Battery "Bounce"

How about equalizing the resistance of the two runs?

Not really practical we are talking super low resistance here. (.0024 ohms at up to 1,000 watts.)

Thinking about what you said then perhaps it is a GREAT idea. perhaps a couple of 200 amp 75mV ammeter shunts in series would work ?

I use a Perko switch for the inverter power main/standby selector. http://www.perko.com/catalog/category/battery_switches/product/148/
A similar model should work for your application. A serial string of batteries works better for me than parellel sets
(my old batteries) so I would use lower voltage cells (large 2 or 6 Volts units) to make the new bank.

**broken link removed**
http://code.google.com/p/solar-monitor/

Exactly the type I had in mind and with an ammeter shunt (Thanks Keepitsimple) then the voltages of my two banks should be pretty close

I've tried various simulations involving dummy loads / shunt stabilisers but none satisfactorily catch the 16V blip, because the bus resistance is so low. So on re-thinking this I'm back to the suggestion in post #7 to slug the controller response. A simple RC circuit with a time constant ~ blip duration could be used at the controller voltage-sense input, provided the controller could tolerate 16V just for that short time. The controller would still respond to longer voltage rises above 15.5V.

Alec; I once went ski-ing down a slag heap on half a suitcase in Cymbran (wish I could spell it). I have stopped the tripping for now but only by limiting "Absorb" volts to 13.9V which is too low for the long term. However since I am resigned now to new batteries (to be ordered on Jan 3rd) I am OK about it.

Thanks for your continued support.

Aahh, the 'half a suitcase' days. I recall toboganning on an old tin tray in my long-distant youth.
I would think a caravan location in Florida beats a slag heap in Cwmbran (though nearly all the slag-heaps have been prettified now)! Hope the new batteries resolve the issue. Christmas present to self ? Nadolig Llawen, as they say in these parts!

All I kew (excuse spelling) was tutsine bogsine. It's supposed to mean; All Englishmen are @#$@%rds

Christmas Merry to you too.

Well, reading all this took about 20 minutes....with thinking included.

I am summarising here:

1) PV charge controller is delivering a 2 second 'overcharge' voltage to the battery bank when it switched modes.
2) Occasionally this causes the batt bank to rise to 16V tripping the inverters (15.5V trip point)
3) Possible load dump ( fridge compressor cycles off) can also induce a rise in Batt voltage and trip the inverters.

Apparent soln is to trap the surge as it goes up (comparator) and enable a resistive load (water heater coil) until the voltage drops back to a safe ( 15V) hysteresis level. Simple circuit already discussed. A decent sized bank of 4 or 5 heatsunk MOSFETs can handle 2KW for a few minutes easily. Circuit might cost perhaps $40 bux in all.

As to the potential root cause. The OP mentions a loss of capacity in his battery bank and that he is using auto batteries instead of RV or deep cycle batteries. A loss of capacity = increasing internal resistance = additional artificial voltage drop across batt terminals under charge = tripping of inverters.

3 month old batteries under daily charge are unlikely to sulphate and lose capacity. But a wrong charging regimen by the PV controller can cause the batteries plates to buckle and lose capacity & add int. resistance by shedding the active lead/dioxide to the battery bottom.

Deep cycle lead acid batteries have a different charging regimen and the plates are solid lead...built not to buckle under high charge.

On the tech support comment of battery bounce. They are misinformed or feeding u a line. Voltage bounce occurs after the battery is allowed to rest after a deep discharge. It has to do with the migration of the chemical ions in the battery. The PV controller is delivering too much charge to the battery bank causing the voltage surge.

Nothing may be wrong with the controller, if it is set up to charge deep cycle batts with large capacity capable of absorbing the charge without the surge.

The internal resistance of auto batteries is actually lower than that of deep cycle batts, but their capacity for absorbing plate abusing charge is also lower.
What this means is that the PV controller is applying a voltage based charging regimen to the batteries, BUT since the int. res. of the auto batteries is low when they are new, the CURRENT being delivered to the batteries is probably TOO HIGH resulting in Plate damage and shedding.
Eventually the shedding can drive up the internal res. of the batteries and reduce their charging capacity thus causing both the 16V surges during charge and the loss of capacity as observed.

Thank you, you have assessed the situation totally correctly, the only point I would add is that as time goes on the batteries are continuing to deteriorate quickly. 3 months ago I could get close to the alleged 3Kw to 50% DOD out of them. Now it is closer 0.5Kw to 50%. Another thing I have noticed now is that on a perfectly sunny day or on an overcast day the inverters don't trip at all, BUT with intermittent fast moving clouds they can get very annoying tripping 20 times in a day. Now this didn't happen at all until they were over a month old, charging currents/voltages are set up as they are supposed to be but who knows with car starter batteries!

I will be ordering the correct batteries ( Trojan T105RE) on Tuesday, in the mean time I am trying another "fix"; Just purchased a 5 Cu Ft freezer, calcs show me that this freezer will pull about 200 watts but only for 4 to 5 hours in a 24 hour period. As a chest freezer will hold food frozen for at least a day when disconnected I shall put it on a time clock so that all power is pulled during good daylight hours.

