An intelligent solar tracker (read 1st post first)

Discussion in 'The Projects Forum' started by umer27, Mar 3, 2009.

  1. umer27

    Thread Starter Member

    Feb 28, 2009
    25
    0
    Read this INTRO first please ..


    Dear All !

    Im designing an intelligent single axis solar tracker as an engineering project. Im doing this for a manufacturer , and its going to compete with a commercially available tracker, my tracker will have learning capability, It will do the learning on the first day, and from then, it would update its data on a daily basis on the amount of power it gathers.

    So, I'd like to discuss lots of things with you guys because I think there are a few very bright people here.Will not be discussing the fine details here...

    A little introduction about my tracker,

    --> Single axis (90 degree east -west)

    --> Uses a 10W solar panel as a sensor and its not connected to the other big panels..

    --> Im using a linear actuator with a reed sensor , amazingly takes ~2A current at 12V for moving this mammoth 300kg load (coz of the ball screw.. )

    --> The Circuit has an on-board 12V, 4.5 Ah battery that im supposed to charge the battery with as well.
    --> Im using an AVR Microcontroller, ATmega8. It has 6 ADCs , an analog comparitor, PWM pins, and lots of other goodies..
    The system has a 8 x 250W panels as you can see. Its hooked up to a sunny boy 1700E grid tie inverter.
    --> The tracker will also have a learning capability , ie it will log data and will become more efficient as time passes by.

    IF IM MISSING A TRICK SOMEWHERE , PLEASE POINT IT OUT.
    The questions:

    1. im measuring the 10W solar panel voltage and current, for this im using a simple shunt, the INA128 instrumentation amplifier and reading the ADC value.any bright ideas on improving this setup are most welcome ...

    2. Please tell me about Solar tracking algorithms that you are aware of.. I have this idea of MPPT, i know its in the regulators.. but I think if I measure the current and voltage of the 10W panel .. and position the panels so I can get the Max power on the IV curves (ie max current and voltage value ).. That would be great.. I have also looked into other algorithms .. as well..

    3. I plan to use the LMD 18200 motor driver , since my actuator's reqs are so small.. anyone know of any protective circuitry in addition with the thermal and current flags on the LMD 18200 , Plz let me know..

    4. Was getting problems with counting the pulses of the reed sensor, but now I think using a Schmitt trigger OR RC filter with ADC monitoring would work..

    Please remember , I dont want any hardware that I dont really want.. The circuit has to be super efficient..

    5. For the battery, I want two charging stages. A simple charge.. and a float charge that I can apply with PWM. to protect the battery from overcharging, I will use a switch .. (in the high or low side, haven't decided yet.. ) . I need Ideas for the switch. The switch will remain closed for extended periods of time, I also want to apply float charge via PWM , so I think I have to use a high side P-channel MOSFET.. help needed with this one..


    Thanks for reading and answering .. sorry if i missed a imp detail ..
    PS dont dwell on the algos for too long..
     
  2. bertus

    Administrator

    Apr 5, 2008
    15,647
    2,346
  3. umer27

    Thread Starter Member

    Feb 28, 2009
    25
    0
    thanks .. a little more explanation would help .
     
  4. HarveyH42

    Active Member

    Jul 22, 2007
    425
    5
  5. bertus

    Administrator

    Apr 5, 2008
    15,647
    2,346
  6. Externet

    AAC Fanatic!

    Nov 29, 2005
    758
    57
  7. umer27

    Thread Starter Member

    Feb 28, 2009
    25
    0
    @harvey

    Yeah, thats me too.. its the same project..

    umer27 posted that on both forums..

    thats me !
     
  8. jpanhalt

    AAC Fanatic!

    Jan 18, 2008
    5,685
    900
    I still don't understand why one would go through the effort of designing a sensor-based tracker when all you need is a watch and a calender, as has been previously pointed out by others. And, the time-based system probably would work better under all conditions.

    There were some hand waving arguments given by umer27 to justify the sensor-based approach, so I reviewed the links provided above. The comparisons are between tracking and stationary panels. No distinction is made between the methods for tracking, except there is one comment that the sensor-based systems do poorly under adverse weather conditions (e.g., clouds and storms). That author goes on to say such conditions are rare in his part of Australia. Such conditions are not rare in the non-desert regions of North America, particularly on the western slopes of our hills and mountains.

    Helping to design an electronic gadget is what this forum is about, even if the gadget is useless (e.g., HHO PWM circuits). In this thread, I don't see any circuits proposed by umer27 for which he needs help.

    Is there something for which you need help?

    John
     
  9. umer27

    Thread Starter Member

    Feb 28, 2009
    25
    0
    @ john

    First of all , ......



