UCC21750-Q1: High Side Power Supply Design

Hi,

Thanks for your questions.

Harsh Songara said:

To see if 2W is convenient for the high-side, I used a reference from TI to calculate the same attached here. For the quantity P(driver), I used the Power Dissipation (Pq+Psw) from the UCC217xx XL Calculator in the product folder based on my design parameters. Is this a convenient approach to design the supply?

I would say that this is a better way to estimate the UCC21750-Q1 thermals, since its only going to estimate how much power is dissipated in the gate driver itself but not outside.

The better way that is outlined in the App Note you have linked, is more considerate of full-system (and is similar to how we calculate power diss for individual driver)

Be sure to add the Quiescent power consumption, any other circuits consuming power and to ensure multiply by # of drivers in your system. Also be sure to use VDD-VEE for deltaGate if you are using a bipolar supply.

I've screen capped it below.

The highlighted RGE / CGE is typically only used if you use an external BJT current buffer, this is used to adjust STO turn-off, you probably would not use such a filter without a current buffer, so I'd ignore that section.

Harsh Songara said:

2. As a part of the learning process, could you please explain on why Rg does not affect the power Supply Requirement?

Rg (external gate resistor), does affect the power disspation of the driver itself, but not really of the total power consumption.

From the perspective of the system, we dont really care how much power each individual component is going to burn. It basically boils down to P=I*V

I=Q/dt or ~ Q*f. So basically, if we know the gate charge of the Mosfet/IGBT (which itself is an based on a sepcific VGS/VGE, so more of an estimate).

Gate driver is fully charging the gate each cycle, if its not, theres a problem! If we know gate charge, and often we are charging it, and how much we change the voltage at that gate cap (VDD-VEE), you can get the power dissipation. So lets say you have Rg_ext =5Ohm vs 6 Ohm, and youre operating at 5 ohm. Even though the PEAK current will be changed by Rg_ext, the gate driver is still going to move the same amount of charge to the IGBT gate, just over a bit more time.

So there is no impact of Rg_ext on the system power diss.

Harsh Songara said:

3. I'm currently also considering the use of the ISO5452-Q1 with a BJT totempole current buffer to increase the drive strength. I would like to know if the Power Supply demands are affected by the use of the current buffer. If they do change, how do I approach the same?

You would need to consider that the BJT also requires some base current in order to charge the IGBT/MOSFET gate at a specified current, so i would also multiply by a factor of your NPN (top) "alpha" parameter into the equation underlined in red, since the IC is going to be slightly higher than IE, you can account for it if you wish.

Harsh Songara said:

4. In the UCC21750-Q1 datasheet, the power dissipation under the Table 6.5 Power Ratings is the worst-case power dissipation that the device can safely handle based on the constraints of Ambient Temperature and the Thermal Resistances of the device right?

Yes, at those conditions this is true. But the ambient temperature is not always 25C, and board temp is not always 25C either. The Excel calculator provides a better estimate of thermal performance since it can take into account the board temp.

Please let me know if I've answered your qustion by pressing the green button. And don't hesistate to let me know if you have any other questions on this topic, or further topics while youre moving forward with your work on UCC21750-Q1 / UCC5452-Q1!

Best

Dimitri

Hi,

Thanks for your questions.

Harsh Songara said:

To see if 2W is convenient for the high-side, I used a reference from TI to calculate the same attached here. For the quantity P(driver), I used the Power Dissipation (Pq+Psw) from the UCC217xx XL Calculator in the product folder based on my design parameters. Is this a convenient approach to design the supply?

I would say that this is a better way to estimate the UCC21750-Q1 thermals, since its only going to estimate how much power is dissipated in the gate driver itself but not outside.

The better way that is outlined in the App Note you have linked, is more considerate of full-system (and is similar to how we calculate power diss for individual driver)

Be sure to add the Quiescent power consumption, any other circuits consuming power and to ensure multiply by # of drivers in your system. Also be sure to use VDD-VEE for deltaGate if you are using a bipolar supply.

I've screen capped it below.

The highlighted RGE / CGE is typically only used if you use an external BJT current buffer, this is used to adjust STO turn-off, you probably would not use such a filter without a current buffer, so I'd ignore that section.

Harsh Songara said:

2. As a part of the learning process, could you please explain on why Rg does not affect the power Supply Requirement?

Rg (external gate resistor), does affect the power disspation of the driver itself, but not really of the total power consumption.

From the perspective of the system, we dont really care how much power each individual component is going to burn. It basically boils down to P=I*V

I=Q/dt or ~ Q*f. So basically, if we know the gate charge of the Mosfet/IGBT (which itself is an based on a sepcific VGS/VGE, so more of an estimate).

Gate driver is fully charging the gate each cycle, if its not, theres a problem! If we know gate charge, and often we are charging it, and how much we change the voltage at that gate cap (VDD-VEE), you can get the power dissipation. So lets say you have Rg_ext =5Ohm vs 6 Ohm, and youre operating at 5 ohm. Even though the PEAK current will be changed by Rg_ext, the gate driver is still going to move the same amount of charge to the IGBT gate, just over a bit more time.

So there is no impact of Rg_ext on the system power diss.

Harsh Songara said:

3. I'm currently also considering the use of the ISO5452-Q1 with a BJT totempole current buffer to increase the drive strength. I would like to know if the Power Supply demands are affected by the use of the current buffer. If they do change, how do I approach the same?

You would need to consider that the BJT also requires some base current in order to charge the IGBT/MOSFET gate at a specified current, so i would also multiply by a factor of your NPN (top) "alpha" parameter into the equation underlined in red, since the IC is going to be slightly higher than IE, you can account for it if you wish.

Harsh Songara said:

4. In the UCC21750-Q1 datasheet, the power dissipation under the Table 6.5 Power Ratings is the worst-case power dissipation that the device can safely handle based on the constraints of Ambient Temperature and the Thermal Resistances of the device right?

Yes, at those conditions this is true. But the ambient temperature is not always 25C, and board temp is not always 25C either. The Excel calculator provides a better estimate of thermal performance since it can take into account the board temp.

Please let me know if I've answered your qustion by pressing the green button. And don't hesistate to let me know if you have any other questions on this topic, or further topics while youre moving forward with your work on UCC21750-Q1 / UCC5452-Q1!

Best

Dimitri

Hi Harsh,

Harsh Songara said:

1. With reference to red underlined term in the screen capture attached.. Just to confirm my understanding, does that represent the total power to be supplied to the Gate Driver High Side System for the purposes of Switching, irrespective of whatever components are used to perform such an activity?

Yes. Things external resistor in the drive path will NOT make a difference in the estimate, because the final charge is exactly the same regardless of the value (unless you make it WAY too large), which is all that matters. The only thing that would make a difference is adding additional CAPACITANCE, such as shown in the photo. Lets say if you have a 1nF cap in // with your IGBT gate you also need to account for it.

Harsh Songara said:

3. So to say, if I were to ignore Alpha Current Gain, it shall not make a huge difference. Hence, another reason why a system based approach to calculating the net power requirement does not require individual device details (say Rg, BJT specs). That is, I can use the same calculation methodology.

Yes, and should be 1/alpha as a factor, but youre right with high enough betait is ~1 anyways. its a similar concept to how how you can estimate power consumption for even a digital IC, just total gate cap, frequency, vdd is all thats needed. Of course you should be sure to include headroom in the power budget when speccing the power supply

Have a good weekened, let me know if you have any other questions. If i've answered your question, please let me know by pressing the green button.

Best

Dimitri