Intelligent battery junction box for voltage and current synchronization
Resources
Welcome to the [INAUDIBLE] management systems seminar session. This is intelligent battery junction box for voltage and current synchronization session. My name is Daniel Torres, and I'll be the moderator for this session.
A few housekeeping items before we get started. All participants are needed for the session, and please use the Q&A box to ask any questions. We will be answering the questions throughout the session in the chat. Also chat if you are having any problems hearing or seeing the presentation. The different windows on the screen are adjustable, so you can make the slide screen bigger or move around based on your preferences. With that, I will hand it over to Sudhir Nagaraj to get started.
Thank you, Dan. Hi, everyone. I'm Sudhir Nagaraj, and I'm happy to talk about the intelligent battery junction box for intelligent BMS systems with you all. We'll also talk about voltage and current synchronization.
So this is the brief agenda for the presentation. So first we will introduce intelligent battery management systems. After that, we will talk about various voltage, current, and insulation resistance measurements. And finally, we will discuss the VI synchronization in battery management systems.
So introducing the battery management systems. So the main function of the battery management system is to monitor cell voltages, pack voltages, and pack current. So what you can see in the green box is basically a battery pack, which has multiple cells stacked in series. So this will be your typical high voltage EV. And the main blocks inside the battery pack are the cell supervisor unit, which has the cell monitors that are monitoring the voltages of the cells as well as the temperature of the cells.
And then the next block, you have the Battery Management Unit, often called the BMU. Sometimes it's also known as the BCU or the Battery Control Unit. This typically has a microcontroller, which manages all the functions within the battery pack.
And finally, the last block here is the battery junction box. The battery junction box is nothing but a relay box or a switch box. It has big contactors that basically connects your entire battery pack to the outside world, which is basically the load inverter, the motor, or even the charging, the DC charging. So that's the function of the junction box. To connect the battery pack to various other loads in the car.
So the two images you see here, on the left side, you have the traditional BMS. And on the right side, you have the intelligent BMS. And the main difference here is actually the junction box. So on the right side, we have something called the intelligent junction box.
So basically, again, coming back to the left image, you can see if you observe the junction box, there are no active electronics inside of the junction box. It is just passive contactors and fuses. All the measurements in the junction box, the various voltages, et cetera, are actually taken-- the wires are taken from the junction box into the ADC readouts, which are actually present in the BMU.
So now looking on the right side, the intelligent junction box, what you can observe is there is a dedicated pack monitor now inside the junction box. This pack monitor measures all the voltages as well as the currents inside the junction box and passes on the information to the MCU using a simple twisted pair communication. So this is what leads to the intelligent junction box. And the benefits mainly is that it eliminates numerous connections between the BMU and the BJB interface.
So a lot of the voltage measurements are sensitive measurements that actually cross the BMU BJB interface can be eliminated. This is also another way of making the measurements more robust. It also simplifies the hardware and MCU software development, because oftentimes pack monitor and cell monitors, they come from the same family of devices, which means their architecture, the register maps, they're all very similar. They also have features which enable better synchronization in the VI measurements. So that basically simplifies the MCU side of things as well.
This is another view. Basically, the pack monitor from TI, which is the BQ76931, it supports both traditional architectures as well as the intelligent architectures. The traditional architecture is supported by UART based communication. And here you can, again, see various voltages are going from the BJB to the BCU over wires.
And the ADC is inside the pack monitor, measure these voltages, and then transmit the data to the MCU over UART. On the bottom, the same pack monitor is right now inside the BJB. And it is measuring all the voltages locally inside the BJB and transmitting the data back to the MCU over a daisy chain interface. So this is what leads to the intelligent BMS architecture.
Coming to various voltage measurements in the battery junction box. So the function of the voltage measurement in the battery junction box mainly is to measure the pack voltage and then measure the insulation voltage-- we will talk more about the insulation detection down the line-- also to measure the health and the status of various fuses and contactors inside the battery junction box.
For example, if you have a load or the link contactor, if you want to check if the contactor is open or closed, then you measure the voltage on either side of the contactor to determine an open or closed state.
