The Solar Power Expert Blog

Testing the Solar Charge Controller Schematic

by Chris - October 21st, 2011.
Filed under: Charge Controller. Tagged as: , , , .

I was finally able to find some time to test my step down converter prototype. The results showed an efficiency of 71%. That’s not bad for a first prototype! Below, I’ll show all the test data, schematics, pictures, and some of the ideas I’ve come up with for improving the design.

Test Results

Please read my previous post on the DIY dual voltage converter to read about the concept behind this circuit. The whole point of my test was to explore the frequency response of the charge controller prototype. I started with a 60 Hz square wave into the gate pin of the MOSFET. I kept the waveform at 50% duty cycle and swept the frequency up to about 100 Khz with a function generator.

While the waveform into the gate of the MOSFET was generated by a function generator for this test, the final version will drive this pin with the output of the microcontroller – like an Arduino or Maple. The function generator simulates the output that a microcontroller would put out.

Test Connections to the Solar Charge Controller

Test probes connected to prototype

As I changed the frequency, I was able to observe the output voltage of the transformer on the oscilloscope. The sweet spot appeared to be right around 33 Khz. Frequencies below this point caused the voltage to drop off and frequencies above this point also caused the output voltage to drop off. I also played around a little bit with the duty cycle (the amount of time the square wave is high), but it didn’t have a huge effect. It did have a little effect however, so I’ll plan on using the duty cycle to find tune the circuit once I get the big stuff nailed down.

Scope shot of the Solar Charge Controller Prototype

Scope shot. Lower waveform is 'V1' to the gate of the MOSFET. Upper waveform is 'Vout', the voltage across the 3 Ohm load resistor.

I limited the voltage across the circuit to 12 volts and the current into the transformer to 4 amps. At the 33 Khz center frequency, the power supply topped out at 12 volt and 3.6 amps. This made for a power input of:

Pin = 12 volts * 3.6 amps = 43.2 watts

Power for the step converter

Power Input to the Solar Charge Controller

I was able to achieve a peek output voltage on the transformer of 9.6 volts (see scope shot). The load was a very large 3-ohm resistor, which means there was 3.2 amps passing through it. This meant that the power out of the circuit was:

Current = 9.6 volts / 3 ohms = 3.2 amps
Pout = 9.6 volts * 3.2 amps = 30.7 watts

Diving power out by power in gives efficiency:

Efficiency = 30.7W/43.2W * 100% = 71%

Like I said, not bad for a first try. 71% is pretty low efficiency for an electrical power circuit. If the efficiency was in the range of 80% to 85%, then this circuit could hold its own against industrial type power supplies. However, this circuit was extremely easy to build using off the shelf parts and a protoboard. I’d be willing to sacrifice some efficiency if I know I could easily build and repair a circuit like this.

While I was pretty happy with this initial result, the circuit got pretty hot! In fact, I let the test run too long and the MOSFET cooked itself to death – not a big problem as I have 19 more. At 71% efficiency and a power input of 43.2 watts, it means this circuit was absorbing 12.5 watts:

43.2 W - 30.7 W = 12.5 W

Moving Forward

The inductor got warm, but most of this energy was going into the MOSFET. That means I need to find some way of handling the heat. This can be done by better heat sinking or by making the circuit more efficient. Realistically, I’ll probably have to do both.

Schematic of Solar Charge Controller Prototype

Initial Solar Charge Controller Schamtic

The solar charge controller schematic above shows the components I used and how I had them wired. All the component values on the schematic above reflect those used in the prototype. From here, I need to figure out how to make the circuit handle more power, run more efficiently, and figure out how to handle the heat. There are a few options for doing this. Here are a few ideas I had:

  • Run multiple MOSFETs in parallel

    As long as the MOSFETs were all made in the same batch, this should work. This would allow me to dissipate more power without the need for heat sinks. It would also improve efficiency by lowering the resistance created by the single MOSFET.

  • Run multiple transformers in parallel

    Running multiple transformers would decrease the inductance, which would lower the resonant frequency of the circuit. This would mean the MOSFETs could be driven at a lower frequency. This would give the microcontroller more time to do calculations and lower switching losses in the MOSFETs. It would also mean lowering the series resistance of the transformer windings, further increasing efficiency.

  • Use a better transformer

    I was really happy with how tightly I was able to wind the secondary onto the torrid. However, a manufactured transformer could probably achieve a tighter coupling. This would also greatly improve efficiency.

I’m going to dig into the math behind the transformer some more in order to come up with a good PSPICE model for the circuit. If I can come up with an equivalent PSPICE circuit, it would allow me to tweak the circuit and figure out where the ‘low hanging fruit’ is for improvement.

What do you think? Share your thoughts below or participate in the mailing list.

6 Responses to Testing the Solar Charge Controller Schematic

  1. Pretty good results! Do you mind if I ask why you put a 1N4004 across the MOSFET?

  2. The 1N4004 diode allows voltage spikes created by the transformer to bypass the MOSFET. It protects the MOSFET. However, I may not need it and I may be losing efficiency because of it. I’ll know whether I am or not once I conduct a PSPICE simulation. For this first go-around, I decided to play it safe and protect the MOSFET.

  3. I think a manufactured transformer will get you more better resulats as you are thinking. Let’s see what the result will come.

  4. Hi Chris, I think Will is right. I think we can remove the 1N4004, as we can see in the IRLB8748’s datasheets, it already has an internal diode (called Body Diode) that has that function.

  5. Sweet! Thanks for letting me know!


  6. To increase power handling. First; a larger transformer with lower gauge wiring and a larger core would be necessary.
    And why a transformer? Wouldn’t a simple buck converter with a single winding coil be more efficient?
    You can get much better power handling out of the MOSFET by heat sinking it. I thing that IRF has an app note that shows some ideas for PCB copper layouts that are effective heat sinks. Also, look for a MOSFET with as low an RDSon spec as possible.
    Mind your gate drive. For best efficiency you need to make sure that you get your Vgs up to the point where you achieve the lowest RDSon. Further, you need to get the gate voltage ramped up to the proper Vgs fairly quickly. As the Vgs lingers below it’s peak, it will cause the MOSFET to linger in it’s high RDSon region. Just some thoughts.

Leave a Reply

Your email address will not be published. Required fields are marked *