Announcement

Collapse
No announcement yet.

Current Flow Through Grid-Tied Outback Radian

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Current Flow Through Grid-Tied Outback Radian

    The blue plot is what 30A "DC" current flow looks like through my Outback Radian GS8048A in grid-tie mode. The green plot is the current flow from a Morningstar TS-600-60 charge controller working as hard as it can in MPPT mode to supply all the current demanded by the Radian.


    shunts.png
    The waveforms were captured with my two-channel digital storage oscilloscope, stored onto a flash drive and read onto my PC, and then run through a 1024-tap digital lowpass filter that I implemented in Python to get rid of the considerable noise that was present for the low-amplitude voltage developed across a shunt with only 100 micro-ohms of resistance. The plot is scaled to Amperes.

  • #2
    Interesting. Would you mind posting the unfiltered low voltage data for this timescale too? What is the cutoff frequency you used in the filter?
    CS6P-260P/SE3000 - http://tiny.cc/ed5ozx

    Comment


    • #3
      Nice information. I am thinking, the Radian is relying entirely on the battery to filter the power peaks.
      Saves them some cost, the panel to grid tie inverters here have to do it internally.

      The charge control current on the panel side should be constant; ideally it vary some on the battery
      side because of ripple on the battery voltage. But my guess is the MPPT is too slow to track the
      ripple, so the result is variation from the ideal MPPT point over each cycle. That could be verified
      by checking the current on the panel side. Guess I could do that here, to see how well my inverters
      stay on MPPT. Bruce Roe

      Comment


      • #4
        Originally posted by bcroe View Post
        Nice information. I am thinking, the Radian is relying entirely on the battery to filter the power peaks.
        Saves them some cost, the panel to grid tie inverters here have to do it internally.
        I think so, too. You need a huge amount of capacitance to filter a hundred amps for even 1/120 of a second.

        Originally posted by bcroe View Post
        The charge control current on the panel side should be constant; ideally it vary some on the battery
        side because of ripple on the battery voltage. But my guess is the MPPT is too slow to track the
        ripple, so the result is variation from the ideal MPPT point over each cycle. That could be verified
        by checking the current on the panel side.
        Yes, and I doubt that there's enough capacitance on the charge controller, on either the panel or the battery end of the buck converter, to isolate the panels from this 120 Hz voltage variation. The CC operates with a switching frequency in the kHz range, so the input capacitance just needs to supply a PV current path for the millisecond or less that the panels are switched off from the series inductor(s). The output capacitance is only there to limit the ripple, again at the relatively high switching frequency. Also, Morningstar has a patent for a charge controller with multi-phase switching with one stated goal to reduce the amount of filtering required.

        Here's another plot I did with less filtering (N=512 FIR filter with fc=3600 Hz, transitioning to a stopband starting at 10800 Hz and weighted at 10000x using the Remez exchange algorithm). There's also a fix in how I was computing the shunt currents. The voltage measurements with respect to ground are very confusing in the Outback GSLC, with the inverter shunt going from inverter negative (and ground) to the battery busbar, and then the charge controller shunt(s) going from there to the CCs. So you have to subtract the first reading from the second, without knowing exactly what's being diverted to the battery. Don't forget the slight offset voltages and gain mismatches of an inexpensive scope running at 2 mV/div sensitivity.


        shunts.png
        The blue plot is inverter current, measured from the scope's CH1 vs. ground. (I suspect the minimum current is really closer to zero than the apparent 4 A.) The green plot is the charge controller current, calculated from CH2 minus CH1. The red plot is the computed inverter current minus CC current, and supposedly shows the current sloshing back and forth in the battery.

        What this all shows me is that there is a lot of microcycling going on with my battery as the system sends power out to the grid, and I don't like that at all. I've been doing simulations on a bank of eight 100,000 uF electrolytic capacitors (rated to 80V, 5.5 milliOhm ESR apiece, about $40 each) that I will probably be hooking up in parallel to the battery to shunt almost all of that AC away from it. It looks like it will reduce it to less than 20% of what it is now.

        Comment


        • #5
          Switchers tend to handle current in gobs, multi phase is a good DC-DC solution. Depending on the bat for filtering
          may not be a bad solution, the phone co has been using it to filter 3 phase rectifiers forever apparently without
          loss of life. Slapping on more capacity eventually hits diminishing returns, the caps must have very low Z and
          the wiring layout gets critical. The final fix is insert another switcher drawing a slowly varying battery current,
          charging a cap to feed the inverter. That is actually what is going on in my grid tie inverter. Bruce Roe

          Comment


          • #6
            Thanks, BackwoodsEE for the plots, they show me yet again that the devil in the details. Here are the current waveforms for my off-grid 24VDC to 240VAC off grid inverter



            Which as you would expect show the same thing. I wonder how much this reduces the battery lifespan?

            I have often wondered how the straight grid tie inverters without any battery to smooth out the AC current waveform are made to work with solar panels. I assume it is a variation on what bcroe is saying in the previous post. By running high DC voltages one cuts down the magnitude of the AC current but higher voltage capacitors are usually larger and more expensive.