Finally, the voltage measurement also measures the DC fast charging port. This is to measure the incoming DC voltage for charging. So you want to make sure that it is in the correct range before you connect this to the battery pack. So that is the function of this measurement. So these are the main functions of voltage measurements within the battery junction box.
And some key properties to these measurements basically are-- just one second. OK. Are the 1% error in accuracy or measurement accuracy. So that is one of the requirements that we are seeing. So in order to support less than 1% accuracy for a high voltage measurement, when I say 1% accuracy, let's say you're measuring 400 volts and 1% is your error budget is basically four volts. So that's the target.
All the voltages, the high voltages, are actually divided down using resistor strings. And these resistor strings then feed into the ADCs within the pack monitor. And the pack monitor is essentially just reading divided down voltages. And because it is reading divided down voltages through these resistor strings, which have large resistors, it is very important to have extremely low leakage in these ADCs that are measuring the voltages, because any leakage current causes IR drops and that can lead to measurement inaccuracy.
So this view basically shows how various high voltages are flowing through the resistor string and are being connected to different resources inside the BQ76960x device, which is the pack monitor. Many of the measurements are differential measurement channels, and all the channel schemes are shown here.
So here, we are measuring both the positive as well as the negative voltages with respect to the pack ground. When I say with respect to the pack ground, what I mean is that the pack monitor essentially sits on the negative terminal of the high voltage battery or the ground of the high voltage battery. And that is the reference point for all these measurements.
So apart from this, there are some temperature measurement requirements inside the battery junction box. So various temperature measurements are needed inside the battery junction box. Some of the temperatures are basically you're measuring the temperature of the shunt resistor to figure out if any kind of temperature compensation needs to be applied in the MCU.
And then all you're measuring the temperature of the contactors. Since they're passing very high current, you're trying to make sure that the temperature is-- they have possibility to heat up. And you're trying to make sure that the temperature is within the range. And so these are various temperature measurements.
And these are typically done through NTCs or PTCs. And these NTCs and PTCs are connected to our TSREF that comes from the device. That's a reference from the device. And we measure them using these GPIO channels. These channels support ratiometric measurement.
So ratiometric measurement eliminates the reference voltage errors that are present in the system. So that gives you the best accuracy for temperature measurements. So the typical min max for this is plus minus 0.2%. So this translates to roughly one or two degree accuracy in the mid to high temperature range.
Next coming to the current measurements. Current measurements are done using two possible schemes. So one is a shunt resistor. So as you can see here, all the current in and out of a battery basically flows through a shunt resistor. So when I say into the battery, that is during charging.
So the current flows through the shunt resistor and when it is discharging through a load, it is again flowing through the shunt transistor. And because the currents in the EV can go up to kilograms, these shunt resistors are extremely small. They are in the range of 25 microns to 50 microns.
And we internally have our primary and the redundant ADCs to support functional safety architectures. Some of the requirements for this measurement, accuracy requirements basically say that we need to support less than 0.3% error from the measurement. So these ADCs are very precise 24-bit ADCs, highly accurate.
And the conversion rate, there are various conversion rate options. These conversion rate options are given to match the characteristics of the filters on the voltage measurement path, for example. This is to give a good VI synchronization between the voltage and the current measurements. We will talk more about VI synchronization in the subsequent slides.
The other method of measuring current in EV is using a Hall sensor. So Hall sensors are also used. And firstly, we are seeing that the modern day shunt resistors are able to achieve higher accuracy across the range. So when I say across the range, right from a few milliamps to kiloamps of current through the shunt transistor. The shunt resistor gives a good performance.
A Hall sensor typically is limited in the sense that its dynamic range is not as wide cover milliamps to kiloamps but using just one Hall sensor. That is the reason sometimes there are multiple sensors, let's say two Hall sensors that are used. The first one could cover milliamps to, let's say, 100 amps. And the next one would cover from 100 amps to kiloamps of range. So two Hall sensors covering different ranges can make up for the entire range.
Also Hall sensors primarily measure the current through a coil by measuring the magnetic field around it. So that is the mode of measurement in the Hall sensor. So this is inherently susceptible to EMI. So if there is stray magnetic field around the measurement, that could be perceived as noise. So that is one of the drawbacks for using a Hall sensor.
