48V to 12V GAN HEMT based DC-DC Converter

Thread Starter

mohtishamali

Joined Jul 9, 2021
55
Greetings of the day to all,

I am working on 48V to 12V GAN based non isolated DC to DC converter with regulated output. I am very confused in selecting topology for my application. Can someone here who has already worked om similar sort of topic suggests me any leads to that, it would be highly appreciated.
Below are the basic parameters of my GAN based Converter.

Input Voltage: 30V to 65V

Nominal Input Voltage: 48V

Output Voltage: 9V to 17V

Nominal Output Voltage: 12V

Output Power: 1KW

Switching Frequency: 1 MHz
 

Papabravo

Joined Feb 24, 2006
21,159
The simplest strategy would be a buck converter. With a simple buck converter you should aim for having a nominal duty cycle of less than 50%.
30 VDC input and 17 VDC output would not be possible under those circumstances. If the simple buck converter topology does not satisfy the requirements there are steps you can take to mitigate that problem.

Regardless of anything else you do, understand that the power levels you are contemplating working with are dangerous and potentially lethal. This is no place for beginners in switch-mode power supply design and construction. Do not attempt this without access to help from a competent professional with experience in this area.

In particular you should be able to determine the ABSOLUTE MAXIMUM CURRENT you can handle with a copper trace on a PC board. Then you need to find an inductor with a core that won't saturate when it handles twice the anticipated output current. Even after you do that I'd want some margin for safety. Can you guarantee that?
 
Last edited:

Thread Starter

mohtishamali

Joined Jul 9, 2021
55
The simplest strategy would be a buck converter. With a simple buck converter you should aim for having a nominal duty cycle of less than 50%.
30 VDC input and 17 VDC output would not be possible under those circumstances. If the simple buck converter topology does not satisfy the requirements there are steps you can take to mitigate that problem.

Regardless of anything else you do, understand that the power levels you are contemplating working with are dangerous and potentially lethal. This is no place for beginners in switch-mode power supply design and construction. Do not attempt this without access to help from a competent professional with experience in this area.

In particular you should be able to determine the ABSOLUTE MAXIMUM CURRENT you can handle with a copper trace on a PC board. Then you need to find an inductor with a core that won't saturate when it handles twice the anticipated output current. Even after you do that I'd want some margin for safety. Can you guarantee that?
Thanks for the detailed and precise information but I don't have to work on Hardware for now. I just have to develop a spice model for my application. I already simulated different topologies but did not got favorable results
 

MrSalts

Joined Apr 2, 2020
2,767
Greetings of the day to all,

I am working on 48V to 12V GAN based non isolated DC to DC converter with regulated output. I am very confused in selecting topology for my application. Can someone here who has already worked om similar sort of topic suggests me any leads to that, it would be highly appreciated.
Below are the basic parameters of my GAN based Converter.

Input Voltage: 30V to 65V

Nominal Input Voltage: 48V

Output Voltage: 9V to 17V

Nominal Output Voltage: 12V

Output Power: 1KW

Switching Frequency: 1 MHz
The Vendors of GANFets should want to support such applications because that is what GANFets excel at (high power, high switching speed (small inductor) SMPS). Have you looked at GANFET suppliers' websites/application notes?
 

Papabravo

Joined Feb 24, 2006
21,159
Thanks for the detailed and precise information but I don't have to work on Hardware for now. I just have to develop a spice model for my application. I already simulated different topologies but did not got favorable results
Maybe I can help you with that. Can you make a zip file with your schematic including the symbols and subcircuit files I will need to run the simulations. I have a few GAN HEMT device modes and some circuits that use them. Primarily the GPI65015TO in a boost configuration.
 

Thread Starter

mohtishamali

Joined Jul 9, 2021
55
Maybe I can help you with that. Can you make a zip file with your schematic including the symbols and subcircuit files I will need to run the simulations. I have a few GAN HEMT device modes and some circuits that use them. Primarily the GPI65015TO in a boost configuration.
The reason i can see that my simulations are not running properly is due to the fact that PWM is not generated properly.
 

Thread Starter

mohtishamali

Joined Jul 9, 2021
55
The Vendors of GANFets should want to support such applications because that is what GANFets excel at (high power, high switching speed (small inductor) SMPS). Have you looked at GANFET suppliers' websites/application notes?
they have application notes but they seems to be very generic in my case
 

ronsimpson

Joined Oct 7, 2019
2,987
GPI65015TO
That is a 650V 15A device. Transistor
You have 65V max and 100A.
Look for a lower voltage part and a much higher current transistors. EPC2206 is an example. EPC2021 or EPC2022
The EPC parts are "low voltage". The resistance is 2m ohms. "90A" but will do 300A at a reduced duty cycle.

If you are doing a simple Buck, then you need a voltage above 65V. (80 to 100) and a current of 100A.
There are other options but they become more complicated. Transformers add a whole new level of complicated. I might consider a center tapped inductor. The current goes down but the voltage goes up.

I am turning on a board this week with a 100V part. Will see how it goes. I have never used low voltage GaN before. I have used 700 to 1500 GaN and SiC.
-------------edited-------------
There are many examples on the internet where people use transistors as diodes. In the case of MOSFET I have used them that way. In a Buck supply you will need a diode. In this case a real fast diode like Schottky. I would not use a GaN transistor for a diode. Looking at the EPC transistors they seem to make a poor diode. IMO Ron Simpson
 
Last edited:

Papabravo

Joined Feb 24, 2006
21,159
That is a 650V 15A device. Transistor
You have 65V max and 100A.
Look for a lower voltage part and a much higher current transistors. EPC2206 is an example. EPC2021 or EPC2022
The EPC parts are "low voltage". The resistance is 2m ohms. "90A" but will do 300A at a reduced duty cycle.

If you are doing a simple Buck, then you need a voltage above 65V. (80 to 100) and a current of 100A.
There are other options but they become more complicated. Transformers add a whole new level of complicated. I might consider a center tapped inductor. The current goes down but the voltage goes up.

I am turning on a board this week with a 100V part. Will see how it goes. I have never used low voltage GaN before. I have used 700 to 1500 GaN and SiC.
-------------edited-------------
There are many examples on the internet where people use transistors as diodes. In the case of MOSFET I have used them that way. In a Buck supply you will need a diode. In this case a real fast diode like Schottky. I would not use a GaN transistor for a diode. Looking at the EPC transistors they seem to make a poor diode. IMO Ron Simpson
I never meant to imply that the part was suitable for this design only that I had experimented with the model. I mentioned this before I knew that his problem was in simulation with generating a PWM "properly". We still don't know what "properly" means.
 
Top