OK. It appears we are ready to go. So welcome to the High Voltage Seminar. This is the "Getting Started with Programming Functional Safety Compliant Gate Drivers for EV/HEV Traction Inverters" session being presented by Gangyao Wang. My name is Will Hadden, and I'll be the moderator for this session.

All participants are muted for this session. Please use the chat function to ask a question and address it to everyone. We'll be answering questions throughout the webinar in chat. And also chat if you're having any problems hearing or seeing the presentation. With that, I will hand it off to Gangyao to get started.

Thank you, Will. Can you hear me well? Is audio OK? Good?

Yes, audio is OK.

OK. All right. Hi. Good morning, everyone. Thank you for attending this seminar. So my topic is about getting started with the programming function safety compliant gate driver for EV/HEV traction inverters. So here's the agenda.

First, I will give you overview about TI part we have released recently, which is UCC5870-Q1. Then I will go step by step about how to program this device, which includes different modes of operation of this device; SPI introduction and SPI commands; different modes of SPI; and how to write and read a register; the frame timings for daisy chain and address mode; general device setup and how to enable the driver to active mode; then what we can program and change in active mode; and last, I will talk about how to exit active mode and different ADC sampling mode, setup, and reading results. So and then I will give a demo of one of the EVM to show how to program this driver with a GUI.

So first, this device is a 36-pin device with a lot of integrated functions. This one can supply minimum 15 ampere source and sink current. And this one has real-time SPI communication. This one is SafeTI ISO-2626 system compliant in terms of integrated data protection, integrated diagnostics, and integrated supply monitors, and the built-in self-test.

So SPI-enabled configuration can do three different modes-- regular SPI, daisy chain, or address-mode SPI. This one has DESAT short circuit overcurrent protection. This one also has an integrated thermal diode or NTC-based temperature measurement and protection. In terms of short circuit protection, this one has soft turn-off after a [INAUDIBLE] is detected, or a two-layer turn-off. This one has integrated six-channel ADC, which can be used to sense temperature, DC back voltage, or phase voltage, or something else.

This one has ASC support, which has both primary side and a secondary [INAUDIBLE] ASC function gives customer a lot of flexibility. This one has integrated configuration dead-time and programmable internal or external Miller clamp. This one also has advanced VCE clamping function. So basically, this one is a one-stop shop gate driver for every EV/HEV traction inverter system. And it will eliminate as much as possible external circuit driving high-power modules.

So this is pretty much the introduction about our device. So since this one is SPI-enabled, the programming about this device is not some simple driver you can use directly. This one requires some steps to compute and move the driver to the active state in order to output the PWM.

So I'll go through step by step about this process. For this driver, there are four modes of operation. So during the power-on, there's a reset mode. In this mode, the internal regulators will [INAUDIBLE] powering up. BIST is running, and nFLT is held low until regulators are ready, like all the power supplies are ready, and internal BIST are completed without failure.

Then it will move to next step automatically, which is configuration 1 state. So what we do in this configuration 1 state is either we write the chip address, because this one supports the address-mode SPI, so we have to assign address for every driver. For example, for three-phase [INAUDIBLE] six drivers, so we have to assign address for these six drivers separately, for example 1 to 6.

This one can support up to 14 drivers in parallel. [INAUDIBLE] to address bit is 4 bit except all 0's and all 1's. So we have 14 addresses can be assigned to any drivers. So in configuration 1 state is where we assign this address to these drivers.

Then we will move to the configuration 2 stage. configuration 2 state is where we program all the registers, the settings. For example, some DESAT threshold or ADC signal, which channels you're reading. And also, we can program which function to enable or disable. For example, gate monitoring or DESAT fault [INAUDIBLE] after a DESAT protection that's going to be used [INAUDIBLE] turn-off or two-layer turn-off. So basically, all resistors, configuration resistors, can be programmed in configuration 2 state.

