Battery "Bounce"

I am sure because the PV controller records max and min voltages in a daily log. Immediately after an inverter trip I can check the log where I find a 16V max. If I check before the trip I see a more normal 14.8V max. Strangley enough (after this mornings other successful fix for battery/shore power switching) it didn't trip once, Just recorded a Vmax of 15.4V Close but not enough to trip the inverters.

I decided to take a cheap stab in the dark. For $4 incl. expedited shipping I have purchased a lot of eight 1,000 mfd 25 VDC caps. So I'll see where this leads with one, three or all eight in parallel.

Only when it's able to record them.... transients can occur well beyond the voltages you've listed so far if they're in short duration and show up as these voltage trips. Switching voltages especially during a load shed can be WAY higher than the voltages you've listed and still only show up as them, because that's what the circuitry is designed to monitor... Yet again, you still have no clue what the characteristically of this pulse are!

I'll let you know how the caps work out.
Thanks.

Be sure to limit turn on current throught the capacitors with a 10-20 ohm series resistor or a couple watts. The transient current can be quite high. There should really be no reason to add big farads of capacitance. If the pulse is really that long, use a zener or other type of active surge suppresser. Don't worry about power draw. Use a zener with a reverse turn on voltage that's higher than normal operating battery voltage.

OK thanks Brownout.

Hi,

I've dealt with high power DC and synth sine power converters up to 30KW (10KW per phase) so i might be able to help here.

The way surges and things work is if you want to damp out a spike that is tripping an overvoltage set point you can either damp out the system at the power entry point or you can damp it out at the sense point. Since you probably dont have schematics for the converters i guess you are going with the former.

I was going to suggest the same that Scead did, that you try some capacitors across the line. What you should know however is that capacitors dont work by themselves, they need some impedance before them. That is, some impedance in series with them before the actual load. Since we dont know what impedance you are dealing with and again the way exponentials with impedances and capacitances work and since you've already bought some caps, the best bet is to try one capacitance value and then if that doesnt work double it, then test again. Since you've bought 1000uf units you could try 1000uf first, then 2000uf, then 4000uf, then 8000uf. If that doesnt work, 16000uf , 32000uf, etc. This is better than trying to add them one at a time because we need more capacitance than passes the with the first try but not too much.
Also, because of the impedance in series requirement, the best placement of the cap(s) is right across the input to the converter, away from the power source. That puts as much impedance between the cap and the source as possible. If you have several converters mounted some distance apart, the best placement is to use multiple caps, some across each converter input.

There is another problem that can come up however, and that is what kind of surge current the input to each converter can take. When they are tested they are tested with a normal configuration most likely, where the input has no additional caps across the input to the converter. Adding caps means higher input surge when turning on because the caps make the source look 'stiffer' for a short time period. Because of this, it would be best to add the cap as above and not try to swamp out the problem with a huge capacitor capable of blowing out the input to the converter. A converter with a nice slow turn on profile might be able to handle the extra available turn on energy, but without testing we cant know for sure. So the trick is to use as much capacitance as needed but not too much.

Also, a circuit that detects the voltage and switches on a light load temporarily is not hard to build either. Basically it's a voltage comparator and transistor driver and MOSFET combined with a power resistor. This of course assumes that the 'surge' does not last long in normal operation.

Thanks MrAl.
I have a very heavy copper buss(probably 2000A) with the 2 inverters extremely close to it (less than 6" of wire) the battery bank on the other hand is more than 6ft of wire length away. In this case I think it would be fine to mount the caps on the buss bar. Any idea what value series resistance I should try? or is it impedance?

If I do need to go the "extra load" route then I am certain (by tests) that the voltage will drop by more than a volt (probably closer 2V) in a small fraction of a second with the already in place 1,800w (150 amp) load I have so I don't think a voltage comparitor is needed as the voltage will immediately dip enough to switch back off a zener. Perhaps all I need is a zener that will handle the existing 150A load 12V 16 ohm relay coil and switch at slightly above 15V ???

Also I am kind of half guessing that I can't parallel zeners to make sufficient power handling as they won't switch at the same voltage? So I would need a 2amp plus 15V zener? All I can find near this is a TVS zener but I am not sure from the specs that they will operate more than once.

Anyway I am getting ahead of myself; Caps first!

Hi again,

Well one of the ideas is to try not to have to actually add any series resistance, but to work the caps against the existing impedance (or resistance) that is already there. It sounds like you've got very low impedance so it's hard to say how good the caps will work. It depends partly on what is actually generating the surge. Is in the battery itself? If so, less capacitance may work because batteries are not that low impedance when it comes to the first volt or so of change ie they load down more easily if the voltage is due to the recovery voltage. It's still hard to say what is going to work and what isnt when it comes to adding the caps however, so you will really have to test whatever you do thoroughly. What the caps wont do is prevent the voltage from ramping up slowly and trip the converter on an over voltage.

