Internet Explorer is not a supported browser for TI.com. For the best experience, please use a different browser.
Video Player is loading.
Current Time 0:00
Duration 36:18
Loaded: 0.46%
Stream Type LIVE
Remaining Time 36:18
 
1x
  • Chapters
  • descriptions off, selected
  • en (Main), selected

Hello, and welcome to this battery management system seminar session. This is a session for battery gauging for high-cell-count industrial systems. My name is Matt Sunna, and I will be the moderator for this session.

A few housekeeping items before we get started-- all participants are muted for the session, so please use the Q&A box to ask questions that you have during the session. We'll have time at the end of the session to answer questions from the Q&A box. You can also use this Q&A box if you have any problems hearing or seeing the presentation.

The different windows on the screen are adjustable, so you can make the slides bigger, or you can move them around according to what your preferences are. And with that, I will hand it over to Xiaodong to start this presentation.

All right. Thanks, Matt. Hi, everyone. This is Xiaodong Cai, System Engineer at TI. So today, we're going to talk about battery gauging for high-cell-count industrial systems.

Let's give you some background about why this topic is important and the structure of this presentation. So a lot of industrial applications require high-cell-count battery pack, and these battery packs are ranging from 30 volts to over 100 volts, which means it has a lot of number of cells.

For instance, power tools require 5 to 10 cells. E-bikes require 10 to 23 cells or higher. Battery packs require 25-plus cells in series for that high power and high voltages. Sometimes, some of these pack even paralleled some of these cells to make them physically very large.

So given these packs being so high number of cell counts within them, the battery gauging requirements is divided into two parts. One is to meet the requirement for safety and standard compliances. That means basic monitoring and protection. And then another part is for gauging the battery to display any state of charge, to estimate the remaining system runtime, quote unquote, "time to empty," or remaining charge time, quote unquote, "time to full."

And some of these batteries are requiring these numbers in these type of features in different requirements. Hence, this topic is highly useful for those battery pack makers or customers to know about how TI's provide solution towards this area.

So basically, this presentation is divided into three ways to solve this problem. The first way is, if customers choose to do a proprietary algorithm using cell voltages, current, and temperature measurements, what TI is offering to help them to solve that issue. That's topic number 1.

And for some of those customers, because they might want to use top-of-stack as a gauge input and then also measure the battery pack voltage, current, and temperature, and then calculated the average gauging algorithm for the entire pack. So that's solution number 2, how to do that.

And then solution number 3 is some of those customer wants to gauge the battery using per-cell gauge. Basically, they look at each cell's voltages and then use that data to gauge the battery status. So that's solution number 3. TI is providing all three solutions in different ways, and that is the structure of this presentation.

At the end, we're also going to talk about if some customers want to use out-of-range battery range, how the customer can scale that battery voltage or current level outside of the solution specs by cheating on the scaling solutions. So that's a way to even further out to the even higher range of voltages or current for the battery packs that is enormous.

All right, so let's look at solution number 1, some of those customers that wants to use proprietary gauging algorithm. So for this solution, TI is basically offering the analog front end, and customer take that front-end data to do their own proprietary gauging algorithm.

So the analog-- the proprietary gauging algorithm usually require a system-on-chip solution that provides accuracy for the different voltages and currents and then also sometimes require peak load, making sure the device is not using too much of a load.

And then some of those batteries, like they're using lithium iron phosphate battery that require more accuracy and higher peak currents-- TI's offering is to provide that analog front end that is precise and providing very good data so that the controller that takes in these data from the analog front end can do their gauging algorithm to gauge the battery pack.

So the first solution we provided for such a proprietary gauging algorithm is the BQ76930, or BQ769x0 series. For this family, the analog front end, which is where my mouse circles-- the first front end is to measure each individual series cell voltages. There's also a dedicated coulomb counter to look at the current measurements using the sense resistor.

And then this solution also supports two thermistors to do temperature monitoring. So this is a basic solution to provide those data into the controller so that the customer can develop their own proprietary algorithm.

For those customer who wants a more advanced or precise gauging algorithm, we recommend to use BQ769x2 family, which is our latest family for [INAUDIBLE] our front-end monitors. For this solution, we provided a 24-bit data for pack analysis for cell voltages and currents.

We also have coulomb counter that do integration of passed charge. This device can support up to nine thermistors for temperature sensing. So that's a lot of thermistors for such an application. Everything that's needed for modern analog input packs, we offer that here for this solution.

If we look at the feature list for such a new family that we released last year, from monitor side, we obviously monitor the voltage, the current, and temperatures. And we have integrated the dedicated coulomb counter for that monitoring.

