|
This is a test for an IC I'm working with for another project, but same idea.
One of the features of that IC is to set a minimum voltage on a solar panel to some constant voltage V_REF * (1 + R_TOP / R_BOTTOM) using a resistor divider as part of a concept called 'MPPC', which is similar to MPPT but dumber. Rather than directly tracking the MPPT, setting the limit appropriately allows you to approximately sweep the 'good' parts of the IV curve of a solar cell with less complicated ICs.
I figured we could make this the same as true MPPT by using some software to control this functionality. So, I have been trying to think of ways to automate that using a microcontroller.
Essentially, the idea is to take a PWM wave, which we can map to a [0 - 100%] duty cycle, not unlike the [0 - 100%] wiper position rule for a potentiometer.
The PWM signal gets filtered into an approximate DC voltage as a control signal for an op amp feedback loop. Essentially, the higher the control signal, the lower the base of the BJT, and as a result, the collector of the BJT increases to track this value.
You'll notice that the PWM signal iand resulting BJT collector are actually a percentage of the panel voltage at runtime, which is an extremely nonlinear relationship.
This is where the second feedback loop comes into play. All it is is a constant current driver to simulate a load *just* strong enough to sag the panel voltage enough.
By enough I mean that with a strong enough load, the panel voltage, and resulting BJT collector from PWM will drop, and through negative feedback, coincide with the 800 mV reference voltage I'm using here.
Putting it all together, the duty cycle of the PWM signal has a direct effect on the regulated DC output (orange) the panel puts out, namely:
V_REG = V_REF x (1 + DUTY / (1 - DUTY))
For instance, with the parameters in use right now at 20% duty, we can expect to see orange equal to:
V_REG = 0.8 x (1 + 20% / 80%) = 1V.
And sure enough we do.
Play around with it, it's pretty silly looking as a circuit and I'm aware that I could just as easily be using a digital potentiometer, but I'm going back and forth trying to think about the best for my needs here, which is probably the simplest solution in a digital pot.
|
