DC/DC Zeta converter power supply short circuit

[Chang96]

Hi everyone, I'm new to this forum.

I currently doing a project to create a zeta converter for solar panel application and my target output of the circuit are 12V and 100W if possible. I simulated my circuit using SIMULINK and fine it works well (for data collection to choose components).
However, I constructed the circuit myself and test using 12V as input, 12V pilot lamp as output load, and set the duty cycle as 50% (10kHZ) , the circuit shorted (input voltage drop from 12V to 4~5V) and MOSFET starting to heat up. I checked my circuit and component but everything seems fine to me.

I begin to doubt my current design has some problem that causes the inductor L1 to be shorted to the ground. (I swapped a new one and still occur ). Can anyone advise me on what I missed from my design?



Here is my circuit specifications:

Input voltage: 10~18V
Output voltage: 12V
Target output power: 100W
PWM frequency: 10kHZ , generated using Arduino UNO Timer1 from pin 9;

MOSFET: IRF9Z34N ; PMOS Vgs(th) = -2~-4V(max)
MOSFET Driver circuit: +12V switching using 2N3904; PWM signal from arduino pin 9 10kHZ;

Diode: MBR30100 CT; Vf = 0.86V;

Capacitor: Nichicon PW polarized capacitor series
C1 & C2 : 4700uF
C3: 470uF;

Inductor L1 & L2: 2mH (current rating: 12A);

really appreciate your help

[SuzyC]

The P-MOSFET is shown in your circuit connected backwards. 

Your switching frequency should be about  50 to 100-times faster.

The inductor values should be around 100x smaller( 15uH )   and wound on the same core.

C2 needs to be non-polarized.

The 2N3904 switching transistor collector resistor will cause(even if Q2 was connected correctly) the P-Chan MOSFET to turn off very slowly and the FET's dissipation will skyrocket and your efficiency would approach zero.

A fix: the gate of Q2 should be driven with a NPN/PNP totem-pole complementary pair. The '3904 could then drive the connected bases of the complementary pair and the connected emitters driving Q2's gate.

If you were to correct these few design errors, you should attempt to test by slowly increasing the turn-on time (starting at 0-dutycycle) of the 2N3904 to achieve the desired output voltage with a suitable load.
With my suggested type of gate drive, input voltage range should not exceed 18-V or you will blow Q2.

Otherwise than that, it should work fine.

[Chang96]

Hi SuzyC,

Thank you for pointing out my mistake. I really had no idea I messed up so much in my work.

Initially, I want to use 100kHZ as switching frequency but i decide to reduce it because I afraid Arduino UNO unable to generate 100kHZ PWM so I design a much lower frequency.
With lower frequency, all my cap and inductor (single core 2 terminal, 2mH checked with RLC meter) ended up becoming huge but I able to brought them from aliexpress.

About C2 I also suspect I use it wrongly because most of the zeta converter used in the paper are reverse polarity or bipolar. Since connect reverse polar will be dangerous (imagine 4700uf cap exploded) I will buy bipolar one to modify my current design.

For the MOSFET gate driver, I will take your advice to build totem-pole complementary pair to drive Q2. I am still new to the MOSFET gate driving circuit and didn't realize it can harm my circuit.

Thanks again for sharing your knowledge with me, I will be sure to share my finding here later if I manage to fix it.

[KT88]

In your current circuit the PMOSFET won't turn off completely (Vgs is -6V when off). The gate pull-up needs to be tied to 18V as well. With a totem pole driver you should also implement a Zener diode to limit Vgs.

Cheers

Andreas

[megajocke]

In addition to the other great comments: the coupling capacitor positive side needs to go toward the output of it is not bipolar.

Also I am wondering what kind of inductor you have. It sounds like one intended for current-compensated common mode filtering applications. Such a choke will saturate at low current. A 2 mH inductor rated for 12A with the fields enhancing each other will be the size of a fist and weigh a kilogram or so. It would likely not be very good at 100 kHz either.