So after configuration 2 state, we can move to the active state, where the driver will respond to the input PWM and output the same PWM. So in this stage, only a few registers can be programmable. I will talk more details later. Basically, some control register and some on-the-fly [INAUDIBLE] change register.

So between configuration 2 state and active state, these two can move between. We can go to either from configuration 2 to active, or from active to configuration 2. But we either require power cycling in order to go back to the reset state. So once this one is powered up, or once-- after the [INAUDIBLE] address [INAUDIBLE] the configuration 2 state, then we cannot-- we will not be able to move back until you cycle the power supply.

So first, I will talk about the SPI configuration for this power device. We support four [INAUDIBLE] modes of SPI-- regular SPI, which requires chip select signal for every driver. So for a three-phase system, as you can see on the right side, the first one is conventional regular SPI, with chip select for each driver, either requires 9 I/Os for a three-phase system-- 9 I/O from MCU part.

For the address-mode is, we parallel for the SPI command line. So it only requires 4 I/Os. The conventional daisy chain mode is the same also for I/Os, but they are-- the connections is by the daisy chain mode. The address mode is they're all-- they're just all parallel, all of them. So from the hardware configuration point of view, this one will save I/Os. Then compared with the conventional daisy chain mode, this one will have faster reading and writing speed.

So for the SPI commands, there are a bunch of them, which includes driver enable, which involves-- driver enable and disable, which moves the driver between the state of configuration 2 and active state. And this one has command for reading data, start configuring, which is moving form configuration state 1 to 2. And the NOP command, basically this one sends nothing but just is a clock to read the SPI data.

This has a software reset data, which will reset the device to the default status. This just resets all the registers. So there's a writing command for the high data bits, and there's a writing command, WRL, for the low data bits over the 16 bit registers. There's a writing command, and there's a-- last one is the writing chip address, because for address mode, we have the SPI address in configuration 1 mode. So this is a command to do that.

As you can see, for most of the commands it has the fourth bit as chip address, because it's a driver-specific command, the last one, when we have all the address 1, and also we have-- when the driver is in configuration 1 state, we can write address, the last 4 bits will be the address it assigned to a [INAUDIBLE] driver.

So for the how to do the daisy chain mode, if we prefer to use a daisy chain mode, the then use 0 for the address for every command. And we can see address, we have like 4 bits from 0 to four Fs, which is 15. So all 0's is reserved for the daisy chain mode. Four Fs is reserved for the [INAUDIBLE] anything between 1 to 14 can be assigned to a chip.

So for the daisy chain mode, use 0 for the address for every command. No address-- addressing is needed [INAUDIBLE] directly from reset state to configuration 2 state, nothing to do for the configuration 1. The device must receive configuration in command one by one in order to enter the daisy chain mode and the SDO passthrough.

So basically, from configuration 2 and configuration 2, that requires a command in-- CFG IN command that will send to all the six drivers when they are connected in daisy chain mode. Now for the address mode. The four lines for SPI are in parallel. So this address occurs in configuration 1 state. So without stacking this address, it should not go to the configuration 2 mode, because once in configuration mode, it requires a power recycle in order to get back to configuration 1 state.

So if address mode will be used, this address assignment should be done in configuration 1 state. So how to do this one? Basically, if we want to write the address for a specific gate driver, we have to pull the in-plus, the input PWM, for one specific drive high, then we know we are going to write the address for this driver.

The pulled in-plus high-- I mean, there are one device each time. Just put PWM high for one device, then write the WR CA command, which is below 1. And the last four-- bit would be the address assigned to this driver. So we do this one by one for all the six drivers. Like, we put up PWM for first one, we write an address, then we do the second one. Once all the six is done, then the addresses-- writing address step is complicated-- it is completed.

So for the address F, which is all 1's, it's broadcast command, all devices will respond to this one. The 0 address should not be used for the address mode, because this one is reserved for the daisy chain and standard operation. And each device should have a unique address. So you cannot write the same address for two devices. For the broadcasting case, never do a read command, because this one is not going to work. Again, broadcasting is writing, but we cannot broadcast the reading.

