1. Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.
    Dismiss Notice

A digital 16F886-based GTI

Discussion in 'Alternative Energy' started by khoa26189, Feb 3, 2012.

  1. khoa26189

    khoa26189 New Member

    Joined:
    Oct 4, 2011
    Messages:
    4
    Likes:
    0
    Hi all ! I am working on a GTI project for my thesis . I have successfully stepped up low voltage 12VDC to 350 VDC .

    The hard part is AC connection side . I used two mosfet drivers named Ir2101 control by opto-isolated signals from PIC16f886 (I don't draw them in the schematic). These two mosfet-driver is uesd to control a H-bridge in phase with the grid . I can create 240V AC from this H-bridge already .

    I try to test this H-bridge to see if it can withstand grid-voltage(220V AC) for some time(a cycle) for my zero-cross detection circuit to work and transmit zero-cross point timing to my PIC . I have carefully connectd a 100W light-bulb in series with one of grid line (so if anything goes wrong , there will no short-circuit explosion ) before doing this test.
    My problem is when I connected AC main into H-bridge output , the 100W light bulb lighted up which indicated a possible short-circuit .

    Can anyone give me some pratical tips on how to succesfully connect to grid if you already have 350 VDC ,a working H-bridge and a working zero-crossing circuit(I take this zero-crossing circuit form Microchip Grid-tie inverter application note) ?

    P/S : Vcc in this schematic is 12V . C5 and C6 is capacitors 104 (630V ) .
     
    Last edited: Feb 3, 2012
  2. tcmtech

    tcmtech Well-Known Member Most Helpful Member

    Joined:
    Feb 22, 2009
    Messages:
    6,659
    Likes:
    429
    Location:
    Sawyer, North Dakota, USA
    I have yet to ever get a reliable constant voltage DC source system to work well as the supply for the AC side of a Grid Tie Inverter system or at least not without loads of precision control and monitoring needing to be done to keep it stable which for me always just seemed like too much unnecessary work to do a simple synchronized feedback process.

    In grid tie you are working with what is essentially a constantly varying voltage and current load which is inherently difficult to to connect to a constant voltage source like what is used for the stable high voltage DC source on the output stages of a common 12 volt DC input to AC output power inverters.
    Although a common power inverter and a GTI are both inverters with similar circuitry layout the function of a GTI is entirely different than that of a power inverter which means what is common design in power inverters is not easily made applicable to grid tie systems. It can and has been done but not without a lot more work.

    To make a high voltage constant voltage source of power feed back into a variable voltage load as is what the grid essentially is in this application takes very precise PWM and filtering is needed which means a considerably more complex and adaptive digital control circuits and related systems to be required.

    I may not be a all knowing master of GTI design but I know from personal work and development simple rugged variable voltage variable current source based analog circuitry works very well in the power handling systems and digital works best in the safety and monitoring side of the system.

    that just my opinions.
     
  3. khoa26189

    khoa26189 New Member

    Joined:
    Oct 4, 2011
    Messages:
    4
    Likes:
    0
    Hi tecmtech !

    The constant voltage source (350 VDC) is made possible through an analog IC named SG3525 which will automatically increase/decrease duty cycle as the load varies . I have now sucessfully connect this H-bridge to grid without anykind of explosion by using BYM26E diode placed beween Vs of IR2101 and output of H-bridge .

    My question is : can one push current into grid by using 350V AC square-wave signal ? Or one need to synthesize a sinusoidal signal by PWM (varying duty cycle) to connect to grid ?

    P/S : the grid's nature (infinite current source) makes me a bit nervous . I think I can wait till my capacitor fully filled then push current into grid and then wait for it to fill . Is this ok ?
    P/S 2 : My design is based on Microchip Solar Grid-tie Micro Inverter Guideline here : http://www.electro-tech-online.com/custompdfs/2012/02/PV_AppNotes.pdf .Reading their app notes makes me think the concept is not that complex to implement .
    P/S 3: I like all of your simple GTI designs . When I finish this one , I will try your design .
     
    Last edited: Feb 4, 2012
  4. dave

    Dave New Member

    Joined:
    Jan 12, 1997
    Messages:
    -
    Likes:
    0


     
  5. tcmtech

    tcmtech Well-Known Member Most Helpful Member

    Joined:
    Feb 22, 2009
    Messages:
    6,659
    Likes:
    429
    Location:
    Sawyer, North Dakota, USA

    I follow how you are making the 350 volt DC source well enough but the concern I have is that with solar and wind your available input power is highly variable which is what I found is the least understood or most overlooked part by people wanting to design grid tie inverters.

    In a typical stand alone power inverter it gets its input power from a battery or similar DC source that is relatively stable and has minimal voltage drop regardless of the range load demand put on it.

    However with solar panels and wind power that source is not the least bit stable or constant both in voltage or available current which is where I have had difficulties understating why anyone would want to try and create a stable rail voltage like you are doing at 350 VDC to drive a grid tie system from an inherently unstable power source then go through the additional efforts to try to rematch those variable levels of available input power to inverter circuit and feed it back from there.

