Can I use PFET Buck Controller IC with External Half Bridge MOSFET Driver?

Thread Starter

electromeow

Joined Oct 12, 2021
9
Hi,
I am trying to make a synchronous buck converter with 5V/6A output, and I'm trying to reach high efficiency with minimum cost. So I want to use a buck converter IC with an half-bridge MOSFET driver and 2 MOSFETs.
The MOSFET driver I plan to use is TPS28225. I am in search of a buck converter controller IC. I found LM3475 is cheap and its control method looks interesting. But this is designed to drive a P-Channel MOSFET, so I'm not sure if it will work with my MOSFET driver.
Also LM3475 and TPS28225's supply voltage ranges isn't enough for my project, which is going to be supplied with a 4S Li-ion battery(up to 16.8V). Can I use a 5V LDO to supply LM3475 and TPS28225 but switch higher voltages with MOSFETs?
Thanks in advance...
 

Papabravo

Joined Feb 24, 2006
21,159
You want to do a synchronous buck converter and you want to use PMOS FETs for both switches, or only one of the switches? IMHO this is a bad idea because PMOS devices are slower which leads to higher switching losses. IMHO you should rethink this design choice and select a suitable pair of NMOS devices. Then deal with selecting and minimizing the deadtime between the main switch and the diode replacement switch. In addition I would get a beefy Schottky diode to place in parallel with the body diode on the diode replacement switch.

The following simulation is a basic open-loop 12V to 5V synchronous buck converter. It does require an isolated gate driver for M1, and such a driver works fine for M2 as well. 95.26% is not too shabby. D1 is the parallel Schottky and it conducts while M2 is turning on to reduce the switching losses. The voltage ripple for the open loop converter is approximately ±25 mV.

1659120359753.png
 

Thread Starter

electromeow

Joined Oct 12, 2021
9
You want to do a synchronous buck converter and you want to use PMOS FETs for both switches, or only one of the switches? IMHO this is a bad idea because PMOS devices are slower which leads to higher switching losses. IMHO you should rethink this design choice and select a suitable pair of NMOS devices. Then deal with selecting and minimizing the deadtime between the main switch and the diode replacement switch. In addition I would get a beefy Schottky diode to place in parallel with the body diode on the diode replacement switch.

The following simulation is a basic open-loop 12V to 5V synchronous buck converter. It does require an isolated gate driver for M1, and such a driver works fine for M2 as well. 95.26% is not too shabby. D1 is the parallel Schottky and it conducts while M2 is turning on to reduce the switching losses. The voltage ripple for the open loop converter is approximately ±25 mV.

View attachment 272518
I suppose there is a misunderstanding here. I don't want to use P-Channel MOSFETs. I want to use 2 N-Channel MOSFETs with an half-bridge driver. But the buck controller I want to use is designed to use with P channel MOSFETs. I want to use a P-Ch MOSFET buck controller to switch an half-bridge MOSFET pair consisting of 2 N-Ch MOSFETs.
 

Alec_t

Joined Sep 17, 2013
14,280
Can I use a 5V LDO to supply LM3475 and TPS28225 but switch higher voltages with MOSFETs?
Should be doable, but you may need a logic inverting stage between the 3475 and the 28225 (I haven't checked the 28225's input logic requirements). Use of non-optimum ICs counts against efficiency though.
 

Ian0

Joined Aug 7, 2020
9,671
What makes you think synchronous will be more efficient?
It will, provided that the current never reaches zero.
At low loads where a MOSFET/diode converter goes into discontinuous mode, a MOSFET half-bridge must remain in continuous conduction so the current goes below zero for part of the cycle so that it is flowing back from the output capacitor into the input. That's not efficient.
ICs that are designed to control two MOSFETs in a synchronous controller have a feature to switch off the lower MOSFET if the current reverses in order to maintain the efficiency.
 

Thread Starter

electromeow

Joined Oct 12, 2021
9
Should be doable, but you may need a logic inverting stage between the 3475 and the 28225 (I haven't checked the 28225's input logic requirements). Use of non-optimum ICs counts against efficiency though.
TPS28225's input is non-inverting. So can I use LM3475 with an external MOSFET driver if I can find a MOSFET driver IC which has an inverting input?
 

Thread Starter

electromeow

Joined Oct 12, 2021
9
What makes you think synchronous will be more efficient?
It will, provided that the current never reaches zero.
At low loads where a MOSFET/diode converter goes into discontinuous mode, a MOSFET half-bridge must remain in continuous conduction so the current goes below zero for part of the cycle so that it is flowing back from the output capacitor into the input. That's not efficient.
ICs that are designed to control two MOSFETs in a synchronous controller have a feature to switch off the lower MOSFET if the current reverses in order to maintain the efficiency.
I am aware of that, but except the times when I'm trying some unnecessary experiments, load will draw at least 2-3 amps, so it will stay at CCM.
 

Alec_t

Joined Sep 17, 2013
14,280
So can I use LM3475 with an external MOSFET driver if I can find a MOSFET driver IC which has an inverting input?
I don't see why not. But as the '75 has only the Pgate output you will need to find an inverting driver which can accept a single-phase input and can provide sufficient dead-time to prevent shoot-through of the power FETs. Some drivers provide very little (if any) dead-time, which can be a problem with big FETs.
 
Top