For protection, we provided a very long list of different protections-- basic overvoltage, undervoltage, overtemperature charge, overtemperature discharge, et cetera. We also offer open wire, short-circuit detections, cell balancing, and watchdog, and more.

If we look at the FET driver for such family, this family is offering a integrated charge pump for high-side driving. It also has two PI open for hook it up to low-side drivers externally if a customer prefer low side. So it's not limited to high side, but it has an integrated high-side driver.

It also offer precharge/predischarge cell balancing, and it has multiple LDOs for different external components, such as external MCU, et cetera.

For communication, it can support I2C HDQ, and SPI. And also, for this device itself, it can tolerance up to 60 volts for the 6922, or 85 volts for the 10s, 14s, and 16-cells products.

All right, so that was the analog front end offering for proprietary gauging algorithm development solutions. Now, we can talk about the second solutions, where customer wants to look at per-cell battery gauging and what TI can offer.

Here, we are showing an example for high-cell-count system that can do per-cell measurements. It's a combo solution of our BQ769x0 family that was showing up at the analog front end at the first solution. And also, we combine it with a BQ78350 device, which is on the right-hand side in the circuit diagram.

These two devices work together flawlessly because we developed them together. The BQ78350 can interpret the data from BQ769x0, the analog front end, directly. And the customer doesn't need to develop their own algorithm in order to receive the data or receive the feature list that is going to show in the next slides.

So the right-hand side device, the BQ78350, is using a compensated end-of-discharge voltage method to integrate the coulomb counter data to-- coulomb counter data to integrate past charge and estimate the state of the charge. And then at the end, when the algorithm interpret the data, it will send out a status to external controller through the SMBus.

If we look at the feature list coming out of this combo solution, especially from the 78350 family, it can output state of charge, remaining capacity of the battery, time to empty, time to charge, state of health. So this is a combo solution that is turnkey. Customer doesn't need to touch the gauging hours of the self.

All right, so that was the per-cell gauging method TI offer. The last solution we offer is the top-of-stack battery pack gauging.

So for this solution, instead of looking at individual cell voltages, the system is looking at the top-of-stack battery voltage and then divide it down evenly through the number of cells to treat the number as the voltage of a single cell. An integrated gauging algorithm uses this stack voltage, and then the pack current, and then the pack temperature to calculate the state of charge, state of cells, time to full, et cetera.

If we look at a proposed solution, we're going to look at the device name itself later on. If we look at the proposed solution, this is the schematic for our proposed solution. Let's dive into each individual block of this solution.

So on the top-left side of the circuit, we see a circuitry highlighted by the red area. It's the input circuit for the whole chip. Basically, it's a small regulator to regulate the voltage for the device.

And then the next section we look at is these two FETs that is highlighted by the red box. These couple external components is to divide down the voltage for sensing the battery pack voltage. By frequently turning on and off these FETs, the battery pack voltage can be divided down to be sensed frequently.

And then there are two alert pins highlighted by the red box again. These two alert pins are for the monitor to tell if there's any alert needs to signal the external controller.

And there is a individual thermistor that can be hooked up directly to this device. This thermistor is hooked up to the TS pin of the device. And then the other side of the battery pin is the REG25 pin to provide that power.

And finally, for this diagram, we're looking at a sense resistor that can be configured above or below the VSS connection. Depending on different customer needs, some customer wants the sense pin to be above VSS. Some of them wants to be below VSS. That has different trade-offs.

All right, so having seen that system diagram, let's look into this BQ34Z100-G1 device. This is the Impedance Track Gauge solution chip we are offering. Basically, this device can support HDQ and I2C communication, and this device is looking at that top-of-stack voltage and then interpret everything into the gauge data and then send it out through the communication port.

We're going to dive into some specific spec of this device. One thing to remember is this device alone can support up to 65 volts. Later on in this presentation, we're going to discuss how to scale up to be further out of the 65 volt for different battery pack voltages, such as 100 volts or even more.

If we look at this solution, this solution is only supporting up to 65 volts. And then the current itself can only support 32 amps. There are some limitations for all these solutions we presented, which is the top of the battery voltage and the highest amount of current device can handle.

For some of those applications that require higher number of either voltage or current, this presentation is also providing a solution to that. The solution we're providing is to scale the calibration. For instance, if we have a current that we want to support up to 48 amps, and the device is only taking in 32 amps at the maximum, there is a 1.5 scale factor that we need to work on.

