Interesting Classic Float behavior with LiFePO4

[openplanet]

When my (Follow Me'd) Classics are in Float, and I start my a.c. vacuum cleaner, WbJr. shows -8 Amps from my LiFePO4 bank even though there's plenty of power available from the arrays.  This implies that Float is not behaving as it should; that is, the Classics are not sending available array power to the load.  But...over about 5-8 minutes the WbJr. slowly drops to -1 Amp, meaning that the load is mostly being powered by the array.  I'll add that if I turn the vacuum off for a couple minutes then back on, WbJr. once again shows -8 Amps, but it takes much less time to drop down to -1.
 
I'm not particularly bothered that it takes a few minutes for the load to be fully powered by the array rather than the battery bank. But I would like to understand it.  Does anyone have thoughts?

As an aside I'll mention that an added benefit of moving to LiFePO4 is that their capacity to absorb high current means that on sunny days they become fully charged sooner than did my FLA bank.  As a result I'm going to (finally!) set up an 80 gallon electric hot water tank as a dump (diversion?) load.  My plan is to send power to the tank by using Aux 1 Float High to turn on an SSR.  This is why I'm paying close attention to Float behavior ahead of time.

Thank you.

[boB]

It is great that you want to "understand" the behavior  !   This is something I always try to do when something comes up that doesn't make sense.

When it goes to Float and these loads are going to come on, turn OFF the loads for a moment and note the resting voltage of the battery system.  If the Classic's Float voltage is set for at least a few tenths of a volt above the resting voltage then you should be charging and supplying the loads I would think.

If the Classic is set to a voltage below the  voltage of your batteries then you'll be drawing from the batteries.

That is, unless you see the Classic go to "Float MPPT" in which case the Classic doesn't have enough power to power the loads and charge the batteries.   (or the Classic isn't tracking correctly)

[openplanet]

Thank you boB.  Yes, Float setting is slightly above resting voltage.
I may not have spelled out my question clearly.

With Classics in Float, 0 amps going into the batteries, but plenty of power available from the arrays, I'd expect that when I turn on the vacuum I would NOT see amps flowing FROM the batteries (or maybe just a momentary draw if the arrays can't handle the startup load).  Rather, I'd expect the load to be handled by the arrays.  But what I DO see is about 8 amps coming from the batteries, slowly dropping to about 1 amp over the course of 5 to 10 minutes.  So the end result is what I expect, namely the load being handled my the arrays.  But why the initial draw from the batteries, only slowly tapering down?  Could this perhaps be a result of the LiFePO4 properties?  Thx.

[ClassicCrazy]

I think it is because with lead acid batteries you turn on a load and the voltage drops - so the Classic starts putting in more power to hold the float voltage.
With the lithium the battery voltage doesn't drop right away and stays up so that is why the Classic behaves like it does with supplying power. It won't add more power if the battery voltage does not sag .

Larry

[FNG]

Some lithium prefer float to be very low, essentially at or around resting voltage, The issue here is when a charge controller goes into float it will product no power out until the battery falls below the target voltage. Essentially until then it sees the battery as over full. I know on the Fortress I am testing here float is still elevated enough so the solar controllers will continue to work and power all the loads all day

[boB]

When you turn on an AC load, the inverter will also suck some 120Hz ripple current from the system.

The voltage you read will be an average of some sort so it might be that the ripple causes some of this anomaly.

[australsolarier]

aha,
that is why the classics should have voltage sensors.
it is exactly what i am seeing. i suspect this is because of slight voltage drop under highish current and that stiff lifepo4 voltage. i also noticed it will then take a long time to bring it  back up to float. it probably does not affect lead acid batteries because of large voltage change.
all in all though it probably only affects 1 or 2% of battery capacity.

[openplanet]

OK, a few of you more experienced than I am have apparently zeroed in on a likely explanation.  My request: Could any you give me a somewhat "slower," clearer, more step-by-step, more detailed, more drawn out version of this explanation?!  I want to take advantage of the learning opportunity!
Thanks.

[australsolarier]

thicker wires between midnite and battery. though then you will find the terminals in the midnite have a certain limit of wire diameter.
i made tails at the end of the wire so they fit  into the termnal holes. however i spent more time on this than the rest of the midnite wiring.

one other thing you might  try is raising the float voltage one tenth of a Volt. (after calibrating the voltage) talking about calibration, did you calibrate the voltage? under no load?

[openplanet]

Yes, voltage is calibrated.

[australsolarier]

then, i checked the actual float voltage that the midnite  actually puts out.
in my case,
after calibrating the voltage,
and  trying to keep 53.6 V float
i actually hat to set the float settings to
1 midnite to 53,7V
3 midnites to 53.8V

all this doen't greatly affect lead/acid batteries. but it certainly has a much more leveraged effect on SOC of a lifepo4 battery.

i would like the jr whizbang junior shunt trying to compensate via current after float. at present the midnite reads the dropping voltage to ramp up.

[boB]

then, i checked the actual float voltage that the midnite  actually puts out.
in my case,
after calibrating the voltage,
and  trying to keep 53.6 V float
i actually hat to set the float settings to
1 midnite to 53,7V
3 midnites to 53.8V

all this doen't greatly affect lead/acid batteries. but it certainly has a much more leveraged effect on SOC of a lifepo4 battery.

i would like the jr whizbang junior shunt trying to compensate via current after float. at present the midnite reads the dropping voltage to ramp up.

Not sure I know what this means...

"i would like the jr whizbang junior shunt trying to compensate via current after float. at present the midnite reads the dropping voltage to ramp up."

Compensate current into the battery ?

In Float stage, only one Classic needs to be running usually to keep the Float voltage steady.  This is because the current requirement is typically so low.  When more loads come on and one Classic can't keep the voltage steady, then the other controllers should come on in MPPT mode to try and bring the voltage back up to set point.

Also, the voltage regulation to stay in Float or Absorb stages is not just one exact voltage pinned down to a say, 0.1 volt resolution, plus or minus 50 millivolts.   If that were the case, it would be turning on and off an on and off all way too often.  In fact, that used to happen some and we had to make it so that 2 or 3 tenths of a volt below that absolute Float or Absorb set point.  Below that, it goes to the Float MPPT or Bulk MPPT stage which is back to pulling all it can from the PV array.  At that point, NOW you want ALL the Classics and other chargers in the system to run and help out.   Except for generators and grid typically.


Now for the future and Lithium battery systems, if the battery voltage wants to be a bit more accurate, we are adding battery remote sense at high precision at least to the newer stuff.  If we can think of an easy way to add that to a Classic, I would love to do that.

[australsolarier]

"compensating current"
well, the midnite is in float, current goes out through the shunt from the battery to the inverter. there is plenty of solar there. but still the midnite does not compensate the full current for a while. (under heavy load it can draw SOC down to 95%. after the heavy load is turned off it takes hours to slowly return to 98% or thereabouts.) like before mentioned the lifepo4 V is very stiff. 0.2V difference can mean like 50% of SOC. that is why BMS's are  reading at least 2 decimals for balancing the cells. and that is in the near fully charged state where the voltage takes off like a rocket, so to speak.

under float typically the load is low or it is high or anything in between!   just think charging your car with 7kw. the midnites changing the loads from one to the other in mad succession ,  partly here partly there, maybe sharing a part for half a second. (this is really visible in absorption mode when whizbang jr current goes up and down rapidly, like from 5 to 12A)
in practice it doesn't matter all that much. (except for the end amps. i added 2A to compensate for that delaying effect)

and yes under light load one midnite out of four might generate 0A, but only for half a second, then it decides to butt in again. (just watched being on zero for five seconds)