What are the advantages of Hall sensor? So the main advantage is the flexibility of placement. It can be placed anywhere either the low side or the high side. And so that flexibility is powerful as well as the inherent voltage isolation. Because the Hall sensor is a magnetic detection or a magnetic sensing, it does provide the high voltage to low voltage isolation. So if you have a readout circuit in the low voltage domain, measuring the high voltage, you don't need to add any other voltage isolation methods.
So the BQ device, the pack monitor, is capable of measuring the Hall sensor output as well. A typical Hall sensor has an analog output. And the BQ device due to its numerous ADCs, it can easily input these Hall outputs and can transmit the data in the ADC converted data to the microcontroller.
Overcurrent detection. So overcurrent detection is a very important feature in-- safety critical feature in BMS. It is responsible to protect the battery pack from overcurrent events as well as provide safety for the operator. And overcurrent detection basically first involves detection.
And the next step can also involve action. Action is basically detecting an overcurrent. There could be local action taken, such as triggering a pyro fuse to disconnect the battery pack from a load that is causing an overcurrent. Or it could also disconnect the contactors, open the contactors that is connecting the battery pack to the load or the fast charging. So these are the reasons why overcurrent detection is very critical.
And the way the pack monitor, the BQ7963x, handles though our current detection is by using an external amplifier and internal OV UV comparators. The scheme is shown here. Basically what is being done here is the voltage across the shunt is amplified and is fed to an input channel that does monitoring and OV UV. So OV could be programmable and it can be set to detect a charge current. For example, if there's an over charge current, then the OV would trip.
And similarly, if there's an over discharge of some loaders suddenly shorted and suddenly takes a lot of current, the UV would trip. And both these would basically signal a fault on a pin, on a fault pin, as well as over the daisy chain to the MCU. The fault can be propagated in both ways. And in fault being local inside the battery junction box, it can be used to trigger something like a pryo fuse. A typical response time for such a system or a scheme is 500 microseconds. And the deglitch for the scheme is about 100 microseconds.
All right. Now, coming to the insulation resistance and its measurement. So what is insulation resistance? Insulation resistance is the resistance of all the insulation materials used to isolate a high voltage battery plus and minus terminals from low voltage chassis ground.
So this is basically if you visualize the high voltage battery pack, the connection to the load. So it has a thick cable where all the current is going. And around that cable, you have the insulation material. And this material represents the insulation. And not just this. Even the insulation in your PCB board and other forms of passive insulation used could all be lumped into this insulation resistance.
And in this image, you can see the insulation resistance is represented by the two red resistances, the Riso_P, which is the resistance from the pack plus to the chassis ground and the Riso_N, which is the resistance from pack minus to the chassis ground. So what we are trying to monitor here is basically either of these resistances, if they fall below a certain threshold, then that means the pack-- high voltage, the high voltage pack could-- the high voltage in the pack could flow to the chassis ground.
And the chassis is where the low voltage electronics as well as a user comes in contact with. So this high voltage forming a low impedance part to the chassis can lead to damage in low voltage electronics, as well as harm to the user. So that is why it is very important to monitor these resistances continuously actually during the operation of the car to detect if it falls below the threshold and then signal a fault to the car.
So the way the scheme in which it is done basically is because we have two unknown resistances, Riso_P and Riso_N, we need two equations essentially to solve for them. So the first equation is obtained by using the switch SW3. So you basically open the SW3 that forms a certain network of resistors. And then you measure the voltage on the GPIO1. And this gives your first equation.
The next equation can be obtained by closing the switch SW3. That basically forms another network of resistance and you again measure GPIO1, the voltage on GPIO1. That gives you a second equation. So solving for these two equations, you can obtain the values for Riso_P and Riso_N. And then once you obtain the values, you can check them against a threshold. If they've fallen below a set threshold, then you can flag a fault. And the car will then take appropriate action.
So we basically spoke about voltage monitoring, current measurements, and insulation measurements. So those are the three main functions of the pack monitor. And now let's talk about the VI synchronization in BMS. VI synchronization is basically synchronization between the pack monitor as well as the cell monitor. So basically we bring the cell monitors into the picture here as well. So let's discuss that.