            In a battery based system if the solar charge controller is running in CV mode rather than MPPT mode it would be nice if the frequency response of the charge controller was high enough to transfer the AC current waveform to the solar panels. I wonder if there are any commercial units out there doing something like this?

            Simon
            Off-Grid LFP(LiFePO4) system since April 2013

            Comment


            • #7
              Here is an investigation that discusses the effect of microcycles on capacity of lead acid batteries, but doesn't look hard at long term lifetime:

              http://www.academia.edu/18165112/Ana...energy_systems

              We saw what the ripple looked like that passed through to the PV side in the previous mppt thread BackwoodsEE had started, for an entirely off-grid small setup (visible, but not significant). Does this higher power / higher voltage version behave differently?
              Last edited by sensij; 12-20-2017, 11:01 AM.
              CS6P-260P/SE3000 - http://tiny.cc/ed5ozx

              Comment


              • #8
                All hybrid inverters rely on a minimum size battery bank, mainly for ripple reduction. I've not seen hybrid owners complaining about shortened battery life.

                My XW has a fault code for over heated capacitor.
                Powerfab top of pole PV mount (2) | Listeroid 6/1 w/st5 gen head | XW6048 inverter/chgr | Iota 48V/15A charger | Morningstar 60A MPPT | 48V, 800A NiFe Battery (in series)| 15, Evergreen 205w "12V" PV array on pole | Midnight ePanel | Grundfos 10 SO5-9 with 3 wire Franklin Electric motor (1/2hp 240V 1ph ) on a timer for 3 hr noontime run - Runs off PV ||
                || Midnight Classic 200 | 10, Evergreen 200w in a 160VOC array ||
                || VEC1093 12V Charger | Maha C401 aa/aaa Charger | SureSine | Sunsaver MPPT 15A

                solar: http://tinyurl.com/LMR-Solar
                gen: http://tinyurl.com/LMR-Lister

                Comment


                • #9
                  Originally posted by karrak View Post
                  I have often wondered how the straight grid tie inverters without any battery to smooth out the AC current waveform are made to work with solar panels. I assume it is a variation on what bcroe is saying in the previous post. By running high DC voltages one cuts down the magnitude of the AC current but higher voltage capacitors are usually larger and more expensive.

                  In a battery based system if the solar charge controller is running in CV mode rather than MPPT mode it would be nice if the frequency response of the charge controller was high enough to transfer the AC current waveform to the solar panels. I wonder if there are any commercial units out there doing something like this? Simon
                  Best panel performance is at MPPT, DC, no ripple. In a straight grid tie, the MPPT input no doubt
                  has some energy reserve to feed the line waveform generator. But a lot of ripple in this intermediate
                  stage should not pass to either the input or the output. The output stage can compensate for a lot of
                  ripple in operation. Bruce Roe

                  Comment


                  • #10
                    Originally posted by karrak View Post
                    Thanks, BackwoodsEE for the plots, they show me yet again that the devil in the details. Here are the current waveforms for my off-grid 24VDC to 240VAC off grid inverter



                    Which as you would expect show the same thing. I wonder how much this reduces the battery lifespan?

                    I have often wondered how the straight grid tie inverters without any battery to smooth out the AC current waveform are made to work with solar panels. I assume it is a variation on what bcroe is saying in the previous post. By running high DC voltages one cuts down the magnitude of the AC current but higher voltage capacitors are usually larger and more expensive.

                    In a battery based system if the solar charge controller is running in CV mode rather than MPPT mode it would be nice if the frequency response of the charge controller was high enough to transfer the AC current waveform to the solar panels. I wonder if there are any commercial units out there doing something like this?

                    Simon
                    Like Backwoods that is some really nasty ugly power. Distorted and full of harmonics. Yours is so nasty the the scope can not sync up correctly.
                    MSEE, PE

                    Comment


                    • #11
                      sensij, Thanks for the link to that paper. Interesting. Here's another one by Carsten Ropeter that reports some hands-on analysis of lead-acid batteries. The paper's conclusion includes the following points:
                      1. Microcycles cause and/or increase existing acid stratification.
                      2. Charging characteristics capable of achieving a full charge under normal conditions are no longer capable of restoring full charge and removing acid stratification fully. Cells with significant acid stratification require more time for recharging. A different end-of-discharge criterion has to be chosen when a battery is microcycled. This leads to significantly longer charging times.
                      3. If the charging characteristics are not changed, cells which are microcycled loose [sic.] capacity and have acid stratification even after the end of charging.
                      4. The loss of capacity can be removed by electrolyte circulation.

                      Ropeter explains "the effects of microcycles on batteries by inhomogeneous current distribution during the charging and discharging phase."

                      I hope to have my PV analyzer box hooked up to one of my 12-panel strings soon and will report back then on what kind of current ripple I'm seeing during the IV sweep. Assuming that the sun will eventually come out again someday here in Eastern Washington.