So how to write-- writing a register? So the configuration register is 16 bits. So we write high 8 bits and low 8 bits separately. There's a writing command WR RA, with address information, with the chip address information and the register address information. So first, we send the WR RA command. Then we send a WRH or RL command with data.

So then this writing will be completed. If only some bits of one register is changed-- like for example, only like high 8 bits changed, it is OK to only write high 8 bits. It doesn't have to read the whole register every time. Only to write half of it is OK.

So this is about writing a register. For reading register, we can-- also, it's also required to send a read data command, which includes ship address and register address. So after reading command, SPI signal should-- SDO signal should be available for the next command. If there is no other command, the controller should send a NOP command in order to clock SDO out from the driver. So for reading register, make sure you don't use the broadcast command. Like, don't use 01 for the address.

So how to compare the daisy chain mode and address mode SPI? How they [INAUDIBLE] different when doing the framing timings? So for daisy chain mode, you have to put the [INAUDIBLE] command for all the six devices together, from 1 to 6. And first one you send a read command, and the second one you will get the data back. So the total time will be like 49.5 microseconds.

For the address mode, you can do reading one by one. So send one command for one specific driver with a specific chip address. It will give-- So this command will be only like 4.1 microseconds, not like this, that you only have to put all six together.

So you would read the data back for the second command. You add them together, So in total, you need seven commands, which adds up to like 30 microseconds. So for reading a single 6-bit register, the timing has been reduced to about 40% between daisy chain mode and address mode.

More importantly, this address mode allows you random access to any driver. You can read any driver saved by sending only two commands-- one read command, and second one, which will read the data back. But you don't have such an option for daisy chain mode. You have to go through all the six devices in order to read the data for a specific driver that you would like to read.

So this is for the reading one. Now for writing one, for writing a single 6-bit register of all six devices, for daisy chain mode, you have to send three commands in group. So each one lasts like 24.6 microseconds. So total would be like 74.6 microseconds for writing a 6-bit register of all six devices.

But for address mode, you can do a broadcast command. If you are writing same register, same value, to all gate drivers, you would do broadcasting and then use 01 address for the writing command. So this will almost save 80% of the time when doing the writing programming, when selecting the broadcasting mode.

So there's a big difference. In order to combine all the time for reading all STATUS registers-- we have total of five STATUS registers-- a daisy chain would take 150 microseconds, address mode 134. With all configuration registers, both probably about 300-- more than 300 microseconds. Write all configuration registers, address mode will be much faster.

Because when we are reading everything-- so for even for daisy chain mode, they can combine together, but we have to go through everything anyway. But the advantage of address mode is random access. For example, if you [INAUDIBLE] one driver report the-- [INAUDIBLE] report the four that you want to just read that gate driver, it will give you much less time than the daisy chain mode. So that's a benefit of the address mode compared to the daisy chain mode-- random access and faster writing.

There are some general device setups about this device. So that star fault P bits are used to mask faults when fault P bit is set to mask fault, the STATUS bit will set be during the fault. However, the fault will be ignored. And the nFLT pin will not trigger. So basically, you can kind of disable some fault reporting.

That fault state bit are used to set the reaction to a fault. So generally, there will be different-- there will be a few options like Pulled Low, No Action. Some has a High Impedance or even some Pulled High options. So when there's a fault, you can select the reaction to that [INAUDIBLE] fault.

For example, like a DESAT particular, you want to do a soft turn-off for like a VCE overshoot, you probably want to just select a No Action that's in the-- so the driver we're currently running, but will report fault bit to the controller. So you have the options to select the direction of each fault for this device, which is quite flexible.

Enable and disable bits are used to enable or disable some function. When a function is disabled, for example, if DESAT disabled [INAUDIBLE], no fault would be indicated.

So write to any reserved bit. There are few bits are not used-- we will call these reserved. If you're writing any reserved bits, may result in a CRC fault. Writing a SPI test register also results in a CRC fault.

