Synchronous buck/boost converter using totem pole gate driver

Status
Not open for further replies.

Thread Starter

ksc2023

Joined Aug 6, 2023
1
we are trying to develop a very low cost buck (day time)/boost (night time) converter for charging LFP batteries (day time) and powering LED (night time)! Due to cost impact we are trying to avoid half bridge gate drivers if possible at all!

Converter circuit - below is the simulation circuit of our buck converter, which will work as a booster for 18V in night time. Input - 25V , 6.5A Battery (fitted with BMS - 12.8V)
1692956762760.png


High side driver:
1692956806473.png


Low side driver:
1692956831714.png

PWM (deadtime of 2us will be managed by MCU PWM generator) - 20KHz

Output:
1692956895939.png

inductor current:
1692956920510.png
1692956707171.png



Issue - we had to put R6 (3.3K) on high side driver & R14 (330) on low side driver to reduce the current flowing out through Q2 & Q4 PNP respectively during PWM fall (around 180mA) on each side due to gate capacitor discharge, which totals to around 6W of power dissipation on high side considering high side gate voltage is coming around 35V with reference to ground. But commonly available smd transistors (SOT-23 pack) are around 0.3W to 0.5W range. which seems not feasible to use and hence we had to introduce the resistors, but in-turn to overcome slow gate capacitor discharge we had to increase deadtime around 2us to reduce the shoot through current due to the high side and low side overlap, and it is impacting the efficiency.

Question - are we missing anything here? we considered using MOSFETs for driving circuit but again the cheap smd packs will be of similar power rating. what are we missing?
 

tindel

Joined Sep 16, 2012
934
What you have depited is only a buck converter - you need another half-bridge to make this buck and boost.
I'm assuming you've considered using the SEPIC and Zeta and have ruled it out for some reason.
If you're doing this discretely, why not use a pMOS for the high side switch?

Is what you're experiencing happening in real life or just this simulation? Asked another way - have you built the prototype and this is what you're experiencing as well?

Something isn't adding up.
If you're charging these transistors to 12V, discharging them through 330ohms = 12V^2/330 = 0.436W if you were turning them off all of the time, but you're not, your only turning them off for fractions of a second at 20kHz, so the power dissipated should be much less. You can use the gate charge curve, below, to do some estimations. I'm guessing you're getting shoot-thru on your drive transistors. Q1/Q2 and Q3/Q4.

I just realized that your high-side driver should be referenced to the switch node and not ground.
 

Attachments

tindel

Joined Sep 16, 2012
934
The gate charge curve shows 80nC at 12V

I = 12V/330ohm = 36mA

Q=t*I
t=Q/I
t=80nC/(36mA/2)~=4.5us

f=1/20kHz=50us

So you have roughly 18mA discharging the FET over ~10% of the time... so you should be dissipating about 20mW in the 330ohm resistors.

Again, I think you're getting shoot-thru on your drive circuitry.

Note also that 330ohms is very large, and you want it to be smaller to to discharge that capacitance more quickly... 1-2A of charge/discharge current is typical.

I have attached a MOSFET appnote that I reference often to help me understand these effects. I recommend you read it a couple of times.
 

Attachments

tindel

Joined Sep 16, 2012
934
One last thought -

Spending $1 or so for a gate driver or two is way more economical that you trying to figure this out - and gets a product out the door more quickly
 

Tracy2022

Joined Nov 22, 2023
5
One last thought -

Spending $1 or so for a gate driver or two is way more economical that you trying to figure this out - and gets a product out the door more quickly
No. In High radiation environment it would be x10K. Plus useless time spent in debugging a bad IC based driver with band-aid ideas from App engineers of vendors.
 

Papabravo

Joined Feb 24, 2006
20,600
No. In High radiation environment it would be x10K. Plus useless time spent in debugging a bad IC based driver with band-aid ideas from App engineers of vendors.
I did not see any reference to a "high radiation environment". Where did you pull that one from?
 
This thread is three months old. and the poster only has 1 post. My guess is that it's a bot.
No Bot. It is real. Looking for a good isolated (transistor, Transformer based) gate drive for radiation environment. Before, for commercial product companies, i always used IC driver (it is convenient).
 
Status
Not open for further replies.
Top