The current booster is almost mandatory I feel, and only costs a few dollars to do.
I have built a few solar pumping systems, and all use some form of current boost, even the 2-3kw ones.
In it's simplest theory, we need to understand that with an electric motor, current is torque... period.
Voltage is rpm.. if you have the torque in place... and one times the other is power.
But to start a pump turning, we need torque. The voltage is immaterial, and is only there to force the current through the circuit.
It is simple in theory and practice.
If we put a capacitance across the output of the panel, the cap will appear as a short for the first fraction of a second, then as it reaches full panel voltage, will appear as an open circuit.. and everywhere in between... so the current from the panel (with whatever light is available) will have a full capacitor with a time proportional to the ambient light (and the capacity of course).. more light, faster to reach fully charged.
The charged capacitor will now when full, carry an energy of 1/2 e^2 C... so now all we have to do is dump this energy into the motor.
The capacitor is happy to dump all it's charge/energy at once, and so is capable in theory of infinite current (think torque)... in practice it will be governed by the impedance/resistance of the circuit... ie the motor resistance, and the circuit (wire and switch) resistance, and whatever .
So all we have to do is place a voltage dependent switch to dump the cap into the motor, and we multiply the current by huge amounts..... but the time between charging up to the jump off voltage will vary with the light/input current.
This makes a poor mans buck converter/current converter. (anyone with a Ghurd dumpload unit will see this switching stuff when the dump load starts to turn on and off at varying frequency ... dependent on input and hysterisis and dump load impedance..... adds up to a kind of poor mans variable PWM)
In fact if we could use the Ghurd voltage switch of his low voltage cut out ... then we achieve the same thing. ie when the voltage is too low, and the cap is charging, the fet is turned off ( low voltage cut out is active.. fet off), when the cap gets to trigger voltage, we turn on the fet and dump all the energy into the motor via the fet ( and freewheel diode) and then rest and wait again... we pick the capacitance to get the on/off times as we want them at low power. We can set the voltage turn on at somewhere around the mppt points, for the best result ( Yes we get poor mans MPPT as well). I find for a 36v panel, that working around 31 volts is a good place to start, as this pulls the mppt voltage point down when the panel is in poor light, but the panel will overtake this point when the light improves and effectively overpowers things and finds it's own point after that. Providing the motor can handle the max power point voltage of the panel, all is well....other wise driving a normal motor PWM with a voltage switch on it's input, and a resistor on that input to limit the max voltage from the pwm works as well.
This means with very little power from the panel (feeble current) we can in fact turn a quite large rotor.... it will go in steps like a stepper whilst the power is very low, but it WILL turn the shaft.... it wont sit and cook.
As the light gets stronger, and more electrons are released for work, the rate of turn on/off will rise, and you will hear the familiar squeal of a PWM driven motor.
When the panel has enough power to run the motor direct, the circuit is turned on all the time, and plays no further part in the system, until the currrent from the panel falls off due to light getting less (clouds, night).
I have also used Curtis 1204 275 amps 36 volt pwm motor controllers for 3hp and more motors, with a simple circuit in front of it to turn the controllers speed control input up and down ( few opto isolators and a zener), and this PWM output mixed with the inductance of the armature and the freewheel diodes, makes a very effective buck converter. I have seen 5 amp panels drive over 40 amps into the motor in poorish light using this system.. the shaft will turn even in fairly sad light. (40 amps at only a few volts when the panels were putting out less than an amp... big motors and fets have very low resistances, so you can get high torque at low voltages) This system can easily make 60 or more liters/min into a 120 psi head.... it's a big mono pump with a 3hp brush motor.
In this case, the down side is that the very substantial caps in the controller cop all the ripple, and so their replacement is inevitable on a bi yearly basis ( internal heat from the switching, and from the caps themselves is problematic in high summer, and 60 amp loads)... external caps are much better.
So it is neither hard or difficult to fashion a current booster for a pump...... but try to keep the switching voltage to the fet gates at at least 13v to keep them turning on and off cleanly with little 12v systems..... 36 volt systems makes this easy, low light into 12v systems... not so much..... though at least when the light is low, so is the power, but the current can still be quite high, but the duty cycle will be low..... alla trade off.
........oztules