So next command. Once all the registers are programmed in configuration 2 state, then we send a driver enable command. This can be sent individually or with a broadcast command, which is like enable all the six drivers. So once the driver enable is sent, the CRC for the configuration register is calculated and stored.

Any bits lost to the configuration register will result in a CRC fault. So [INAUDIBLE] able to make any changes once this one moves to the active state. So but there are only a few exceptions for-- we'll talk about this one in the next slide. But this is the driver enable command, which we move driver from configuration 2 state to active state.

So for SPI communication in active state, there are a few things we can write. Like, in active state, all registers are readable. And also, there are a few registers that can be written, which is control 2, clear fault bit, [INAUDIBLE] 8, IOUT SEL-- selector bits. This one are used for control the output driver strength. We can program from 1 to 1.6. And also, we have CRC disable command. We can write this bit in active mode.

So when we want to change the driver [INAUDIBLE] on the fly by programming CFG 8 IOUT selector bit, you have to follow these four steps. First, we need to disable-- to set CRC disable bit to 1, then change the IOUT selector bit to select the required strength, then write the CRC disable bit back to 0. So if we don't do the CRC disable bit to 1 and then to 0, this step will result in a CRC fault.

The control1 and control2 registers are all used to perform diagnostics for some functions. So to use these registers in active mode, the CRC disable bit must be written to 0, similar to the previous [INAUDIBLE] gate driver strength change case. So in general, these diagnostics are only used to during a key on cycle. So write the CRC disable bit once, go through all the diagnostics, then write the CRC disable bit back to 0, and proceed with the normal operation.

Exiting active mode. So active mode is exited in one of two ways. First, writing a driver disable, which will turn off the driver and leave the resistor intact. It will move back to the configuration 2 state. Second is write the software reset command to turn off the driver and reset all the registers to the defaults. Now, this doesn't change the chip address, though. So the difference in driver disable and software reset is disable doesn't change the register values. Reset will change all the register [INAUDIBLE] back to the default state.

So after this one has been moved to the configuration 2 state, it requires-- so drive disable and software reset will move the driver out of active state. But in order to get back to the configuration 2 state, it does still need another configuration in command to enable the normal read/write access. So first one is the driver disable command, and then the software reset command.

Lastly, I will talk a little bit about the ADC sampling modes. So for this device, we have three sampling modes we call center sampling mode, edge sampling mode, and hybrid sampling mode. So for the center sampling mode, is we only do sampling during-- in the middle of one PWM cycle. You can select the on cycle or off cycle, or both. So we will calculate the PWM and do the same thing in the middle of this PWM.

The edge sampling mode is, we put some-- we [INAUDIBLE] transition at low to high or high to low, and do some delay here-- the delay is programmable-- and then start sampling the signals, whichever channels are enabled. We can do both during the on time, PWM on sampling or off sampling. And you can select which one you want to sampling-- run the sample during the on time, or [INAUDIBLE] you want to sample it during the off time. Or you can sample in both, essentially during both on and off PWM cycle.

So hybrid mode is a combination of these two. So it controls-- so it will have-- it makes some judgment about the pulse width and determine some optimized way to do the sampling. So these modes are trying to avoid sampling during the PWM edge, which may have more system noise, so in order to get a better, clean result. That's the motivation about this design.

So for the ADC setup, we have to set up-- because we have eight ADC channels plus one die temperature, so total we have like seven ADC registers can be read. We have this setup, which channels can be read and which part of PWM cycle we want to read.

So that bit is called ADC enable, which enables the ADC function. They are like ADC sampling modes. I mentioned there's like three sampling modes-- edge mode, center mode, or hybrid mode. For the edge and hybrid mode, we can select the delay time. There are a few options. I can show later when I do the demo.

So that's the V reference selector bit, which is used to select the internal ADC reference or external ADC reference. So internally, we have 4-volt reference for the ADC. But this internal reference is like 1.1% accuracy. So if you want to use some very high accuracy, for example 0.1% reference, you can connect an external one. And we have option to select. So all the ADC results are stored in the ADC data register.