So VI synchronization in BMS. So what are cell voltages, pack voltages, and pack currents used for? They're mainly used for battery monitoring. They used to calculate state of charge, state of health, and these are done through electro impedance spectroscopy. So basically impedance of the cell.
So you're measuring the impedance of the cell by measuring the V and I across the cell. And also power. So you want to monitor the instantaneous power of the car. So nowadays, cars are telling you when you accelerate what is the instantaneous power going out of the battery pack. And when you're charging, what's the power coming into the battery packs. For all those accuracies, your V and I both are important. So again, diagnostics as well involve various pack voltages.
One point to mention here is a pack monitor is typically used to monitor the voltage across the entire pack. But in some few cases, the sum of the cell voltages can also represent a pack voltage. That is possible as well. So the cell pack voltage as well as the pack current, they need correlation and time to provide the most accurate power as well as impedance estimates. So samples taken within a certain time interval, that is called the synchronization interval or the VI sync interval.
So smaller the synchronization, more accurate is the power estimate or the impedance estimate. That is being shown in the examples in the bottom. So in the left side, you can see the VI synchronization window is still small. So in the small window, the voltage and current is measured. And this provides a good estimate or an accurate representation of the Vi in the system at a particular time instant.
In the right side, what you can see is the voltage measurement happens before a dip event. But the current measurement happens after the dip event. This is because the VI synchronization window is wider and there is a possibility of events happening between the voltage and current measurements. And this basically leads to when you have a power or an impedance estimate, which is the relationship between the voltage and the current, this leads to an inaccuracy there.
So non-synchronized data can lead to errors. So this is directly proportional. For example, 1% error in estimate of the state of charge directly translates to typically a 1% shorter range in the vehicle operation. So the more accurate you are, the less margin you need. And you can squeeze the battery or utilize the battery to a larger extent, which means more range.
So what are the requirements and challenges in synchronization? Some of the requirements are these days we are hearing requirements for synchronization less than one millisecond. And also all the samples, the cell voltages, the pack voltages, the pack currents, all of them need to be measured within 10 milliseconds.
And what are the challenges in doing this? So the cell monitors, you have multiple cell monitors stacked one on top of the other. They are non-synchronized. So between each cell monitor, they have three running clocks and they're not synchronized. And so that is one challenge.
The second one is the readout. So each cell monitor has it's measuring typically 12 to 18 cells these days. And each of the cell data is like 16 bits long. So there's a lot of data that needs to be transmitted from one cell monitor to the next. And so if you have a free running ADC, what that means is by the time you transfer data from in the daisy chain at least from one cell monitor down to the next, that itself could exceed or come close to the VI synchronization budget.
And by then, your first ADC has already moved to the next sample. So you missed an entire cycle or a few cycles of voltage. Or there is a few cycles of voltage sample difference between one cell monitor and the next cell monitor. So this is how the data readout actually affects the synchronization.
Next is the influence of filter. So any filter in the path, like a voltage filter or a current filter, we need to carefully consider how the filter characteristics contribute to the VI synchronization. So if you have different filter characteristics in the voltage in the current path, that could lead to different phase shift in the current and the voltage data. And so that will lead to in time the synchronization will be off.
Command propagation delay. We will talk about that in the next slide. Then bandwidth and communication interface. This is, again, to do with the similar thing I said. If your data rate, the readout rate is slower than your synchronization and in case of free running ADCs, this will lead to an issue in synchronization.
So let's discuss some synchronization with the BQ7961x family of battery monitors. So this is a typical scheme that we see today. So you have the cell monitors on one daisy chain that is connected to the MCU. And then you have the pack monitor that is measuring the current and the pack voltage that is connected on another daisy chain to the MCU. So the MCU now becomes the common point between the two daisy chains. So it is now responsible to synchronize the cell voltages to the pack current.
So basically, all it needs to do is monitor the time or maintain a time relationship between issuing a go command or ADC start command to the cell monitor and the ADC start command to the pack monitor. And if it does that within a few microseconds, that synchronization should be sufficient to maintain very tight overall system synchronization.