If we increase that scale factor to 2x, then we take an example of, hey, if you have a 5-ohm sense resistor, and we calibrate, we can calibrate a 4 amp out of this 5-ohm resistor instead of the 2 amp out of this 5-ohm resistor. So that means we're scaling up a initially supposed to be 2 amp out of a 5-milliohm resistor. So in that way, we're scaling up 2x out of this by cheating that calibration factor of 50%. In that way, we can support 2x of the maximum battery pack current.

Same idea for the battery pack voltage. If we have a 90-volt battery pack voltage, and then this device is only taking 65 volts, we can use a voltage divider to divide down that voltage of the 90 volt into the battery pin so that the battery pin is sensing much lower voltages to scale down that voltage. And then when we do calibration, we calibrate it out using that factor.

This is our proposal to any battery pack that has a higher voltage or higher current than our device packs, by changing that battery scaling factor. All right, that's the presentation I have.

Thanks, Xiaodong. And thank you, everyone, for joining us. Everyone will receive an email with the link to the on-demand recording of this session, so all sessions are recorded. There will be a brief post-event survey that pops up after the end. Let's take some time to answer any questions that people have. So if you have questions, please go ahead and post to the Q&A box on your screen, and we can take those as they come in.

Xiaodong, I have just one question to you while we wait for the questions to come in that may be helpful for others. How do the different methods-- so you talked about three different methods. How do they differ in terms of accuracy for measuring state of charge?

Yeah, sure. So the first solution, which is the proprietary analog front end with the customer dedicated algorithm is a hard-to-say solution because that depending on how customers do their algorithm.

But if we look at TI's proposal as a turnkey solution for the rest of the two, which is per-cell gauging versus top-of-stack gauging, the per-cell gauging solution is generally more precise and easy to use. The top-of-stack solution is a single-chip solution, which is easier to implement. So it has this trade-off in terms of the accuracy. Per-cell gauging is normally more accurate.

OK. Thank you, Xiaodong. And another question-- do we have reference design using gauges for high-cell-count applications?

Yes, the BQ34Z100-G1 product folder has a reference design to look at for such an application. I think that's listed as one of the external resources on this session. If you guys can click out of the session, look at the external resources. One of the link is for that.

OK. We have a few more questions coming in. "Do these solutions provide per-cell balancing?"

Yes. So our analog front-end chip, most of the time, especially if we look at any solution that has the BQ769x0 or BQ769x2-- those two chips provide cell balancing, for sure. And then our latest cell balancing for the BQ769x2 can support up to 50 milliamp cell balancing current.

Yeah, so I think also-- I'll go ahead and make a comment that also the BQ769x0 family, the previous generation, does have cell balancing, just does not have the automated algorithm. So the host will determine when to balance and when not to balance.

But if the BQ78350 gauge is used, it does have an internal algorithm, so it will automatically enable cell balancing, although this algorithm for this device only works during charging, so not during a relaxed mode.

OK, I have another question that came in. And this one, this question is, "Can the BQ79616 balance high-capacity cells with its internal balance module, such as 107 amp hours?"

And I think I'll just comment that this question is a good question. I think we do have the automotive track coming up, and the experts for the BQ79616 will be present in that. I believe the answer is yes, but I think the experts here today in this session are for the industrial battery monitor, so I would recommend reasking that question during the automotive session.

Mhm.

Another question that came in is, "How does the per-cell solution characterize state of health? Is cell characterization data required by TI, and how does it compensate for partial discharges?"

Yeah, I think this question is divided into three different questions. So the per-cell-- yeah. Matt, please comment.

[INAUDIBLE]. It's muted.

Yeah.

Yeah. So the state of health-- so the first question is, "How does the per-cell solution characterize state of health?" So this is a common parameter reported by our battery gauges. So the BQ78350, for example, is a per-cell gauge.

And the state of health is simply reporting the learned capacity of the battery divided by the original design capacity. So as the battery is charged and discharged many times, and as it ages, it will learn the new capacity as it does decrease. And so then it will update the state of health.

There is cell characterization data required to optimize the gauging parameters. So really, it's the gauging that indirectly will result in a state of health change. And for an impedance track gauge that involves either selecting a ChemID from the TI database if it already exists, or there's a tool called GPCCHEM that can be used to find the closest match.

Or in the most extreme cases, you can work with your sales representative to get your cell characterized at TI and add it to the database. For a CEDV gauge, it's a different type of characterization which just involves six discharge curves and uploading your data to a different online tool.

And I think the third part of that question was, how does it compensate for partial discharges? And that really depends on the type of gauge, again. For a CEDV gauge, it really needs to go through some full cycles down to 7% in order to learn the new capacity. So as long as it does that occasionally, it will update the SoH, because that should not be changing very fast, the health of the battery.