And there's a timestamp for every ADC register. So this one, time stamp, has-- will change for every sampling. There are a total of 64 different values. So if a timestamp has not been changed, we can assume this is the same data. This data has not been updated. So it's not a new sampling. If this one has been changed, we can say this one is new data.

So unless there's some extreme case-- so sampling happens at exactly 63 PWM cycles. That case will be the same. But it's a new sampling. So this is-- it's a very rare case. We have equation to get the value of the data when using an internal VREF 4 volts. So for the AI pin voltage, it's VADC in decimal times 3.519 millivolts. For die temperature, we have this equation to get the internal die temperature for the ADC part.

So that's my introduction about how to program this part. So next part, I will give a demo about how to program this driver with-- through a half-bridge board. As you can see, this is a half-bridge EVM. It's available from our TI website. And there's also a quick start demo video available on TI website.

So this will help you to get from this device very quickly to save you time, because it requires some effort to write programs. So this one is ready to use. So you can play with the [INAUDIBLE] driver and your hardware to do some switching tests, double pulse tests.

This driver-- this board is designed to fit one of the infinite power modules, half-bridge modules. So I will move forward to this demo. So this is a hardware picture. So when we do demo, we just need two power supplies for the connection. And we need this USB to be to the computer and one oscilloscope. The steps are to plug in MCU board, connect the MCU board to the USB to the computer, and then enable two power supplies, and then connect the oscilloscope probes. And then you can do the measurements.

So let me switch the camera. I want to show this one to this demo. [INAUDIBLE]. So I have this one, the EVM. I have two power supplies here. And I have an oscilloscope. Let me just turn on the power supply, and another power supply, and turn on this sampling [INAUDIBLE] oscilloscope.

And we have a GUI program. Let me show the GUI program. So I have started this GUI program. I have done the connection for the USB. Now we can see this hardware connected.

There are sometimes when it will show this one is not connected. You have to switch accounts, and then it will-- it should work. So I can see this one, hardware is connected. Then I click this, Explore Device Modes.

There are three options for the SPI-- regular SPI, daisy chain SPI, and address mode SPI. So select one of these. Because my this-- MCU board setting is for regular step, so I just select the regular SPI, click Finish. So then you'll be in configuration 1 state, because we are now using regular SPI, so we are not doing any address writing here. So we just go ahead and program one of the drivers.

We have high side and low side, because this is a half-bridge. So let me-- I need to show-- OK. I need to go on my camera here and then show this [INAUDIBLE] here. Hold on one second. Yes, here.

So that's the hardware. Here's the GUI program. Let me get this [INAUDIBLE] more over here. So as you can see, this one is in configuration 1 state right now. So we're not going to program the address. We just move to the configuration 2 state.

OK. So in configuration 2 state, now we've got the register map. We need do a read command to read all the registers. So this program will be updated about all the default values here. And you can see here this writing, these are changes. So there are total, I think, 31 registers. So we have to read all of them.

Now we go back to here. We can see a bunch of things we can program. This one is for two-level turn-off, ADC configuration, UVL settings-- like for 5 volts, 3.3 volts, a bunch of things. I would like to go ahead to-- because my EVM board doesn't connect any power device, so after disabling DESAT function, otherwise it will report a fault.

And on this EVM board, I have a load 124. So this one will trigger bit [INAUDIBLE] fault by default. So I disable this one for now. And I also changed the default output to Pulled Low.

So I just changed-- I disabled this one and changed the gate [INAUDIBLE] monitoring. Then I can move to active state. Now, in active state I can generate a continuous PWM. I can also-- like for example, just maybe 50 femtosecond and whatever frequency. But we can see we have fault states. You can also see this red light is on.

We can take a close look at what this fault is. We can read all the registers and go back to the STATUS register and find out which fault this one is. So if it's 1, nothing; 2, nothing. So we can see, you have UVLO2 fault. Why there's this fault?

