Originally Answered: Servo or stepper for solar tracker?
The design you show uses azimuth elevation mount as far as I can see. The panel rotates and tilts, like a gun on a ship. This means a keyhole is present, a zone of confusion, which would be when the sun is right overhead, which happens anywhere in the tropics. Also to move from any position of the sun to another, both motors move. A calculation may be needed to move to this given position. The first link refers about the calculation
A better arrangement is the X,Y mount. Here the Y axis tilts the panels in the EW direction. It has to rotate 180 degrees from dawn to dusk depending on the horizons. The X axis tilts in the north south direction. It sets the angle of the sun each day. The second and third links show this mechanism clearly. Only the Y axis moves most of the time. The sensor can be an X pair and a Y pair (quadrant) that is able to drive an x, y movement directly. No computer, just servo can do this.
Servos seem better in that there is a sensor and feedback to indicate where the panel is. Also the motor is smaller for a given torque. However small motors are fast, so huge (lossy) gearing is required. The stepper may seem simple to implement, but it needs to be driven by something that converts position to pulses, and needs a reference position (home switch) to start from in each axis. If wind or other mechanical force causes steps to be lost or added, the position is lost unless it also has feedback.
The position of the sun can be calculated, which eliminates problems tracking during rain and clouds. The position is easier to translate to an X,Y mount, considering you may be using a simple microprocessor. Note there needs to be a sun rise algorithm, something that positions the panels ready for the morning sun. This is one advantage of using a microprocessor, it is much easier as this can remember where it was "yesterday" (saving power).
The driving torque depends on the motor, gearbox losses and gearbox ratio. The load torque depends on the mounting arrangement of each axis, the worst case angles, the weight, wind loading etc. The motor torque and the holding torque (braking) are separate issues. A servo has the full motor power available to hold it in position, but that is not what you want to save power. The holding torque of a stepping motor is specified (no current situation), but be aware it drops with aging as the magnets weaken. Go for bigger than expected with stepper motors and avoid disappointment.
The power required is about how fast the panels move. This is so slow it will mainly be to overcome losses with the likely gear ratios. The sun moves one degree east to west every 4 minutes. An error of 5 degrees in x or y represents a loss of 0.4% from the panels.
Torque is in newton meters, 2.5kg has a force (weight) of up to 2.5kg * 9.81m/s/s = 25 newtons. The radius from the axis of rotation gives the torque. The 2.5kg may be distributed evenly or not, depending on the design, but could have 1 meter radius. The whole mass could even be dynamically balanced so the torque applies to acceleration only. Without balancing, the force could be of the order 25Nm or more. A 1W drive has to turn the shaft at maybe 30 degrees a minute maximum (0.0833rpm), so the torque from 1 watt would be 114Nm at such a slow speed (but consider losses in gears). The reduction for a 5000rpm small motor is very high at 60,000 to 1. The torque depends on the gear losses. It could be that 10 watts and more might be used driving the gears.This makes a stepper attractive as it is slower. If high friction gears like a worm drive is used, the holding torque is increased. The stepper motor gears can be some arbitrary ratio like 20:1 that is a trade off with losses, holding torque and driving torque, remembering steppers have maximum torque when stepping at low rpm.