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 11:26
Loaded: 0.00%
Stream Type LIVE
Remaining Time 11:26
 
1x
  • Chapters
  • descriptions off, selected
  • en (Main), selected

Hello, everyone. In this training video, we will cover TI's GaN portfolio and the differentiation of TI GaN power stage, which is called LMG3410x family of products. So in general, GaN has much better electrical characteristics such as higher electro mobility and wide bandgap energy. So these result in having high [INAUDIBLE] compared to silicon FETs, which effectively reduces the [INAUDIBLE] capacitance.

Since all the [? stray ?] capacitance is smaller in GaN, GaN FETs can switch much faster than silicon FETs. And this increased switching speed reduces the switching losses and enables higher switching frequencies that reduce the overall volume of the converters. So GaN-based converters can achieve twice the power density of best-in-class superjunction FETs.

One of the common questions is whether TI is in production with GaN, and the answer is yes. TI is in full production with its GaN portfolio, and many customers have already started using them in their designs. The GaN-based designs are not only compact and provide higher efficiency, but also achieves cost parity with a silicon-based design solution. So in this training, we will cover them all.

So the following couple of slides will talk about the differentiation of TI GaN. So first of all, TI does not make the [? discrete ?] FETs. All of our portfolio is called power stage where we integrate the driver with the smart protection features. So the integration provides us fast protection response, as well as very clean switching waveforms.

The second [? pillar ?] is the robustness. Our devices can protect themselves under short circuit and over-temperature conditions. In addition to this, the reliability under surge condition has always been one of the concerns of the industry, and we addressed this by testing our devices under 720 volts switching transient, similar to this stress observed during a lightning strike. And we have specified this in our data sheets.

Thirdly, we spend a lot of time to make sure our devices are reliable. Up to date, our devices have experienced more than 20 million of reliability hours and have less than one FIT rate for 10-year lifetime. And lastly, TI is in production with three of our devices, which comes in 50 milliohm, 70 milliohm, and 150 milliohms. So they are 100% TI-owned and uses TI internal high voltage GaN process, fab, and assembly.

This slide shows the block diagram of LMG3410x family. As I mentioned earlier, our GaN technology includes the GaN FET, but also the driver and the smart features. So first of all, integrated driver eliminates the [INAUDIBLE] inductances on the gate path, and also on the common source path, and allows switching with almost zero ringing. So in this graph on the right, the GaN switches from 0 to 400 volt with 100 volt per nanosecond slew rate with less than 25-volt ringing at the drain terminal. The power stage also integrates a very high speed overcurrent and short-circuit protection with less than 100 nanosecond response time.

Another smart feature is the R-Drive pin. So this pin mimics the gate current and allows us to change the slew rate between 30 volt per nanosecond and 100 volt per nanosecond on the hardswitching turn on edge.

We have also put time to optimize our daughter card to achieve the minimum cost. The backside of the board is covered with the heat sink. And on the front side, there is only one high side digital isolator. And the high side GaN FETs is powered by a bootstrap circuit. So the board only uses a 12 volt unregulated supply to provide the power supply for the high side.

So one of the questions that comes to mind is why does TI go with the integrated approach and how is it different than a discrete GaN FET with external drivers. So first of all, one of the most critical reliability part of the GaN is the driving voltage of the gate. So if gate is exposed to higher voltage then maximum allowed voltage, the lifetime of the device can be greatly reduced. Since we internally regulate the gate voltage from an unregulated 12 volt supply, it actually never sees a high voltage. So thus, integrated approach improves the reliability because all the gate voltage is regulated internally and is not exposed external to the transients of the circuit.

Secondly, the overcurrent protection is quite difficult with the discrete components due to the large saturation current of GaN and the small packages that we use for GaN. Those are low inductance packages. And our integrated protection feature is able to protect the devices even under harsher circuit condition because of the fast response time and high accuracy.

And lastly, the elimination of the loop inductances such as gate loop inductance and common source inductance allows us to switch much faster without any ringing. So this lowers the switching losses, allows you to go to higher DVDTs, at the same time increase the reliability because the ringing on the drain terminal is pretty small. And also, it reduces the generated EMI.

This slide is showing how the device behaves under extreme conditions. So the one on the left induces 720 peak voltage on a converter that is running from a 400 volt bus. And actually, the device voltage goes up to 720 volt on the fly. But as you can see, the device still continues to switch at this high bus voltage without any interruption of the operation.

And the graph on the right hand side is showing how the device behaves in a overcurrent event. So all the TI GaN power stage comes with two overcurrent protection schemes. We have two part numbers for these.

So this one, particularly, shows the case when a cycle-by-cycle protection scheme is selected. So once current threshold is exceeded, the overcurrent protection scheme in the circuit shuts down the GaN FET. And the driver waits PWM signal to go low before resetting the current protection. So, here, actually you're seeing the wave forms of cycle-by-cycle protection scheme. So every time the PWM goes low, the protection scheme restarts.

And we have another option, which is called latch overcurrent protection, where the PWM signal should be kept low for 300 microseconds to resume the operation. So these two options have different use cases in applications. And we offer two different parts for these.

This table here compares the TI GaN overprotection scheme versus the discrete implementation in terms of system performance. The most competitive discrete protection scheme is to sense the voltage across a shunt resistor and use a threshold to protect the device. However, this definitely lowers the system performance.

First of all, at high power levels, the high value of the shunt resistance would cause a lot of power loss. On the other hand, if it's too low, then it would be very difficult to measure the millivolts across it in a noisy environment. So in this example, we chose 12 milliohm resistors, two of them in parallel. So effectively, it's 6 milliohm resistance.

And if you look at the added PCB area, including the comparator, it's about 233 millimeters squared. So because of the addition of the shunt resistor in the power loop, we increase the power loop. So the power loop, the increase is about 1.2 nanohenry.

So under this circumstance, you can no longer switch at 100 volts per nanosecond with the same overshoot. So you have to slow your device down. So if you slow your device down, for instance to 80 volt per nanosecond to get the same ringing across the drain of the device, you will increase your IV crossover switching losses.

So because of the conduction loss on the shunt resistor as well as increased switching loss, the total loss increases by 0.9 watts. So as a result, you increase the BOM cost and you increase your space and you increase the losses in a discrete implementation. And yet it may not be as protective as an integrated approach.

I also want to quickly touch base on the reliability of TI GaN FETs. So TI's GaN devices have over 20 million device hours with less than 1 FIT rate for 10 year lifetime. The JEDEC standards, all the qualifications hardswitching tests, dynamic RDS-on tests, long term in-application, and extreme condition tests are all completed. And TI is in mass production with these devices for about a year now.

With three products in LMG3410 family, we can cover a wide range of applications and power levels ranging from 65 watts all the way up to 10 kilowatts in a multi-level approach. And on the low power side, our 150 milliohm GaN FET is well suited for wall adapters, USB PD chargers in digital TVs. As the power goes higher than 500 watt, we can cover a wide range of applications with our three devices on the portfolio. These would be the PFC and DCDC stages of server telecom and 5G applications.

At higher power end, our portfolio covers single phase inverters and three phase multidrives. For higher voltage and higher power levels greater than 5 kilowatt, our 17 milliohm and 50 milliohm FETs can be used in a multi-level approach. This would enable to use them in three phase high voltage grid tight applications, like PV [INAUDIBLE].

So this is the end of the section. And I hope it gave you an idea of how TI GaN is different from the discrete FETs and what the applications are that can be covered with the TI portfolio. And thanks for watching this video.