# Forward converter switching frequency

#### Imre Norbert

Joined Feb 24, 2021
22
Hello.I'm working on a Forward converter with two outputs and I'm wondering what frequency should I use in order to minimalize my losses.
Here are the converter details:
-Vin:18-30V
-Vout1:15V Pmax1: 15-20W
-Vout2:7.5V Pmax2: 8-15W
Currently I'm using 360kHz and I think this is a bit fast and I dissipate more power when switching that with a lower frequency.
My question would be: How can I estimate a switching frequency? What are some genral guidelines for this?
Thank you.

#### BobTPH

Joined Jun 5, 2013
3,129
Calculate the power dissipated with the switch fully on. The calculate the switching time, and therefore the time in linear mode, and the average power dissipated in that time. I would aim for keeping that power down to maybe 10% of the on time dissipation.

MOSFETs can switch quite fast, but it depends on the current you can supply to the gate. A better gate driver should improve performance.

360K is not that high, typical DC to DC converter chips operate at higher frequency than that.

Bob

#### crutschow

Joined Mar 14, 2008
27,018
As BobTPH noted, it's the MOSFET switching time that mainly generates the switching losses, so you have to determine that in the circuit.

#### Papabravo

Joined Feb 24, 2006
15,804
MOSFETS have an associated behavior called the Miller Effect where the gate voltage on the way up and on the way down is on a relatively flat plateau. This places the device in the linear region where it creates nearly a dead short between the supply and Ground. You can minimize this effect by using a current source to drive the MOSFET gate. There is a datahseet parameter that tells you how much charge you have to provide to turn the gate on or off. It has units of Coulombs

#### Imre Norbert

Joined Feb 24, 2021
22
Calculate the power dissipated with the switch fully on. The calculate the switching time, and therefore the time in linear mode, and the average power dissipated in that time. I would aim for keeping that power down to maybe 10% of the on time dissipation.

MOSFETs can switch quite fast, but it depends on the current you can supply to the gate. A better gate driver should improve performance.

360K is not that high, typical DC to DC converter chips operate at higher frequency than that.

Bob
Thanks for your help.I tried reducing the frequency to about 325kHz but it is a bit worse, so i suspect that a higher frequency could be better.

#### Imre Norbert

Joined Feb 24, 2021
22
MOSFETS have an associated behavior called the Miller Effect where the gate voltage on the way up and on the way down is on a relatively flat plateau. This places the device in the linear region where it creates nearly a dead short between the supply and Ground. You can minimize this effect by using a current source to drive the MOSFET gate. There is a datahseet parameter that tells you how much charge you have to provide to turn the gate on or off. It has units of Coulombs

#### Imre Norbert

Joined Feb 24, 2021
22
As BobTPH noted, it's the MOSFET switching time that mainly generates the switching losses, so you have to determine that in the circuit.
Thank you.

#### Papabravo

Joined Feb 24, 2006
15,804
Since this is a new design I would look at a switching frequency in the 1.2 to 2.5 MHz. region.

#### Imre Norbert

Joined Feb 24, 2021
22
Since this is a new design I would look at a switching frequency in the 1.2 to 2.5 MHz. region.
This isn't a new design, I got it as it is and I need to make it more efficient, right now is about 61 to 68 %.

#### BobTPH

Joined Jun 5, 2013
3,129
Well, then, the first thing to do is determine is where the losses are.

Bob

#### Imre Norbert

Joined Feb 24, 2021
22
Well, then, the first thing to do is determine is where the losses are.

Bob
Yes.I did that.I have chosen new lower RdsOn MOSFETs,i did replace a diode with a MOS,lowered the current sense resistor, and I just wanted to make sure that the frequency isn't the problem.Another problem is the secondary side RC snubber that I can't figure out.

#### BobTPH

Joined Jun 5, 2013
3,129
What about the inductors? Their parasitic resistance can also limit efficiency.

Higher frequency allows smaller inductors, which can have lower resistance.

Bob

#### Imre Norbert

Joined Feb 24, 2021
22
What about the inductors? Their parasitic resistance can also limit efficiency.

Higher frequency allows smaller inductors, which can have lower resistance.

Bob
On one output I have 12 turns and on the other 6.

#### Imre Norbert

Joined Feb 24, 2021
22
The snubber does get hot, I think I should reduce the capacitance to have smaller currents.Right now is 470pF.I did some calculations and got as result about 27pF. Next step I will try to change that too.

#### BobTPH

Joined Jun 5, 2013
3,129
On one output I have 12 turns and on the other 6.
That sounds way too small for your switching frequency. What is the ripple in your output? A large ripple would explain why an RC snubber is dissipating a lot of power.

I think it is time for a schematic.

Bob

#### Imre Norbert

Joined Feb 24, 2021
22
I didn't measure the ripple, but when testing the output voltage on DC Volts I always got a stable voltage for example 14.987 and it did not move at all, I have tried this with small and large consumption.

#### Imre Norbert

Joined Feb 24, 2021
22
However after the inductor I also have a 3 stage capacitor setup and it works fine.

#### BobTPH

Joined Jun 5, 2013
3,129
You can produce a stable voltage by running a PWM signal through an RC filter. But that will not have the efficiency of a switching power supply. In the latter, the current is limited by the inductor, which is lossless. If the inductor is too small, the current is limted by resistance, and is lossy.

If you could post a schematic of the critical parts of the output stage, leaving out the control logic, I could simulate it and see what is going on.

Bob

#### Imre Norbert

Joined Feb 24, 2021
22

#### BobTPH

Joined Jun 5, 2013
3,129
Okay, I am not familiar with that topology, so I cannot be of any more help. Looks interesting though. The inductors are not performing the function I imagined, so you can ignore what I said about their sizing.

Bob