    Going from a variable input power source to constant level back to a output that has to be forcibly controlled to keep it properly proportioned in reference to the primary input source is not necessary and adds a middle stage that is not needed in my opinion.

    The method I have found to be the simplest and easiest to work with is to let the middle stage rail voltage, the 350 volt level in your case, simply float in relative proportion to the available input voltage and power opposed to trying to keep it at a constant voltage and then try and vary the output power being fed back to the grid from that.

    By eliminating the control loop that tries to produce a constant 350 volt middle rail voltage, peak voltage limiting at this stage may be necessary though, you can also eliminate part of the output stage control loop that has to be referenced to the input source. Both circuit parts counts drop and complexity drop while picking up some efficiency.

    If you are planing to use a SMPS type first stage to boost your input voltage up and provide line isolation, good practice for safety, you can set up your control circuit to basically float the mid point circuit voltage while still using your PWM wave form shaping in the output stage to get a clean and peak current limited/controlled waveform matching effect without having to do any referencing to the input power stage. If that makes sense.

    Basically you run it as two independent inverter stages. the first one does your primary voltage step up and isolation duty and the second one does all of the line end synchronizing, waveform shaping and power switching duties without concern over the input power source fluctuation other that simply being told when to connect and disconnect from the mains lines.
     
  6. khoa26189

    khoa26189 New Member

    Joined:
    Oct 4, 2011
    Messages:
    4
    Likes:
    0
    Iknow this design is way more complex than yours .
    This concept is from Microchip Grid-tie Design Guideline : trying to create a constant high-voltage source from highly variable power source then fed that power into grid .
    Everything seems fine till now , I need some tips on what to do next . Should I create square wave or modified sine-wave 350V AC to be able to feed power into grid ?
     
    Last edited: Feb 4, 2012
  7. tcmtech

    tcmtech Well-Known Member Most Helpful Member

    Joined:
    Feb 22, 2009
    Messages:
    6,659
    Likes:
    429
    Location:
    Sawyer, North Dakota, USA
    If you have the means to do PWM wave form shaping to shadow the actual grid wave form then by all means do so.

    Dumping an actual square wave directly into the grid does not work so well unless you have a fairly robust LC tank circuit or transformer to run it though to clean it up before it hits the actual line.

    Doing PWM waveform shaping/shadowing will cut down on your potential line distortion considerably and will also let you get by with considerably less post LC waveform clean up before your power hits the main grid.


    My concern/curiosity is how you intend to reliably limit the grid side feedback power in relation to available DC side input power while maintaining your 350 VDC mid point voltage.

    If it was me I would be tempted to run the output switching circuit in a wide square wave mode with a low zero cross dead band but control the effective feedback power by modulating the first stage inverters PWM duty cycle to produce a shadow copy of the grid side waveforms. By doing so the the four switching devices on the HV inverter side would be seeing minimal turn on and turn of current peaks which would likely pick up a bit of efficiency and greatly reduce the chance of device overload.
    Granted doing so would have you modulating your 350 VDC midpoint voltage from near the zero dead band cross over voltage point to slightly over what ever your peak grid side line voltage is at double your line side frequency but in exchange will reduce your second stage inverter switching losses and peak switching currents while also reducing the amount of line side filtering and clean up work as well.

    In a way it would be turning your first stage voltage step up circuit into a proportional current control that limits the DC side power draw to what ever is available from the source without excessive amounts of second stage control work. Granted you loose your stable 350 volt mid point DC rail but your overall control circuit process loop issues drop off considerably in exchange.

    Its just something to think about.

    As far as keeping the constant 350 VDC mid point in place I just see it as an unnecessary step that just puts additional parts and work load on the control and switching circuits without any justifiable gains in the end. It certainly can be done but I am not the one to talk to about how to refine it.
     
  8. khoa26189

    khoa26189 New Member

    Joined:
    Oct 4, 2011
    Messages:
    4
    Likes:
    0
    Since my high-frequency high-voltage AC is generated by analog IC , I can't create a positive part of a 50Hz sine-wave for my H-bridge to slowly unfold and flip it to fed into grid . Replace that analog IC will delay my Thesis and the dead-line is near . I can't do that . My only option now is to create a 50 Hz square-wave signal and use LC tank to shape it .

    Any advice on how to design an robust LC tank for this task , tcmtech ?
     
  9. tcmtech

    tcmtech Well-Known Member Most Helpful Member

    Joined:
    Feb 22, 2009
    Messages:
    6,659
    Likes:
    429
    Location:
    Sawyer, North Dakota, USA
    Basically standard inductor and capacitor impedance calculations will get you close enough to work. After that its just a matter of finding components that are physically capable of handling the power levels they will be dealing with in energy transfer or filtering losses.

    Your theoretically ideal L and C values may come up a bit off from whats most commonly available so don't get too wound up if you have to cheat the values slightly higher or lower to best match whats available in manufactured components.

    Whether you lean your values slightly high or low is up to you but don't be surprised when you build the actual unit that the as built and as tested numbers may come up noticeably off from the theoretical numbers. ;)
     

Share This Page