High Accuracy Solar Tracker

Oct 24, 2021

Considerations when constructing a high accuracy solar tracker to achieve 1/100 degree accuracy.

Outlined below are issues that can affect accuracy.

 

(1)

    The performance data stated here is a result of 10 years of laboratory calibration and observation of Kinematics slew drives and Venture Manufacturing liner actuators with brushed & brush-less motors. This has been done independently and without involvement of the drive manufacturers. Claims of achieving 0.01 degree accuracy with brush-less motors have been confirmed by Rhineland TUV in 2016.

Hall Feedback Resolution

Most motors have hall feedback which to calculate position from. One complete rotation of the motor produces a number of pulses multiplied times the combined gear ratio.

This is expressed in pulses per revolution (PPR) for the slew assembly.


A typical 3″ Kinematic slew drive produces 29265 pulses for each 360 degree rotation. A quick calculation reveals that a PPR of 29265 does not have the resolution down to 0.01 degree since pulse counts are binary. The best theoretical accuracy is 0.012 degrees.

A resolution of 1 pulse per 0.01 degree is still not ideal in terms of the controller. In this case an optical encoder would be advised to produce more counts per motor revolution.

Hall PPR Resolution Calculation

29265 / 360 * 100 = 0.81

Inclinometer Feedback Resolution

Inclinometers are easier to evaluate for accuracy because of good data published by the manufacturers but analog devices are subject issues not found with hall feedback.


In this example a Turk model B1N360V is examined.

The resolution (0.14) immediately indicates that this is would not be enough for the intended purpose. It is true that this example is for 0-360 span and our application only requires 90 deg. and there are other inclinometers with smaller spans which increases the resolution but there are other factors as well.

Linearity and temperature drift also adds inaccuracy. We should examine the repeatability as an important factor. If the inaccuracy is consistent and the sensor repeats itself then this can be mitigated to an extent.

One consideration often missed is in the controller. Is the Digital to Analog converter 16, 32 or 64 bit? The simple conversion in the controller processor could also introduce small inaccuracies.

If the inclinometer communicates through a serial protocol such a Modbus it is possible to introduce errors due to poll rate.  The poll rate should be less that the rate of change of the solar position between .01 degree behind the sun to .01 degree ahead of the sun.

 Typical Inclinometer Specifications

  • Measuring range: 0-360 deg.
  • Resolution: <  0.14 deg.
  • Linearity deviation: < 0.6%
  • Temperature drift: < 0.05%
  • Repeatability: < 0.2% of measuring range

Poll Rate

    The time in seconds for the controller to poll a Modbus device and receive values into registers.

Brushed Motor Run-On

Brushed motors have no internal stopping mechanism like brush-less or servo motors. There can be slight rotation after power to the motor is OFF. This motion is translated to the gears advancing the tracking frame. It is hardly perceptible to the observer but the error introduced can be as much as 0.1 – 0.2 degrees (1)

By reducing the motor speed this run on can be reduced which can be accomplished by adding resistance to the control signal powering the motor or by a PWM. This then brings up the question: “How then to provide full speed signal to the motor for long moves?”  This is answered later on in this article.

Gear Backlash

Brushed motors rely on the slew drive or linear actuator worm gear to lock into place when the motor is not turning. The final position is somewhere between the calculated position plus or minus the backlash. Normal wear can also increase backlash. 

If a constant load, always in one direction, could be maintained on the actuator then the backlash becomes a constant and can be calibrated out.

The best example is the linear actuator for tracking elevation. The tracking frame is always experiencing a downward load because it is opposed by gravity.

The same cannot be said for the Azimuth axis. Theoretically each move should be repeatable but slew drives need even lubrication throughout their travel. Dry spots are common and can add significant hysteresis. This can potentially change the final position within the backlash.

Gear backlash

    The clearance between mating gear teeth. This clearance or slop causes lost motion between motor input and drive output, making it difficult to achieve accurate positioning.

Tracker Frame Slop

The tracker frame should be robust enough to not deflect under mild winds or other external forces.

VT-IPM2M 113 SPA Series (Solution 1)

The 113 series has 4 relay outputs that control 2 axis. Two for direct motor movement and 2 for reverse movement. When used with servo motor controllers and brush-less motors can eliminate motor run on inaccuracies. The 2 speeds are achieved within the motor controllers with build in speed ramping settings. This was the design that was confirmed by Rhineland TUV.

VT-MIPM 135 SPA Series (Solution 2)

The 135 series has 8 relay outputs that control 2 axis. Two for direct motor movement fast, 2 direct motor movement slow, 2 for reverse movement fast and 2 for reverse movement slow. The slow relays output are governed by a PWM and the fast relays output are full power. The control logic selects the fast or slow mode based on the size of the move in degrees. No other motor controller is necessary.

Control Program 1220-800 (Principle of Operation)

The solar control program 1220-800 calculates tracker position based on the National Renewable Energy Laboratory (NREL) Solar Positioning Algorithm (SPA) and is able to calculate solar position within the required accuracy.  It is agnostic of any error introduced by the tracking system but can position the tracker accurately during any weather condition.


Optical solutions alone can be very inaccurate on cloudy days and chase bright clouds on sunny days. Creating a hybrid of the 2 technologies provides a promising solution for consistent high accuracy and allow the tracker to learn as time passes how to compensate for all the above problems outlined. Tests have shown that this method could also work to correct for foundation settling or even displacement due to earthquakes.

Solar MEMS hybrid tracker

Control Logic when Tracking using Solar MEMS

Calculate Solar Position from NREL (Sp).

Calculate present position value of tracker (Pv)

Calculate deviation c_dev = Sp – Pv .

If solar radiation is greater than 300 W/m2

Get deviation from Solar MEMS sm_dev

Calculate  error err = sm_dev –  c_dev

Store err.

Otherwise use last known value of err.

If Pv + err > 0.01 then command motor MOVE

The parameters for setting accuracy is through the deadband setting. This setting is plus or minus therefore if the desired control accuracy is 0.01 degree the deadband should be set to 0.005. Once the Pv falls behind the sun by the deadband then a MOVE command is initiated.

Tracker Time Constant

Once the MOVE command is received the output to the relay is governed by:

      • time constant
      • overshoot
        Time constant if tuned with motor speed control can be adjusted to fall within deadband as shown by shaded area.
    • Overshoot must be much greater than deadband to prevent cycling.

Time Constant

A newly calculated position Pv may not occur fast enough to stop a short motor move. The position could be well past the target once this happens so a configurable time constant value limits the time the motor can make any short move independent of Pv.

The controls will determine before the move if it should be long or short and not limit long moves. When a long move comes within the range of a short move then it should switch to a time constant limited output.

Time Constant

    The maximum amount of time in milliseconds the motor can make a short move. This cannot be less than the scan rate of the control logic.

Permissive

Before any motor all movement criteria must be checked. Some examples are if limit switches or soft limits are reached. Is a hall feedback failure detected or is there a time delay between motor direction reversal?

Permissive

A binary bit (TRUE or FALSE) that results from a scan of internal alarms which would deny motor movement.

    Overshoot

    A newly calculated position Pv may not occur fast enough to stop a short motor move. The position could be well past the target once this happens so a configurable time constant value limits the time the motor can make any short move independent of Pv.

    The controls will determine before the move if it should be long or short and not limit long moves. When a long move comes within the range of a short move then it should switch to a time constant limited output.

    Overshoot

          The maximum amount in degrees that the Pv can over the deadband after a MOVE without reversing direction.