    3. I plan to use the LMD 18200 motor driver , since my actuator's reqs (takes 2 A @ 12VDC) are so small.. anyone know of any protective circuitry in addition with the thermal and current flags on the LMD 18200 , Plz let me know.. anything that would always keep the IC from getting burned at all times..

    4. Was getting problems with counting the pulses of the reed sensor, but now I think using a Schmitt trigger OR RC filter with ADC monitoring would work..


    5. For the battery, I want two charging stages. A simple charge.. and a float charge that I can apply with PWM. to protect the battery from overcharging, I will use a switch .. (in the high or low side, haven't decided yet.. ) . I need Ideas for the switch. The switch will remain closed for extended periods of time, I also want to apply float charge via PWM , so I think I have to use a high side P-channel MOSFET.. help needed with this one..


    These questions first pls, I will answer the useful argument as soon as possible...
     
  10. jpanhalt

    AAC Fanatic!

    Jan 18, 2008
    5,685
    900
    As for the battery charger, your specifications are generically extremely common. Charge, then switch to trickle charge. Since you mention float charging, I assume you are using a lead acid battery. If that is not the case, float charging may not be the way to go. NiCd's and lithium don't like it nor, for lithium, is it needed; NiMH's are an open question, but probably should not be done.

    If you are using a lead acid battery, is its voltage 12 V? What will be your supply voltage? What is the current capacity of the battery?

    I eagerly await an answer to why.

    John
     
  11. thatoneguy

    AAC Fanatic!

    Feb 19, 2009
    6,357
    718
    Couple questions:

    Are you using a stepper motor in the linear actuator, and a magnetic reed switch reading magnet position on a shaft, or is it a standard motor told to run fast then stop?

    How is the seperate panel anchored to the main array to ensure alignment doesn't change? Is the power from the "sensor" panel used for anything other than sensing? Have you tried adding small (single cell) panels angled 10 degrees from each "horizontal to sun" plane? The reason for the last would be enabling the system to "see" a cloud bank incoming, and not scramble looking for the sun while under a cloud, also, it would give a tip to the processor as to a direction that won't work better without moving the entire array to find out.

    How does the system respond if the reed switch fails open or fails closed? Why not a hall effect sensor or enclosed optical rotary encoder? It sounds like you are having bounce problems with a brand new reed switch in a known environment.

    Regarding the battery, if it is intended to last a while, you don't want to treat it like a Car Battery. There are many charge control ICs on the market, such as used in UPS Power systems, that maintain a small Gel Cell or motorcycle size wet battery much better.

    --ETA: Regarding motor driver, you want a larger one (at least 2x). ALL sorts of things will happen when others are using it/changing an item, usually towards the overload side. The protection should NOT be "Motor driver protects the rest by burning".
     
    Last edited: Mar 6, 2009
  12. umer27

    Thread Starter Member

    Feb 28, 2009
    25
    0
    Im using a standard linear actuator. It has a stroke length of 900mm, Voltage 36 VDC and has a built in read sensor. current is about 2 A max .. which is very cool !

    The sensor panel is fixed on the same fixture and on the same alignment as the array.

    The basic idea is to find the MPP of the sensing solar panel for each orientation and get max power out of the array ..

    As far as cloud cover is concerned, The controller knows the sun will move from east (morning) to west (evening) so it doesn't scramble to look for the sun. Even in cloudy conditions, its moves west if its getting a higher MPP value than before. I hope this clears the confusion...

    @ John,
    Im sorry I forgot to mention, and its a sealed dry battery.. (lead acid) . It usually works well for 3-4 years before dipping in terms of performance..
     
  13. umer27

    Thread Starter Member

    Feb 28, 2009
    25
    0
    Regarding , how to measure the MPP of the sensing panel,

    I have an idea, need input from people on whether it would work..

    Im going to short the panel outputs .. not for very long , only for a very small amount of time.. there would be a 0.1 Ohms wire wound shunt in the circuit however...

    Now.. By, approx calculations, the MPP is at 90 % of the Isc and at 70-75 % of the Voc .


    what I needed to do was to vary the load resistance on the panel, and I would get the IV curve..

    what a resistor essentially does, is to limit the flow of electrons.. Im going to use PWM @ 3Khz or less to do just that .. lets say I take the PWM to 100% for about 50ms .. keep it there.. I should get Isc... right.. then I take the PWm to 0 % ... calculate.. Voc..

    I think if a vary the PWM little by little.. i.e take it from 0 to 100 % or vice versa .. and every 5 % or so, take a reading.. I should be able to generate the curve.. Im going to bulid a small model to test this theory .. Any ideas are welcome..

    We also have a sun simulator at the factory, It uses a neon lamp , flashing around 1000 Hz , and has a reference cell next to it (whose, IV curve is known), with this data , they calculate the MPP, Im going to find the MPP with the solar simulator.. then with my setup.. if the results are similar.. I win !
     
Loading...