I think this is because when I power up this device, I power up the VCC 1 first, and then the secondary VCC 2 is not available. So that will give you a fault here. We can clear this fault by sending a command. So now it's good to go, as you can see.

This is the low side. Let me also clean the fault for the low side. So we can go back to-- clean the fault for high side. We can go back to high side. We can also move to configuration 2 state. I can do the same thing with all registers. So I can also move this onto active state and clear this fault.

As you can see, there's two are the red light is not shown here. So if I move back to the low side, I have this option to get a test function. So basically, this one would generate a double pulse. So we can program the on time, for example t1 10; t2, 5 microseconds; t3, 5 microseconds. Then we hit this Test.

Now you can see from this oscilloscope, you can see this double pulse waveform. First pulse, 10 microseconds, then 5 microseconds off, and another 5 microseconds for the second pulse. So this one shows the double pulse test of this driver.

Now, how about if we do a DESAT fault? For example, when I go back to the configuration state, let me enable the DESAT function. So this one was disabled previously. Now I just enable this one, DESAT detection and the DESAT current enable.

So move to active state again, now do same double pulse test. We can see this one is protected. We don't see a double pulse, we just see a single pulse. And this one is protected. And the driver should report the fault. This one reported a ping.

Now when you go back to the register map to see exactly which fault, so that's just register-- STATUS register 3. We can see that a DESAT fault has been reported. Now I want to show a DESAT fault with [INAUDIBLE].

So I clear this fault. I go back to configuration 2 state. with this, DESAT is enabled. I select the [INAUDIBLE] for the DESAT fault. So I select self turn-off short circuit and DESAT. So I can select different soft turn-off current, 0.3 to 1.2. So I just select the 0.3.

Now I move to active state, I do another double pulse. So I hit Test here. Now I can see this waveform will sense the DESAT fault, but turn off slowly with a constant current 0.3 amperes. So this is a DESAT with soft turn-off.

And this also-- this one also has functions for short circuit test. So for either high side or low side, different combinations. This one also can generate the continuous PWM. Let me just give it a try. Oh, this one is DESAT protected, because I didn't disable that one.

So let me go back. If I disable this one, I should be able to do a continuous test. So you can see this is a continuous waveform for 10 kilohertz. I think we can do 20 kilohertz. So I think this is quite flexible about the GUI, how to use this one, this program-- low side, you can do the high side.

So let me see if any questions about this demo for-- so I think this is-- this one also has the self [INAUDIBLE] if we move this device back to-- somewhere between default-- default and configuration 2 state. And these are register functions, which we mentioned. We can do the different IOUT selector, like it's all segments, or 1/3, or 1/6. So these are the options to change driver strength on the fly.

Yes. So now, I think this is for the half-bridge demo. Now let's move back to my presentation. So by the way, this software is available for download. And there's a quick start demo video which also just goes through the steps about how to using the GUI program and how to initiate the driver. And these are all available from ti.com

We also have a three-phase board, [INAUDIBLE] three-phase EVM. So the purpose of this one is more for debugging software for three-phase configuration. So this one supports both address-mode SPI or daisy-chain mode SPI. This one can fit the TI launchpad directly so it's easier. It's very easy to do the programming, we are also having-- have some example code. We can initiate this program to writing all the registers to set up a device address.

We will make this one available sometime later. So it's easy, very easy, to get a start with programming these devices, and to test, to explore different features of this driver. So I think that's the end of my presentation. So I'm open for any questions about this demo or about this topic.

So any questions from the audience?

I do not see any questions in the chat. So yes. So thanks, Gangyao, very much. And thank you, everybody, for joining us. All of these session recordings and presentations will be available to view later this week on ti.com/highvoltageseminar.

You'll also receive an email with links to an on-demand presentation and a post-event survey. We really appreciate your feedback so that we can continue to improve our content for future seminars. Thank you again. And have a great rest of your day.

Thank you, everyone.