"So, I can use the output section of this schematic. "..... yes you could, although I would prefer to see a totem pole of pnp npn transistors to drive the gate from whoa to go and back again... This will ensure proper gate turn on and turn off.
The frequency would be better at 20khz so you don't have to hear it squirm at lower light levels..... unless you want to hear it, and then go for 5khz or so.
The step off voltage I refer to is simply a poor mans mppt. The problem with solar panels is that they are essentially current sources with an internal resistance.
You will have heard on here before that if you connect a 18v panel to a 12v battery, the panel voltage will drop to meeet the battery termonal voltage, and you lose the 30% or so between 12 and 18v.
When you direct connect a panel to the motor, it does the same thing, except it has no terminal voltage, and so drags the panel down to whatever the panel can support at that light level.. The current will remain the same, but the voltage will drop..... so you throw away power, and are limited to the current dependent on the light conditions.
At low light, your motor was seeing short circuit current, but no voltage ( almost), so it sat there and cooked. On better days, this would not last long, and maybe it would start turning before the thing cooked, once away, the voltage rises, and your motor ahs back emf to balance the incoming voltage, and it runs normally. If the transition is too long for the physical mass of the motor to absorb the current before it starts to motor... then poof...
To stop this, we can use one of the op amps in the 494 to sense the incoming voltage. We then compare a sample of this to a fixed voltage. We can get that from the 5v regulated output from the tl494 ( pin 14 i think ). Even if it is just a 5k trim pot from 14 to ground with the slider going to pin 16 At mid point this will give you 5/2 or 2.5v on pin 16. That is our reference. Then it is a simple matter of taking the Array + and using a divider on it to give us about the 2.5v at the divider point.
Feed this to pin 15.
What should happen now, is if the motor speed control is set for say 3/4 speed (using pin one and 2 for that), if the pulse is too skinny, the pwm unit will try to widen the pulse by dropping the voltage on pin 3 ( op amp output ) As pin 3 gets driven down, the output if fed to another comparator internally, and beaten with a triangle wave.
The result is that when ever the pin3 voltage touches the triangle below it's peak, the pulse is the width of the line drawn across the triangle at that point. If pin 3 is zero, the full triangle is the pulse width... etc etc.
So at the time when it want to drive the motor up to full speed or whatever, it will try to pull 3 down, but the second comparator (pins 15 and 16) are orred with the same pin three.
In our case, we want the result from pins 15 and 16 to count... so as the voltage at pin 3 dropped by turning on the motor pulse, the flimsy panel voltage will sag... and this is caught by the pin 15 and 16 comparator. It then raises the voltage on three, and shuts the pulse width back down... so they fight over the same pin, and whomever puts the highest voltage on that point, will lower the pulse width.
What this means is that the voltage at the array will NEVER drop below what you set it for with the trim pot, as as soon as it tries, the pulse will narrow and it won't happen... no matter how hard the pin 1 and 2 comparator try to make the motor run fast, it just cant because of the voltage limit of 15,16.
This means your operating point is going to be always at or above the voltage you set. The resultant pulse width will be the compromise between what you wanted, and what can be achieved with the voltage available...... but we also get the buck converter effect. The capacitor on the array output ( yes put a cap on the input to this thing), will charge and discharge into the motor via the fets via the pulse width... so 1A @ say set to 16v will suddenly become perhaps 5A@2-3v.... so the pump will turn.... because the current is the torque, the voltage is the speed, we now have a current amplifier... it is going to turn no matter how low the light is ... until it gets very very overcast/night..... because the cap will discharge it's energy in bursts into a low impedance load ( locked rotor motor), so the current can get very high.... albeit wih a very low voltage.... so very low watts , but will still grunt and groan, but still turn. At low speed the pump will provide low load to it too.
The last thing to control is the max speed, so the pump runs for a long time, at reduced voltage... you can use the voltage across the freewheel diode across the motor as the reference, and this time you could use the pin 4 ( pulse width control max). I last did this a long time ago, but used a little opto isolator to look at the voltage across the motor, and then use the output to drive some of the 5v regulated into pin 4 proportionally... pin4 high, no pulse, pin4 zero, max width.
None of the above is what Elec Engineers would do, but I can vouch for the effective result. There is a 2hp solar water pumping unit here running one of these for the last few years.... probably not the right way, but a way anyhow.
I no longer have the circuit, or the board drawing, but thats how i remember it.
http://www.ti.com/lit/an/slva001e/slva001e.pdf Use this link to learn about the chip. They are in nearly all the computer power supplies I have torn down, also KA7500 is the same thing.
Which ever way you go, you do need to control the step off point to get any useful result short of lots of sun, and you do need that for the current boosting to stop the burning smells.. The max voltage control will give your motor long life, and the speed control to give you the feeling you have some control over all this .
Someone who knows what they are really doing like Opera or Joe will be able to explain better, and may have some circuits to try.
.......................oztules