                      Comment


                      • #12
                        Originally posted by BackwoodsEE View Post
                        sensij, Thanks for the link to that paper. Interesting. Here's another one by Carsten Ropeter that reports some hands-on analysis of lead-acid batteries. The paper's conclusion includes the following points:
                        1. Microcycles cause and/or increase existing acid stratification.
                        2. Charging characteristics capable of achieving a full charge under normal conditions are no longer capable of restoring full charge and removing acid stratification fully. Cells with significant acid stratification require more time for recharging. A different end-of-discharge criterion has to be chosen when a battery is microcycled. This leads to significantly longer charging times.
                        3. If the charging characteristics are not changed, cells which are microcycled loose [sic.] capacity and have acid stratification even after the end of charging.
                        4. The loss of capacity can be removed by electrolyte circulation.



                        Ropeter explains "the effects of microcycles on batteries by inhomogeneous current distribution during the charging and discharging phase."

                        I hope to have my PV analyzer box hooked up to one of my 12-panel strings soon and will report back then on what kind of current ripple I'm seeing during the IV sweep. Assuming that the sun will eventually come out again someday here in Eastern Washington.
                        You are correct with a couple of caveats.

                        Stratification is more problematic on cells where the height is greater than width and length. Example L16 type cells.
                        AGM stratification is not an issue

                        To add to what you have said about micro cycles, to prevent stratification requires the battery to be charged at a minimum rate of C/12, and this is the biggie, Absorb stage equal to or greater than GASSING VOLTAGE. Having said that stratification is not a big problem if proper charge rates and EQ cycles are performed as required. Thus is why I tell folks do not use battery manufactures recommended charging voltages, use a hydrometer to find the right Bulk/Adsorb voltage. You want to see bubbles when the battery charges like a good champagne.
                        MSEE, PE

                        Comment


                        • #13
                          Below is a plot generated from the simulation I developed of a 6-capacitor ripple filter I'm thinking of putting together. Each capacitor is a United Chemi-Con 100,000 uF electrolytic, rated at 80V with 5.5 mOhm ESR. The simulation has them connected in two lines to a common negative bus bar and two separate positive busbars. The busbars are 1.5" wide aluminum, 1/8" thick, and about 12" long, and contribute no significant resistance or inductance to the circuit (though I did model their resistance).

                          The positive busbars connect to a pair of ganged 80A breakers via two 18" lengths of 4 AWG wire. The negative busbar connects directly to the GSLC's current shunt busbar, where the battery negative connects, via a single 2 AWG wire. The other ends of the 80A breakers connect to the GSLC's positive plate via two 8" lengths of 4 AWG wire, bundled as closely as possible to the wires leading to the other ends of the breakers to minimize loop inductance. (I modeled this as a 5 mm average gap, plus insulation thickness.) The breaker resistance was modeled as 2 mOhm.

                          The battery bank consists of eight Rolls S6-460AGM batteries connected in series by seven 12" lengths of 4/0 cable, with the overall negative lead being 60" of 4/0 and the positive being 36" of 4/0. This is both what I have out in the equipment shed, and what I modeled in the simulation, including self-inductance. For the batteries, I used a 2nd order Randles model with R1 = 1.3 mOhm, and R2 = 0.3 mOhm in parallel with C = 200 Farads for double-layer capacitance. This was selected to have my simulated battery voltage drop match what I observed while keeping R1+R2 at the 1.6 mOhm specified in the Rolls datasheet.

                          Here's the result:

                          mcf.png
                          The ripple current in each capacitor is well within the 30A specification, and power dissipation is around a Watt and a half apiece. They will barely get warm under full load. Power dissipation in the battery will be reduced down to less than 10W peak, probably a tenth of what is being burned up in the batteries now due to the microcycling going on without any ripple filtering.

                          As bcroe pointed out, there are diminishing returns for using more capacitors. Based on the top subplot and the fact that ripple current per capacitor is very manageable, I see little reason to use eight of them as opposed to six.

                          Comment


                          • #14
                            How do you treat the aluminum bus bar so it does not instantly oxidize and become a resistive connection ?

                            Did you include any provision for fusing the caps, or just rely on their leads ?
                            Powerfab top of pole PV mount (2) | Listeroid 6/1 w/st5 gen head | XW6048 inverter/chgr | Iota 48V/15A charger | Morningstar 60A MPPT | 48V, 800A NiFe Battery (in series)| 15, Evergreen 205w "12V" PV array on pole | Midnight ePanel | Grundfos 10 SO5-9 with 3 wire Franklin Electric motor (1/2hp 240V 1ph ) on a timer for 3 hr noontime run - Runs off PV ||
                            || Midnight Classic 200 | 10, Evergreen 200w in a 160VOC array ||
                            || VEC1093 12V Charger | Maha C401 aa/aaa Charger | SureSine | Sunsaver MPPT 15A

                            solar: http://tinyurl.com/LMR-Solar
                            gen: http://tinyurl.com/LMR-Lister

                            Comment


                            • #15
                              A fun exercise; I think you will find, the real world is never quite as good as the model. Sometimes
                              the connections to a bus bar have more resistance than the bar. Make sure the filter input leads
                              have no common wire with the filter output leads. Can't do much about the cap resistance, except
                              choose a good cap. Bruce Roe

                              Comment

                              Working...
                              X