1
\$\begingroup\$

I'm trying to design a buck converter to get 3.3v from a 6v solar panel. I am thinking of using this as my p channel mosfet:

http://www.farnell.com/datasheets/196719.pdf

I was thinking of the IRF6218PbF because it has a low Rds and needs only -3 to -5 volts to turn on according to the datasheet. The pwm to control the mosfet in my circuit comes from an attiny85 run at 3.3v with an fsw of 32kHz. I wanted to make sure of two things:

  1. As the mosfet requires a negative voltage to turn on this means that my mosfet would turn on when my pwm wave falls to 0v and turn off when it goes up to 3.3v correct? So whatever I set the duty cycle in my pwm to the actual switch duty cycle be 1 minus that right?

  2. If I'm right about part 1, how likely is it that this switching circuit will be too unreliable with a variable output from the solar panel, requiring me to find a mosfet with a very tight tolerance to avoid the switch not turning on in low light conditions.

If I left out any necessary information or my question sounds just plain stupid please let me know and I'll change or add what is needed, thanks for any help.

Edit:

The buck will be used to charge a 3.7v lithium ion battery which is running some microcontrollers, the overall power draw from the load is small, about 40mA. I believe the attiny85 will be fast enough, it has an 8 MHz clock and looking at the pwm output on an oscilloscope it is very clean at 32kHz. The source V2 is in place of the attiny85 in the schematic.

Schematic of Buck Converter

\$\endgroup\$
  • \$\begingroup\$ What is the intended current for this buck converter? What your gate driving circuit of going to be? Are you sure that a garden variety microcontroller is going to be fast enough to do closed-loop control of a buck? Last but not least: post a schematic of what you currently have in mind. Even a pencil (or white board) drawing would do. A buck converter shouldn't have too many components. \$\endgroup\$ – Nick Alexeev Feb 11 '16 at 4:36
  • \$\begingroup\$ Hope I got everything you needed, for now I don't have any feed back involved but hope to implement it after testing the circuit as is. \$\endgroup\$ – elecStudent Feb 11 '16 at 5:06
  • \$\begingroup\$ Why do you design a buck? Take an existing one, with all recommendations inside the datasheet \$\endgroup\$ – Gregory Kornblum Feb 11 '16 at 5:07
  • \$\begingroup\$ I know I could just buy one but I was hoping to build one instead, for the experience. \$\endgroup\$ – elecStudent Feb 11 '16 at 5:10
  • 1
    \$\begingroup\$ So from what I understand a P channel will simplify the circuit as it requires a -Vgs to turn on. An N channel would require a +Vgs so my control voltage would need to be higher than my source, which is at my input voltage, so an extra driver circuit is needed. \$\endgroup\$ – elecStudent Feb 11 '16 at 5:18
2
\$\begingroup\$

As the mosfet requires a negative voltage to turn on this means that my mosfet would turn on when my pwm wave falls to 0v and turn off when it goes up to 3.3v correct? So whatever I set the duty cycle in my pwm to the actual switch duty cycle be 1 minus that right?

Correct.

If I'm right about part 1, how likely is it that this switching circuit will be too unreliable with a variable output from the solar panel, requiring me to find a mosfet with a very tight tolerance to avoid the switch not turning on in low light conditions.

If you fear that the panel voltage will fall too low for the buck converter under low illumination, the IV curve is actually quite flat and it likely won't be a problem.

PV IV curve from adafruit

In simpler terms, the voltage will stay relatively constant under varying illumination, while the maximum current you can draw will fall drastically. As you can see at any level of illumination there is a sweet spot of voltage and current where the panel delivers maximum power. Since you are charging a battery, you can control the charging current to match that spot. This is known as MPPT (Maximum Power Point Tracking) in the industry. To do this, you need to monitor at least the charging current and the charging voltage, preferably also the solar panel voltage.

With the basics out of the way, let's look at your preliminary circuit design.

1mH is a huge value for an inductor, and will either mean a horrendous series resistance (several ohms) or a physically imposing component. Even at such a low frequency as 33kHz, you need much less inductance. You generally want the smallest inductance you can get away with (that won't go into saturation).

The MOSFET you selected is really not that good. Its on resistance at low gate to source voltages is huge, and it won't even turn on at -3.3V.

MOSFET

Instead, it is rated for -150V drain to source, which is completely unnecessary for something that won't even see 20V. While MOSFETs that you can drive directly with the microcontroller at 3.3V at good efficiency are available (e.g. IPP80P03P4L-04), it could make more sense to use a lower performance N-channel part with a driver.

Your circuit completely lacks any kind of input capacitor, so the solar panel would be subjected to a huge amount of current ripple. You want a low ESR capacitor at the input. The output cap at 50uF is too small as well, but at least the battery is a fairly undemanding load.

\$\endgroup\$
  • \$\begingroup\$ Thanks for all the great feedback. I did see the IPP80P03P4L-04 when I was looking at mosfets my one worry was that since the range for VGS was -1 to -2 volts typically, wouldn't it be on all the time for me? Assuming I have 6v in and 3.3v at the gate when the pwm goes high that's still only -2.7v, or am I misunderstanding something? So if I'm right about that maybe the N channel with driver is the way to go. \$\endgroup\$ – elecStudent Feb 11 '16 at 14:41
  • \$\begingroup\$ @elecStudent You could either tri-state the AVR pin and have a pull-up resistor turn the transistor off, or capacitively couple the AVR pin to the gate (with a large value pull-up, e.g. one megaohm). If I were you I would get a proper high and low side driver and two N-channel MOSFETs (one would replace the schottky diode for increased efficiency) \$\endgroup\$ – jms Feb 11 '16 at 15:45
  • \$\begingroup\$ Last question, I appreciate all your help, if I chose a driver like this cds.linear.com/docs/en/datasheet/lt1157.pdf (I'm assuming 2 independent channels means I can configure one as high and the other as low?) and added that capacitor to protect the panel from ripple current as you mentioned, is there any reason I couldn't hook the driver up directly to the panel to get the full six volt input instead of 3.3, which wouldn't give me the necessary Vgs to run my mosfets at an ideal Rds? \$\endgroup\$ – elecStudent Feb 11 '16 at 18:53
  • \$\begingroup\$ That is an interesting chip, but it's not good for PWM. Look at the turn on time, it takes 200 μs for it to drive a 1nF gate high, and for smaller voltages than 5.5 V it's even worse. The intended purpose is to drive MOSFETs that are used for routing power (that stay on continously for minutes to hours). \$\endgroup\$ – jms Feb 11 '16 at 19:41
  • \$\begingroup\$ I completely missed that, you're absolutely right. Saying I did find a better chip could I run it directly off of the solar panel? \$\endgroup\$ – elecStudent Feb 11 '16 at 19:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.