Hi, and welcome to part two of this training videos covering battery and system monitoring in smart meters. This is the detailed agenda of this training. In part two, we will discuss the principle type of primary battery used in smart meter, and how battery health can be monitored. I will also describe the characteristics that will effect their performances.

There are two main types of primary batteries that are used in smart meters. The first one is lithium manganese dioxide, or LIMn02. And the second one is lithium thionyl chloride, or Li-SOCl2. Both of these battery provide very high power density and a very long lifetime, which makes both of them perfect for powering smart meters during decades. The main advantage of the lithium manganese dioxide battery is the capability to deliver a higher current pulse. For the lithium thionyl chloride battery, the main advantage is the higher power density and the higher starting voltage of 3.6 volt.

This slide also shows the difference of the discharge curve of these two batteries. As you can see, the lithium thionyl chloride battery has a very flat voltage of a discharge curve that makes it harder to give a precise estimate of the remaining power just based on this voltage. I will introduce later TI gauging solution for these type of batteries.

Some part of the application can effect the batteries' performances. First, high temperature can generate passivisation, which will increase the internal resistance of the batteries. The discharge rate required by the system can significantly reduce the battery voltage in case of high current spikes. The system background current has to be taken into consideration, since the weather and gas meter will stay a very long time on the field. Even a few micron can have important impact on the battery capacity.

The cut off voltage will depend on the voltage drop due to high current spike required by the system. Indeed, if the system needs 1.8 volt in the battery, voltage is 1.85. In the case of a current spike, the battery voltage could drop to 1.7 volts and cause the [INAUDIBLE] to the system to work properly.

And finally, the battery has a self-discharge current that has to be taken into account since the impact over a long period will be important. For instance, here, with 1.1 microamps self-discharge on the battery, it will be equivalent, after 10 years to 91 million power loss.

Lifetime prediction of primary battery is difficult, since a lot of independent variables need to be taken into account. Indeed, the battery lifetime is very dependent on the actual field usage, like temperature, discharge rate, cut-off voltage, and self-discharge. The current solution that I use to make sure the system will not run out of battery is either to oversize the battery or use a second battery.

TI battery gauging team developed the BQ35100, which is a battery health monitor front end that is able to evaluate the battery condition. Three parameters are used to monitor the battery. The battery voltage, the current coming out, and the temperature. It's able to calculate the battery State Of Health or, SOH in percent, and to detect a low SOH state. It's also able to detect the end of service of the battery, the low voltage state, high or low temp states, and the battery disconnection.

Here are the challenges of monitoring a lithium manganese dioxide battery. The photograph on the top left shows the effect of a current profile on the battery. The low profile used on this graph is 30 million peak current for 5 seconds, followed by 20 minutes of relaxation. If the battery is fully relaxed, at the first load spike, the voltage will drop significantly and then recover little by little.

The graph on the bottom left shows the voltage drop versus the SOH in percent. As we can see, it's harder to have an exact estimation at the beginning of the discharge, since the voltage drop necessary to decrease the SOH of one percent is very small. The ADC used to measure the voltage has to be very precise to see those variations.

The last graph on the right shows the effect of the temperature on the SOH. When the battery is fully charged, the temperature has no important effect on the SOH. But when the battery is discharging, the effects of the temperature become bigger and make it harder to estimate the remaining battery. This is critical, since it's very important to have a precise estimate at the end of the battery life.

Here are two examples of micro-controllers that could be found in the metering field. On the left, you can see the specification extracted from the data sheet. To get a precise estimation of the SOH from 100% to 0% it's required to have less than 1 millivolt of resolution. With most MCU internal ADC, the resolution is 12 bits, a bit lower than 11 bits in effective number of bits. So the smaller measurable change is not easily reaching the required resolution.

Here is an example on how you can save power with BQ35100 converter using the MCU internal ADC to monitor the battery. BQ35100 can independently monitor the battery while the MCU is in standby. And these allows, in some applications, some interesting power settings. indeed, the BQ35100 consumes between 0.06 microamps and 0.35 microamps of average current, depending on the monitoring mode. The MCU just need to send an I2C command to get the monitoring data already calculated by the BQ35100.

And here is another example on how you can get better estimation of the battery voltage with the BQ35100, compared to using the MCU in a normal ADC to monitor the battery. By using an MCU to measure the battery voltage, current is consumed to power the internal ADC, and it creates a voltage drop. Indeed, most MCU internal ADCs consume between 0.3 and 2 milliamps, which can cause a significant drop that could lead to an imprecise battery state estimation. The BQ35100 needs much less current, and the effect on the battery is nearly imperceivable. This allows a more precise estimation on the battery remaining lifetime.

Here is a quick comparison of using an MCU plus internal ADC to monitor our battery voltage versus using the BQ35100 and the ADS7142. For monitoring with the MCU's ADC, the only advantage you will get is that you will not have to have extra components. But on the other end, the disadvantage-- you will have less accuracy in the ADC, and you will need at least 14 bits. You will have a higher current consumption. The battery voltage will drop while you are measuring it. You will be unable to detect abnormal behavior, because you will not measure the current. And you need to adjust your gauging algorithm every time another type of battery is used.

But if you monitor your battery with the BQ35100 and the ADS7142, you will get a lot of advantages, like getting higher accuracy, allowing the temperature compensation, having ultra-low power consumption, the possibility to detect abnormal discharge with the ADS7142. TI can also provide you a learning cycle of your battery if you want to change your battery type. It's available on TI.com on request. This solution will also provide you an accurate SOH and abnormal discharge detection, and this will enable predictive maintenance and allow you cost settings. And the only disadvantage you will have for this solution is that you will have to add one or two components to your design.

Here is an example of test results done with the BQ31500 and the ADS7142. To perform this test, we simulated a harder transmission load by shorting the battery to the ground through a 100 ohm resister for 5 seconds every 2 minutes. The standby current is simulated by a 10 mega ohm resistors.

The BQ35100 and the ADS7142 monitor the battery in the system during the discharge. The graph on the left shows the battery voltage, the State Of Health of the battery, and the temperature is below. The duration of the test is seven days. And you can see the effect of the temperature on the battery voltage, because the temperature is changing in the building during the day. The BQ35100 perform voltage, current, and temperature measurements to compute the SOH. The ADS7142 measures continuously the current during our transmission and during standby. And we are using micro-controller to count the number of cycle, get the data from the BQ35100 and the ADS7142, and write the data in a CSV file on an SD card.

This slide shows a very important feature of the BQ35100, which is temperature compensation. On the top graph, the battery voltage is represented in red, and SOH in grey. The graph below shows the temperature.

In the middle of the graph, we can see that the voltage is constant around 2,978 millivolts, but the SOH is going down at the same time. The reason is because we are discharging the battery at the same time as the temperature is rising. The battery discharge is supposed to reduce the voltage, and the temperature rising is supposed to increase it. The result here is a constant voltage. But from this voltage, the BQ35100 is able to monitor our discharge.

To summarize the second section, the remaining lifetime of a primary battery is not easy to monitor, because a lot a factor can alter it. A dedicated solution like the BQ31500 help in getting a more accurate result by using voltage, temperature, and current measurement to estimate the battery life time. It also allows a lot of advantages compared to using an MCU internal ADC to monitor the battery, and can do so with lower power.

So this is the end of part 2 of this training. Please now watch part 3 of this training series to know more about how our new platform for battery and system monitoring. Thanks for watching.