For Impedance Track, it doesn't need as much of a change. It doesn't need a full charge/discharge cycle. So it can do this with only partial discharges. So that's one thing to consider when choosing which type of gauge you use.

OK, so one other question that's come in is, "The accuracy seems to be a function of the ability to test the chip at the top voltage. So how is the measurement error of less than 1 millivolt obtained?"

Yes, I think that 1-millivolt accuracy is the analog front-end accuracy we mentioned in one of our product page. If we want to look at the total solution, we probably want to answer that question by looking at different solution individually. I don't know whether this one milliamp is listed with specific device that this customer is looking at for 1 milliamp.

But in general, I think I'd defer to some of my experts, my expert colleague, which also sit in this room, Terry [? Sully, ?] to answer this question further.

Yeah, one comment. So the device is able to measure-- it has a limited accuracy if you're measuring a low voltage. Say if you only have a 1s battery. So you can measure that one. It can measure directly with less than 1-millivolt accuracy.

When you're measuring a higher-cell-count system-- say you have a 10s battery, and it's measuring top-of-stack divided down. Say each cell is 4 volts, so you've got 40 volts there-- you're dividing that down. And then when it gets divided down, it's effectively gauging it like it's one cell, but it's kind of be average of all 10 cells. And so what you end up with is you end up with less than 1 millivolt of accuracy, but at the equivalent one-cell level.

So that then translates in a 10s system, that's kind of like the top-of-stack accuracy, is going to be less than 10 millivolts. But when it translates it down, as it divides it down, the accuracy of that average cell that it's gauging will then be within the 1-millivolt level. Hopefully that's kind of clear. I know I rambled a bit.

But that's average.

Yeah.

The emphasis is that's average.

Because you're not measuring every cell, you're really getting the average of all the cells together. Then you're doing gauging on that value.

Thanks.

All right. Thank you, Xiaodong. Thank you, Terry. OK, one question that came in is someone said they're facing an issue with the cell voltage measurement in a monitor for 16 cells. "Two cells show a mirroring effect in measurement. So one cell voltage goes up, and another cell voltage goes down by the same amount. Is this a known issue?"

I think I can probably take this one. I think that it is common to, if you look at the pins of the device during cell balancing, you will see a type of mirroring waveform, as there's an internal FET that pulls or an external FET that pulls the [? double ?] [? fuses ?] together so that [INAUDIBLE] discharge can happen for the cell that's being balanced.

So a battery monitor should stop that balancing whenever it needs to make a measurement. And so it should not affect the cell voltage measurement. So if it is affecting the voltage measurement, there's something else going on.

One common issue is if the cell input resistors and capacitors which form an RC filter-- if they're quite large in value, they can have a long settling time. So if a battery monitor tries to disable cell balancing and quickly take a voltage measurement, if the filter is still settling, then you may see some effect due to that settling, that it's not really at its final value yet.

So that's one possible issue that could be happening. So that's probably the first thing to look at. I think if you search our E2E forum, you'll see a lot of different users who are trying to debug cell balancing. So I think that's probably a really good place to look if this doesn't describe your problem exactly.

OK, and one other question is, "What reliability issues need to be a concern for an AFE in the multicell battery pack system?"

Terry, maybe you can take this one. I think this will be covered in your session, probably.

Well, I meant so much reliability. I think the primary thing, number one, is to make sure that the ICs are used in their recommended operating conditions. We have those specified maximum and minimum voltages, currents, and so forth that the device can be-- are specified over. And with those, we have good reliability on the products.

We also have an absolute max spec in there that says, if you exceed recommended operating conditions but are within the abs max level, the parts should not be damaged, but it may not operate properly. So that may not be a reliability concern for you here. But ultimately, at least for the semiconductors, it almost always comes back to keep them in their recommended operating conditions.

OK, thank you, Terry. I do have one question that came in specifically on the BQ34Z100. I'm going to recommend Kevin, who asked that question, to possibly post this onto our E2E forum. I think, unfortunately, the experts we have on this panel are mostly experts on the monitor side, so a specific configuration becomes a-- on the BQ34Z100, we probably don't have an answer for you.

I will point to a few-- I think that's all the questions we have right now. I will point to a few reference designs that do use the top-of-stack and cell-by-cell gauging solutions.

So just a reminder that you will receive an email after the session with a link to the recording. And also, please answer the post-event survey that pops up after the end. We'd like your feedback, so we can continue to improve our content for future event. Thank you again for attending, and have a great rest of your day.

Thank you.