HELP:Solar MPPT->Battery->Inverter

[NorthGuy]

But during float ,the voltage is to be reduced to 53v by increasing duty. If there were no load on the charger won't the battery try to over charge or boil?

No. When the battery is charged, any current into the battery will increase its voltage. If you keep voltage at 53V, it'll be nearly no current into the battery.

When there's no load, you'll be producing very little (if anything). It is achived by decreasing duty and thus driving the panel voltage way above MPPT point.

If there's a load, you produce just enough to feed the load with battery doing notthing.

If the load demands more than you can produce, you go to MPPT, produce your best, but the battery voltage still falls and the battery starts discharging - typically you get this when sunset is near.

BTW: You can't really charge batteries with panels with Vmp=60V. You need more voltage. Your best bet is to rewire your panels into two strings (each string with 4 panels in series). It'll be even better to buy another panel and re-wire them as 3x3.

 

[abicash]

understood
Thanks a lot

 

[Tony Stewart]

No. When the battery is charged, any current into the battery will increase its voltage. If you keep voltage at 53V, it'll be nearly no current into the battery....

BTW: You can't really charge batteries with panels with Vmp=60V. You need more voltage. Your best bet is to rewire your panels into two strings (each string with 4 panels in series). It'll be even better to buy another panel and re-wire them as 3x3.

More? Actually, MPPT is normally 80% of 67 Voc = 54V not 60V.or 90%.
So it is well matched as is with just a PWM buck reg..

 

[NorthGuy]

More? Actually, MPPT is normally 80% of 67 Voc = 54V not 60V.or 90%.
So it is well matched as is with just a PWM buck reg..

54V is not enough. It is some drop in the controller, so your 54V becomes 53V. If you only have 53V, you cannot properly charge the battery.

60V would be slightly better, but still not enough because the conditions are not always perfect, and Vmp goes down substantially when the panels get hot (as they normally do).

Batteries need to be equalized from time to time, and this requires 62V or more. Add to this controller drop and temperature drop and you get a number around 75V.

OP has an MPPT controller, so higher voltage is not a problem. Getting 3 panels in serieas (rated Vmp around 90), or 4 (rated Vmp around 120) ensures that the controller can always produce enough voltage for charging.

 

[Tony Stewart]

Edit: delete calc error

This results in ;
13.5V ( "12v") or 54.0V ("48v") for 100%SoC at rest and
14.2V ("12V") or 56.8V ("48v") for max charge voltage.

MMPT is closer actually to 78% of Voc +/-5% which is fairly flat and the loss of 1V or 2% of power in NorthGuy's arbitrary regulator is significant.

I would challenge anyone to prove you can get better efficiency than direct charge from Array to battery with no drop and only OCP and OVP with Voc =1.25* V(100%SOC).

I think the advantage of MPPT is that when you oversize the PV to the system, you can fully charge in less time in case of low time duration for full sun.

All you need an extra 5% Voc for PV panel or Vmpp to increase charge current with a discharged battery. There is no need to choose a MOSFET with more than ~50~100mV drop for OVP at worst case current.

I believe , the requirements are that the battery is not drained less than 2.875/cell nor charged more than 3.550/ per cell.with 100% SoC at 3.375V/cell.

 

[NorthGuy]

It's pure mathematics.

MPPT: you get Vmp*Imp*Efficiency (97% or so)
No controller: if (Vmp > Vbatt) you get roughly Vbatt*Imp, very little otherwise.

Assuming you keep Vmp > Vbatt without controller, the difference is:

Vmp*Imp*Efficiency - Vbatt*Imp = Imp*(Vmp - Vbatt) - Vmp*Imp*(1-Efficiency)

To make it even, you must make these terms equal, or Efficiency = (Vmp-Vbatt)/Vmp. To beat 97% efficiency, you must have your Vbatt just 3% below Vmp. With both Vbatt and Vmp having considerable variation and such a small difference between them, it is impossible to avoid (Vmp < Vbatt) situations, in which case you lose most of the production. Typically, people with PWM (no MPPT) controllers, use 18V (Vmp) panels to charge 12V (nominal) batteries to avoid the (Vmp < Vbatt) situation.

Cannot understand what you say about voltages. 3.55V/cell will definitelty boil it out very quickly, my guess is within minutes.

 

[abicash]

In my MPPT , the algorithm constantly tracks by one reducing and once increasing the MPP .
With load on inverter, if MPP is at 53v and battery is at 49v, the MPPT adjusts the panel correctly at 53v.
But without a load on the inverter, the reduction of voltage on mppt never works. Since maybe it is tied to battery voltage.
So I think, the MPPT charger will work from battery voltage and above ,irrespective of MPP volts, if there's no load on the charger.
Am I right?

