configurations (two cells in series). They often implement rudimentary passive balancing. For example, the Apple MacBook series of products often use the bq20zxx family of fuel gauges with integrated cell balancing from Texas Instruments -- such consumer-grade fuel gauges can handle cell balancing for small packs up to 4 cells in series.

Monday, April 6, 2015

59. THE ANATOMY OF AN IPHONE 6

NASA landed in July 1969 the Apollo 11 on the surface of the moon. Along with the famed astronauts was the Apollo Guidance Computer (AGC), a 1960s-era computer to navigate the spaceship to the moon and back. A reliable machine for its time, it stands today only as a relic. A modern smartphone, for example the Apple iPhone 6, is about 25,000 (yes, thousand) times more powerful, yet a a tiny fraction of the cost and size of the AGC. This early computer had to be powered by three gigantic silver-zinc primary (non-rechargeable) batteries with a total capacity of 3.4 kWh (enough to power a modern house), whereas the iPhone 6 takes its power from a modest 7 Wh rechargeable battery. This picture puts into perspective the marvel of modern electronics: in essence cheaply packing immense computational power into a small volume, and drastically changing our daily lives. As we have peeked in the past inside a Tesla electric vehicle, let's take a closer look at the inner working parts of the iPhone 6, an icon for modern smartphones.

This blog is not about making another teardown of the iPhone 6 -- these are abundant on the internet. Instead, I will build on them to give you additional insight into how these various components tie together, and ultimately, how they relate to power consumption and battery use. Opening an iPhone 6 reveals, in addition to the battery and display, a printed circuit board, called the main logic board, that includes a large number of integrated circuits (IC) and other small components. This is the "brain" (so to speak) of the mobile device, and is responsible for the bulk of its power usage. Let's take a closer with photographs of the front and back sides, then begin to decipher the various sections and their functions.

Photographs of an iPhone 6 teardown showing the main logic board, both front and back. Courtesy: iFixit

First, one observes that the battery takes up a substantial space of the mobile device. This is exactly why energy density is so important. Space, hence volume, is limited yet the energy demand on a mobile device is increasing. This dictates that energy per unit volume, i.e., energy density should increase, at least as fast as the energy demand of the device will in the foreseeable future. That has been and will continue to be a challenge for the industry.

Next, one observes some obvious large sections. One the front side, the main processor, the Apple A8, is very noticeable and takes up a significant part of the logic board. What you see is actually a little module that incorporates the A8 processor chip along with its 1GB of RAM (memory). At its peak operation, this processor can consumer several watts of power as I discussed in an earlier blog. The nanoSIM card holder is also quite substantial.

To the far left of the A8 processor is the wireless section highlighted in the yellow rectangle. This is where the wireless radio frequency (RF) signal is received from the antenna, then amplified before being read by the electronics, and in the opposite direction, amplified before being transmitted by the antenna. These power amplifiers (because they increase the power of the incoming signal) are provided by companies such as Avago, Skyworks and RF Micro Devices; they are also responsible for significant power consumption (i.e., heat) with the mobile device (up to several watts), especially if the cell tower is distant. One also notices a small chip called envelope tracking. This particular IC is made by Qualcomm (QFE1000) -- albeit there are several other suppliers too -- and is responsible to adjust the voltage supply to these power amplifiers only to the amount needed to efficiently operate them; the result is some nominal power savings and less heat generation. Envelope tracking did not exist a few years ago; this added chip (and ergo cost) is strictly to help conserve the limited battery resource.

At the core of the wireless functionality is the LTE multimode baseband chipset made by Qualcomm (MDM9625M) shown in the photograph above by an orange rectangle labeled "LTE radio." It is responsible for managing, encoding and decoding the wireless signal, both data and voice, on the 3G and LTE bands, and do it at an extremely fast speed -- 150 Mbps in this particular case. This type of products is big business for companies like Qualcomm, Intel and MediaTek.

Above the Qualcomm baseband chip one finds a gyroscope and accelerometer combination chip made by Invensense (MP67B). This MEMS device incorporates a total of 6 sensors: 3 accelerometers to measure acceleration in the 3 basic directions, and 3 gyroscopes to measure rotation. The teardown also identifies a 2-dimensional low-g (i.e., sensitive) acceleration sensor from Bosch (Sensotec BMA280). I suspect that is the inclinometer -- it can tell which orientation the screen has been tilted.

The back side of the logic board is home to a number of functions, some visible to end users, and others less so but equally critical. Let's start with the visible functions. First, there is the M8 coprocessor. This is a low-power ARM processor that can handle basic functions like motion without the need to wake up its more power-hungry big brother (the A8). There is also the flash memory for data storage -- this is where you store your music and photos. Then, there is the WiFi module, built by Murata (339S028). It provides fast WiFi connectivity 802.11a/b/g/n/ac. That same module appears to also provide Bluetooth connectivity.

Apple Pay functionality is centered around the NFC module which includes both NFC wireless connectivity (that's the wireless link between your iPhone and the payment station) as well as the secure element that Apple says provides the encryption behind this secure payment method.

Now we transition to lesser visible but quite critical functions. The touchscreen controller made by Broadcom (BCM5976) handles the display functions including reading the finger swipes on the screen. There are also two power management integrated circuits (called PMICs), a primary one made by Dialog Semiconductor and another one made by Qualcomm (PM8019). The primary PMIC provides a number of critical functions such as providing voltage rails to the logic board as well as battery charging. This is also where the primary battery management functions tend to reside though details tend to vary across PMIC suppliers. It is likely that the PM8019 provides and maintains the voltage rails specifically to the Qualcomm baseband chipset. It is not clear whether the fuel gauge functionality is integrated within the Dialog PMIC or whether it is integrated into another fuel gauge chip within the battery itself (and hence not visible by this particular teardown). Public teardown reports of prior iPhones and MacBooks have shown that Apple tends to place the fuel gauge, often made by Texas Instruments, right on the connector cable between the battery and the logic board. It's not clear whether the iPhone 6 follows this same architecture.

Naturally, what a teardown fails to reveal is the richness and depth of embedded software that resides with the multitude of chips on the logic board. Many of these chips above have some type of intelligence on board; of course, the A8 processor is by far the most capable, but even the NFC chip has a small processor (called a controller) on board, and it too needs a certain amount of intelligent software to have it run efficiently.

While the iPhones have been some of the most observed and torn down smartphones, the concept of high density integration applies to all modern smartphones. The space is tight and cost is paramount, and that will continue to drive innovation in this market. In parallel, and equally important, there is a race to implement ever more intelligent software to extract the most performance and operation from the immense amount of computational power that now resides inside a mobile device. Richard Feynman's challenge in 1959 on "There is plenty of room at the bottom" can now take on a whole new meaning and aspiration.