This might help, I needed a freezer anyway.

An update.

I now have a new battery bank (6 x Trojan T-105 RE's) and my problem has all but gone away. Not quite totally gone because once a week I should charge my bank to 100% (normally only charged to 90%) and in order to do this I need to increase charge voltage to 15.5V for several hours at which point both inverters will trip. What I was thinking is this;

I believe a typical silicon diode will drop about 0.7V pretty much irrespective of current flow? My inverters will never see more than 150 amps max input current and more typically only 30 amps. 0.7V is about exactly what I need to reduce by to prevent tripping. If this works I was thinking a stud type diode bolted directly to the copper bus bar for cooling and feeding the inverters. Now with 14.8V instead of 15.5V.

I have no idea of the best type of diode to ensure this voltage drop any suggestions appreciated.

I am considering a 1N4045 available on Ebay for $20. It appears to suit my needs but I am an electronic illiterate! I am not sure that this would also work at inverter quiescent current of about 0.5A or if the inverter would fail to start?

The datasheet on it says the Vf is 0.61V max. Not 0.7V. Would that matter to u?
It will carry 0.5A as well as 50A or 150A, other than heating losses (P=VI)

GREAT thank you. 0.61V will work too. Heating losses are insignificant when compared to the power I am losing by only being able to make the final charge at 14.8V instead of 15.5V, this amounts to maybe 20 amps difference for several hours. It will also mean I can make a 100% charge within daylight (PV) hours. I truly appreciate the help Mosaic and will order that diode now. I shall install it permanently but have a shorting switch in parallel to it so normally I will not incur any losses except when I need to top off the battery bank.

Hi,


The regular silicon diodes drop more than the Schottky diodes. You could also check into that if you like. Sounds like you have it all under control now.

Thanks MrAl, After my 1N4045 arrives I'll post the results here. Below is a bit more detail (I wrote for someone else) of why I need this diode.

Addition of power diode to inverter input. (Feb 14th 2012)

Virtually all Off-Grid Solar systems employ a lead acid battery bank to store energy for use after the sun goes down. In order to maximize charging currents modern charge controllers utilize 3 stage charging with a voltage gradually rising to 14.8V (29.6V for a 24V system) where it then stays unless an equalize charge is selected when voltage is increased to 15.5V.

The result of these stages coupled with the finite day time sun hours is that unless almost no power is used during the day the batteries will rarely become 100% charged. The reason for this is that as SOC (State of Charge) % increases the charging current drops off quickly unless charge voltage is increased too. In my system for example to attain 80% SOC might typically take 4 hours the remaining daylight hours will only produce a further 10% SOC when limited to 14.8V so the batteries are only charged to 90% SOC.

While 90% SOC is OK for a while a battery bank does need to be brought to 100% sometimes (once a week?) or it will suffer a shortened life span. It appears most people either ignore this need or shut down their loads for the day and go out somewhere while their batteries charge to 100%

A 15.5V finishing charge will allow me a 100% SOC within daylight hours BUT all inverters I know (I've had 3) will shut down at this voltage (some even lower) and I will be left without power for the fridge and the freezer. The solution is to somehow have a slightly lower voltage going to the inverter than to the battery bank for example charging at 15.5V and reducing that voltage to the inverter by 0.7V would permit the inverter to run happily at 14.8V all day long while the battery bank charges more quickly and to 100%. The trouble is that inverters take very high and widely fluctuating input currents and any form of resistance across their inputs would not effectively work to reduce voltage.

The solution I believe is to use a 1N4045 power diode in series with the input to the inverter. The specs indicate that it has a forward voltage of 0.61V which means it will reduce voltage by 0.61V pretty much irrespective of current being drawn by the inverter. This, if it works, is almost ideal for many Off Grid systems and I shall install a switch across the diode to short it out for normal charging, only opening the switch when I need to equalize charge the batteries at 15.5V.

Note caution should be used to not over-charge the batteries when applying 15.5V charge. Limit the time to an hour or so first until you have used a hydrometer to check the actual SOC and learn how long you may safely apply this Equalize charge for to get no more than 100% SOC.

PERFECTION!

1N4045 is dropping exactly 0.7V (not 0.61V) which is EXACTLY what I wanted. Presently above 80% SOC but still getting 12A charge where normally now I would be at about 2 or 3A.

Thank you all for your help.

Just curious. Presently I is 19A, ambient is 28C and diode is 47C. Does that give us any idea of I max with the existing heat sink?

Let me add that although the diode is rated at 190C my bus bar which is used as it's heat sink has other wires attached which by NEC should not exceed 105C, I personally would be happier with 90C max.

In my practical example 19 amp flow = 19 C temperature rise so obviously if the relationship were linear then with an ambient of 28C current would need to be limited to 57 amps if I don't want to exceed my preferred diode temperature of 85C.

Some research has confirmed my gut feeling that the relationship is not linear but with a larger T diff. cooling is quicker (Newton's Law of Cooling). Therefore I.max is more than 57 amps, however my calculus is way to rusty to figure anymore than this. I guess I just need to do a practical experiment. I would wager that I.max is at least 100A though.