[RE-wrenches] Fwd: Re: OB Battery Charging

[Michael Welch]

Hi Wrenches. Here is a post from Julie Haugh, who uses the archives. If you respond, be sure to include Julie's address as she is not on this list.

From: jfh at greenhousepc.com


R Ray Walters writes:


> I still haven't figured out just how the GVFX works, (i.e. when it 
> goes thru a full charging sequence, whether it uses the renewable 
> sources or the grid, etc)
> because the AC Kwh meter wiring configuration I used doesn't show AC 
> input to the batteries.
> but what ever it does, it works.

The "Support" and "Sell" functions are driven by battery voltage, much the same way that an MPPT charge controller works, only in reverse.  The "Sell RE" voltage is the target voltage for the battery and the inverter works to keep the voltage from rising above "Sell RE".  As the voltage rises, the inverter output increases as well.  If the voltage falls, the inverter's output falls as well.

The decision to use the grid directly is likewise based on battery voltage -- when the battery voltage rises above "Sell RE", the inverters begin to produce power (all other things being equal -- there are Mate settings which can affect this).  Prior to this the inverters are in "Pass Thru" mode and there is no inverter output.  The Mate will then show "Support" for the inverter mode.  Increasing the power output requires increasing the voltage to "force" the power backwards through the grid.  When the current begins to flow out through the AC in terminals, the inverter changes from "Support" to "Sell".  This is the goal -- keeping the voltage from rising above "Sell RE" -- and the only limitation is inverter output and grid voltage.  For those of y'all installing Xantrex inverters, they have similar settings and can have similar problems.

"Charge" decisions are based on two different voltage setpoints -- "Rebulk" and "Refloat" as well as timers which track how long the battery voltage was below the setpoint.  "Refloat" can be configured via the Mate, "Rebulk" is fixed at 12.4, 24.8 and 49.6 volts and has a 90 second timer.  It can get complex, but I can explain if needed.

> Anyone have a better meter configuration? We're using the recommended  
> 5 jaw digital meter from Austin International, and the base is wired  
> per Outback's white paper on the subject.
> It cancels energy readings that just pass thru to the loads and  
> tabulates sell back, but it doesn't seem to register AC use from  
> battery charging. (my system shows 0 Kwh after 2 years)

For a single inverter installation, a 5 jaw meter wired as Outback shows is best.  For digital meters, such as the one you have (I'm assuming you have one as it apparently has multiple registers for exported (sold) and imported (charged) power, you may have needed to request that "negative" power flows be supported.  Some vendors ship their meters by default with that feature turned off because it's an unusual situation.

The Outback whitepaper doesn't work for dual G-series inverters and you'd need to use a revenue grade meter that supports CTs.  My recommendation is E-mon D-mon meters.  They have a "Green Class" meter that supports measuring bidirectional power and you can put the split core current sensors around multiple conductors.  Plus, they have been certified as "utility grade" meters.

In a different posting, Ray also wrote --

> I think that periodic Equalization is "old school", a concept left  
> over from the days of C30 controllers that didn't fully charge the  
> batteries on a regular basis.
> I now only recommend Eq charging if:
> a) the batteries have sat for over a month in a deeply discharged  
> state, or
> b) checking individual cell voltages indicates more than a 2%  
> difference between batteries. (over a tenth on a 6 v battery)

The need to equalize isn't just from excessive discharge, it can also occur on batteries that are not heavily used.  Both charging and discharging result in the electrolyte circulating within a flooded batteries as the density of the electrolyte changes.  Grid-interactive systems can get into this situation (more in a second) because the batteries are not typically being actively used.  In a typical grid-interactive installation, the battery voltage is raised to "Sell RE", it doesn't rise above "Sell RE", and then it naturally falls back to its resting voltage, waiting for the next day to happen.  Net current into and out of the battery bank is minimal because "Sell RE" is typically set close to "Float", which is a net zero current maintenance voltage.  Meaning, current just keeps up with self-discharge.  Without the circulation that happens due to changing electrolyte density -- charging and discharging -- the electrolyte in flooded batteries can stratify.  That's the other reason for equalization.  It's my understanding that chronic undercharging is a problem with grid-interactive systems for a variety of reasons.

Finally, the internal self-discharge rate varies between batteries, resulting in inequalities between cells and blocks.  Equalization charging is the only way to correct cell-to-cell SG differences as well as battery-to-battery voltage differences.  It's not needed as much as it was in the past when chargers were dumb, but it's still needed.

In another post Ray writes:

> In 7 years of Outback inverter installations, we've never had to go  
> out to adjust something. The Outback defaults are pretty good out of  
> the box, and not including the Mate keeps them from messing with the  
> settings.

The default values are conservative to the point that chronic under-charging is possible for some configurations.  In particular, when the ratio between array size to battery bank is small -- low percentages for available current at the "Absorb" voltage -- the absorb time limit needs to be set longer than the defaults.  The default is 2 hours, which for a system where absorb current at the start of the absorb cycle is not much higher than the typical 2% return amps value for the end of "Absorb" is entirely too short.

Another area where the default values result in chronic under-charging is the "Sell RE" voltage.  The default is 13 / 26 / 52 volts.  The correct value has to be determined experimentally and adjusted quarterly as the Sell RE voltage is not temperature compensated.  Those values might be acceptable for summertime "Float" values, but for winter temperatures the batteries will be seriously undercharged.

Wallace Stahle writes (responding to a post by Ray) --

> That works on 2 volt cells but do you occasionally verify your conclusions
> with a hydrometer?
> 
> I have been these days and discovering that "fully charged" parameters dont
> always translate into fully charged battery chemistry.
> 
> Still fine tuning my understanding here.

There's no substitute for periodic specific gravity testing -- unless one has AGM or GEL batteries, then load testing and resting voltage have to suffice.  But, yeah, "fully charged" according to some model (or according to an FN-DC) does not mean the batteries really are "fully charged."

Without testing SG, the easiest tests are the morning (pre-dawn) temperature compensated battery voltage for backup and grid-interactive systems, and dropping the grid on a grid-interactive system while selling and monitoring the time to reach "Absorb".  A fully charged battery will have a significantly higher morning voltage, and will reach "Absorb" significantly faster (seconds, typically), than a battery that is actually less charged than indicated.  A battery monitor that records minimum temperature is a good substitute for waking up early.  A monitor that records minimum temperature compensated voltage is even better.

A more difficult way is looking at the net current to the batteries during an "Absorb" cycle.  The average net current over a 15 minute interval should not decline when "Absorb" is actually complete and the batteries are as 100% full as possible.  The only exception to this rule is when the charge controllers are struggling to support both loads and battery charging.  Plotted at 15 minute intervals, the charging current should rapidly fall off, leveling off over a period of time.  The flatter the line at the end of "Absorb" the more fully charged the batteries.
--
Julie Haugh
Senior Design Engineer
greenHouse Computers, LLC