 

[Tony Stewart]

3.55V/cell is what most cars are tuned to for alternator voltages (14.2V)

Most modern MPPT's are around 93-97% efficient in the conversion, but I don't know the cost difference.

I took worst case bottom range and showed about 6~7% below peak to see the range of battery voltages.

In this system , the result was 52V to 66V, which covers the required range.

So efficiency needs to be checked carefully in MPPT vs losses and never assumed it is always better than a lossless CV regulator until near full charge if there is surplus power, although this is for shallow discharge systems that extend battery life.

If the battery is discharged deeply then MPPT charger will certainly drop the voltage with some regulator loss to improve charge rate and capacity.

 

[NorthGuy]

Car battery has 6 cells, so 3.55V/cell would be 21.3V.

I agree that inefficient MPPT controller is useless.

 

[Tony Stewart]

oops edited calc error... corrected below

Flooded Acid Battery from my experience...

1 cell 6 . 24 .%of Vmax State
2.367 14.2 56.8 100% V_Max Charge
2.250 13.5 54.0 95% Float Charge
2.083 12.5 50.0 88% Rest 100% SoC
2.000 12.0 48.0 85% Rest 50% SoC
1.917 11.5 46.0 81% Rest <10% SoC

For good maintenance, keep a log book of Specific Gravity for each cell, (s.g.) and record rest V.

When batteries get Sulphated, ESR rises rapidly and s.g. drops. resulting in weak link and more rapid ageing.

This can be avoided by avoiding long durations near or below 10% SoC and avoid any temperature rise above room temp.

s.g. Correction** can be done using pulse charging > 10kHz with several Amps. ~1us duration or 1% duty cycle while charging. ( We used to build these for Solartech and I verified on Motive Power batteries that it works.

ref. Solartech **

 

[Tony Stewart]

In my MPPT , the algorithm constantly tracks by one reducing and once increasing the MPP .
With load on inverter, if MPP is at 53v and battery is at 49v, the MPPT adjusts the panel correctly at 53v.
But without a load on the inverter, the reduction of voltage on mppt never works. Since maybe it is tied to battery voltage.
So I think, the MPPT charger will work from battery voltage and above ,irrespective of MPP volts, if there's no load on the charger.
Am I right?

If MPPT cannot tell if battery is discharged just heavily loaded, it does not matter. But if there is no load and low PV power avai, l PMMT hunting CAN cause instability with hunting or oscillation. Since the dielectric has like a capacitor absorption loss with a long time constant, ( several minutes with no load) tracking may have difficulty finding a sharp power peak as ESR is higher in batter when low in charge and MPPT point drops in Voltage.

It obviously depends on design nitty gritty details of MTTP under adverse conditions..

Courtesy; ZyMOS wiki

 

[NorthGuy]

In solar setup, battery is never at rest, so rest voltages are pretty much useless. If voltage is lower than some threshold (say 53), the controller should go all out. It might be charging a battery, or supporting big loads, but it doesn't matter - in either case more energy is needed.

Good control algorithms don't hunt.

 

[Tony Stewart]

When there is hysteresis in power measurement response then the algorithm will have to choose an average to avoid error or a relaxation oscillator effect.

Hysteresis is when the peak power moves depending on if you sweep with voltage rising or falling. If current rises with voltage and falls exactly the same, it is predictable and stable. If current rises when voltage is reduced ( by an equal amount i.e. constant P) and then rises when voltage is increased ( i.e. rising power) at the same operating point, then you have hysteresis. So time response and non-linearity of system, memory effects etc, all contribute to hysteresis and potential hunting. the "gain" of the system also changes the stability margin of the system for different loads and PV % power available as the MPPT point shifts downward with P as much as 25% but still follows closely to around 78%+/-5 of Voc.

Not saying an optimal algorithm cannot be achieved, but saying there are plenty of opportunities for failure and testing for operating point stability and regulator losses must be considered in any overall system.

 

[NorthGuy]

Commercial MPPTs are pretty good, even with partial shade and such.

The response time is quite fast and hysteresis is almost non existent. Your turn FET off - panel voltage starts increasing in less than 1us. You turn FET on - panel voltage starts decreasing equally as fast. If your input capacitor is not too big, you can get the full I-V curve in a matter of milliseconds. Lighting conditions don't change nearly as fast.