All right. So coming to cell voltage measurements and timing in the BQ7961x device. So voltage synchronization within multiple cell monitors. We are talking about just synchronization within cell monitors right now. Only cell monitors. So using a single conversion mode, all cell supervisor units can synchronously measure the cell voltages. So the BQ7961x supports delayed ADC sampling in order to compensate for CSC to CSU propagation delay when transmitting the ADC go command down the daisy chain interface.
So what it means is basically if you look at the image to the left, so all the commands basically go through the Cell Supervisor Unit, the CSU. There is a certain amount of delay when it goes from one CSU to the next CSU. For example, here S1, S2, S3, and S4 each represent the CSU. So S2 is a CSU on top of S1. S3 is a CSU on top of S2. A command that goes from S1 to S2, it takes about 14 microseconds to basically-- that's a delay between the command reaching S1 and the same command reaching S2.
And after that, the delay is shorter. For example, the command reaching S2 to the command reaching S3, the delay is just four microseconds and so on and so forth. So now basically what happens is when you say ADC start, so S1 receives the command. And if you look at S4, it receives the same start command for the ADC 22 microseconds later. So this 22 microseconds now needs to be considered within the VI synchronization window.
But we have a trick and a feature. Basically what we allow to do is each of the devices, we can program a delay in the ADC start. So that is shown on the right side. So basically what we're doing here is the S1. We program a delay of about 20 microseconds. S2, about 10 and S3 about five microseconds. So by the time the go command reaches S4 and S4 ADC starts, the S1 also because of its delay, it starts around the same time.
Similarly, the S2 and S3 also start around the same time. So this essentially looks like all the ADC start within the plus/minus two microsecond window or a skew. So that is what the delay function allows you to do. Minimize that 20 odd microseconds or depending on your stack, it could be even higher. Minimize all the delay down to something in the order of two microseconds. So that's the function that the BQ7961x device provides.
So coming now to the pack monitor. So we spoke about synchronizing all the cell monitors using the delay function. Now integrating the pack monitor into the equation. So the pack monitor and cell monitor, they can all be synchronized within approximately to 15 microseconds. This can be done by using a conversion rate in this example of one millisecond for the current measurement.
So the blue line represents the cell voltage readings and the yellow below represents the pack monitor readings. The top is the voltage. The bottom of the current. So the peak of the current or rather the weight of the current is in the center.
So now, basically I think what is shown here is all the voltages in the pack monitor, this pack monitor has about-- and the cell monitor has about 16 channels. Each of these channels is measured serially. So the delay within the device itself from the first channel to the last channel measurement is about 120 microseconds. So within 120 microseconds in every device, all 16 measurements are complete. So that at least is a minimum synchronization time for all the voltages.
Now, we have already seen that all the cell CSUs can be synchronized with each other using the delay function. So that all falls within this 120 microsecond budget. And finally, the current sample being-- the weight of it being in the center.
If you look at the start of the V1 to the center of the current sample, that entire delay now can be within 253 microseconds. So that overall is the entire VI synchronization capability of the system, of this BQ796x system. So this is a VI synchronization.
So there is a reference board for the pack monitor. This reference board basically has various high voltage inputs. It has multiple communication options like UART and daisy chain. It also has the isolated low voltage supply to supply the pack monitor. This is available for sampling right now.
So in summary, firstly to conclude on the VIR measurements. The voltage and current measurement accuracy improvements will directly result in optimal utilization of a battery. So that is the key to voltage and current accuracies. Overcurrent detection and local autonomous triggering of pyro fuse will enhance the safety in vehicles. Integrated insulation detection capability provides enhanced safety for operators and sensitive electronics inside electric vehicles.
And moving to the conclusions on the VI synchronization. So effective VI synchronization enables precise state of charge, state of health, and electrical impedance spectroscopy calculations that will result, again, in optimal utilization of the battery, to extend the battery life, as well as squeeze more miles out of the battery. So that is the overall conclusion and summary.
Thanks, Sudhir. And thank you everyone for joining us. You will receive an email with a link to the on demand recording for this session. There will be a brief post event survey that pops up after the end. We'd like you to feedback so we can continue to improve our content for future events. And again, thank you for your time today, and have a great rest of the day.