Solar panel regulator design question

[JoeP]

I'm designing a circuit to deliver power from a 25W solar panel (which will never output very much power here in the UK) to a 24V lead acid battery. Since the power is so low, I'm not particularly concerned about trying to keep up a constant current phase, I just want to keep the PV at its optimum voltage (18V). I've designed the circuit shown below. The theory is that if the voltage is >18V, the transistor turns on and more current is delivered to the battery, reducing the PV voltage, and vice versa. What I'm concerned about is unpredictable oscillations, as the change in current is having to propagate back through an SMPS, which could take a long time. I've considered a few possible solutions, such as putting a capacitor across the B-E junction of the BJT in order to keep it in its active region, but I'm not sure whether this would work very well (if the PV takes a while to respond, it could take a huge capacitor). Efficiency isn't vital, so I don't mind if some power is dissipated in the transistor. Does anyone have any ideas?

[NiHaoMike]

Just limit the duty cycle of the boost converter so that it doesn't try to pull the input voltage too low.

[JoeP]

Interesting idea, but I'm struggling to see how that would keep the PV voltage at 18V. It would cap the output power, but depending on the light intensity, the duty cycle limit would need to vary (correct me if I'm wrong, I'm just guessing). Also, it wouldn't be very easy to do as I'm using a boost IC and not a microcontroller.

[mikerj]

Configure the error amplifier as an integrator and make the bandwidth much lower (order of magnitude) than the switcher.

You need to think about protecting the Vbe junction from reverse breakdown if/when the output of your op-amp falls well below battery voltage.

[David Hess]

It would be just as easy if not easier and more efficient to just use an 18 volt buck-boost converter.

You need to think about protecting the Vbe junction from reverse breakdown if/when the output of your op-amp falls well below battery voltage.

A diode in series with the base will protect the transistor however base-emitter breakdown is not a serious condition for a power transistor which will have low gain at high current anyway.

[NiHaoMike]

Interesting idea, but I'm struggling to see how that would keep the PV voltage at 18V. It would cap the output power, but depending on the light intensity, the duty cycle limit would need to vary (correct me if I'm wrong, I'm just guessing). Also, it wouldn't be very easy to do as I'm using a boost IC and not a microcontroller.

Do you have a link to the datasheet of the controller you're using? Limiting the duty cycle is often as easy as clamping the compensation pin to a certain value. Or go to basics and use a 555 as the controller with a circuit to force the output low if the output voltage reaches the limit.

You'll be better off implementing full blown MPPT in a microcontroller if you want to optimize the performance. If you don't trust your code to always behave properly, add some comparators to force the PWM signal low on an overvoltage condition. (A 393 will work nicely and double as the level shifter, then add a NPN transistor and diode to speed up the rise time.)

[Seekonk]

Just use the boost converters power electronics.  I do this all the time.  Your boost converter reference is likely 1.25V.  When the panel is below 18V, fake the boost controller into thinking it is an over voltage condition.   It can be as simple as using a TL431 with a pullup resistor. The output feeds thru a couple diodes and/or a LED to get about a 2V offset or more. When panel is over 18V, the TL431 drops to about 2V and no voltage feeds the boost converters sense input or drains from it. Under, it drives it higher and lowers the boost converters voltage. It is like a linear current booster keeping the panel above power point.

[JoeP]

Lots of good ideas. Mikejr, I'm struggling to see how that would work any better than what I've got with a capacitor in terms of reducing oscillations (I haven't been into electronics long, so I might be missing something obvious). David Hess, what do you mean by an 18V buck/boost? I'm trying to regulate the PV input not the output. NiHaoMike, I'm sure limiting the duty cycle would be achievable with a few extra components, but I'm still not sure that it will keep the input voltage constant. An MPPT would be best, but my micro is already pretty busy (both PWM channels are in use) and I'm not concerned about performance (this is a fairly small project). Seekonk, I'm struggling to visualize this circuit; could you please clarify?

I've started designing the PCB for this, so I'd rather stick with what I've got if possible (even if it sacrifices efficiency), so can anyone confirm whether or not my current circuit will function as expected?

[Seekonk]

This will give you the general idea.  The 5.6K resistor should be powered by the battery (not the panel) and be 15K. This is absolutely the best way to do it. If you don't have a TL431 handy (they are in almost any old wall wart) you can even use a power FET with changes in voltage divider to see how it works. Those will have gate voltages from 2.5 to 4.5V but not be temp stable.

[JoeP]

Right, I'm starting to understand. From the equivalent circuit in the datasheet, it seems like that circuit is acting in a fairly digital way, either pulling the feedback pin of the boost IC low or letting it go high, but presumably it doesn't do this due to the negative feedback. How did you derive the value for the 5.6K/15K resistor? Also, I don't understand the function of the diodes; can the TL431 not pull the output all the way to ground?

[Seekonk]

Lots of questions.  Think of the TL431 as an opamp, voltage reference and a transistor.  You have to supply some power to run this.  Everything works best if you stay above 2.5V.  Drive it hard and it will go just under 2V.  Input/reference is high impedance, but go over about 2.6V and it acts like a zener drawing current.  Up to about 10ma and then it destroys the chip.  Choose your voltage divider so that doesn't happen.

It is an analog chip used to make linear power supplies, but the switching is fairly abrupt.

15K.  I don't know how you do it, but I just guess since I don't know the resistance values of the boost converters voltage divider. It isn't critical, just need to drop the boost converters voltage just below that of the battery.  That won't take much current.

As said, the TL431 can't go to zero.  An offset of a couple volts is needed and at least one diode so the circuit doesn't act on the  boost converters voltage divider when the panel is above voltage. A traditional red LED is about 1.7V and is nice as it provides some indication, it still glows some when holding the panel at power point.  This circuit is so useful in solar.

[JoeP]

Right, I see. So when the PV is above 18V, the TL431 conducts, putting the diodes into reverse bias and making the boost regulator go to the voltage specified by its feedback divider (which should be about 30V), then when the PV is too low, the TL431 acts like an open circuit, allowing the 15K resistor to alter the boost's feedback divider so that the output voltage is lower (about 22V). The negative feedback then allows the TL431 to reach a sort of equilibrium. Is it safe just to connect this output to the battery via a diode? Thanks for all the help, and sorry for the questions.

[Seekonk]

The nice  thing about boost converters is they naturally isolate the panel from the battery. You still are powering the TL431 which is a little over 1ma and the controllers voltage divider. Or loose the power of a diode drop. Your decision.

Here is the nice option, A cheap PWM can be added.  This is one I use to maintain batteries.  The buck puts out 14.5V to the PWM controller. It says 20V, but that was when it was connected to a grid tie inverter.