A voltage comparator is a chip that can accurately measure when the voltage surges, and quite quickly too typically within a couple microseconds or less. That would be the device that would measure your voltage and determine if the load should be applied or not, and if it should be disconnected. A zener isnt as good in this kind of application because the accuracy varies over temperature and device selection, so in this kind of application in an industrial setting you wont find zeners being used unless the accuracy doesnt have to be especially good (maybe plus or minus 20 percent). A voltage comparator chip combined with a temperature stable voltage reference will be able to detect voltage changes as little as 0.1v in microseconds and will do it each and every time with the same accuracy with very little variance over time (aging) and temperature. The comparator would drive a small transistor and the small transistor would drive a power MOSFET which would turn on a load made up of a load resistor.

Yes paralleling zeners wont be a good idea. If you really wanted to use zeners you could use multiple zeners each with their own resistor, but again you may find that it works sometimes and not other times, and if it doesnt work sometimes it wont be easy to adjust it up or down say 0.1 volt. The comparator can be made to be adjustable to detect almost any voltage desired, so if you find that 15.1 volt works better than 15.0 volt you have no problems.

Just so you know, the converters themselves are usually built with comparator circuits internally to detect over and under voltage. That allows them to detect quickly and accurately without worry that when they heat up a little the over voltage setting changes.

BTW if you were going to use zeners in a higher power application you would probably opt for a low power zener and a bipolar transistor where the combination creates a high power zener. This could work if you didnt need too accurate a voltage setting, but you really dont want this thing turning on unless you really need it to and so you do want at least some way to set it up properly (adjustment) and know positively that it will stay that way for years to come.

If there is a refrigerator/freezer involved that has to store food long term then you may also want to consider a back up system to monitor the whole process. Depends on what you stand to loose in the event of a primary system failure. For example the Space Shuttle had 7 independent systems working together...if 4 systems decided to take action #1 and 3 systems decided to take action #2, the system would take action #1 not #2.

Thank you for your clarity. I have nothing that is critical to continuous power, it's just annoying. Now my computer used to be quite critical but I simply use a $40 UPS off the inverter now and I have a full 10 minutes to reset a trip.

I modified (re-programmed) the controller yesterday so that below 12.1V there is a 15 second delay then my home made ATS switches to shore power. After the batteries reach 12.7V again there is another 15 second delay and the supply is switched back to battery. Now although I see no reason why, since that mod the inverters have not tripped. Vmax yesterday 15.4V, today so far 15.3V. Too early to be sure yet, but it is a little strange as normally the inverters trip several times per day and almost always recorded 16Vmax.

According to the controller manufacturer they are aware of this over voltage problem but insist it is due to the "wrong" lead acid battery type. Well I think they are probably correct but I can't afford to change these batteries yet although if I can't cure the problem I may be forced to.

I can if you choose explain what exactly is "wrong" with my batteries ?

Can you provide a link or some information for the specified batteries?

No link available. The point is that they are 3 x 185AH inexpensive car batteries, now a car starter battery is designed to deliver a very high current for a short time period and then after a short rest (if car doesn't start) to quickly recover enough to supply a second short high current discharge. to start the engine.
True Deep cycle RE batteries (or golf cart batteries) have much thicker and less porous lead plates so will sustain a smaller discharge for a much longer time and to a higher depth of discharge, (typically 80% discharge instead of my 50%) They also cost 3 times more for the same capacity.

It may well be this quick recovery that I am seeing as a spike? However changing them out is not an option at present.

I doubt the recovery is that quick. We're usually talking about several minutes. I don't know why the battery should make a difference, in this regard.

Hi again,


A theoretical power source has no resistance meaning you can load it to several million amps and the voltage wont crack even a millivolt. Real life sources unfortunately have internal resistance. They also have more than one theoretical capacitive element so the behavior is more complex then we usually have to think about in many applications. One of the unusual things is that they can recover after several minutes to some higher voltage level. Also, the internal resistance makes the voltage jump down when loaded and up when unloaded. These things can be a pain in the neck, but for your application what has to be done isnt that complicated. It might sound like it is, but to us guys that have studied these matters in great detail and worked in the field for many years in various capacities this is a piece of cake. Literally, this is not a hard thing to fix, so dont get too worried about what you have to do to fix it. It probably wont even be that expensive because most of the parts these days are rather cheap, even the higher powered MOSFETs for example. The most expensive thing would probably be the power resistors and a little metal enclosure to hold them safely, or possibly even some 12v light bulbs :chuckle: . Light bulbs however have other undesired characteristics unless you use one rated for much higher voltage, so power resistors would he the best bet assuming the caps dont work.

Lets take a quick look at the simplest solution since you say you dont need great accuracy.

A zener connected to a transistor and possibly a second more high power transistor operated in the linear mode. Since this only has to kick in when the voltage is too high, you want the zener to conduct at the upper voltage level. That turns the smaller transistor on part way and that turns the bigger transistor on part way. The result is the output is voltage regulated. This is your typical shunt regulator. Transistor needs a heat sink also, but a well sized load resistor will take some of the power out of the transistor allowing it to run cool while the resistor runs a bit hot. Add an op amp and you've got a control circuit. Disadvantage is not too good for accuracy when the temperature changes.

Second up is the same configuration, only instead of using a zener you actually build a shunt regulator. A simple one with a higher power transistor. This isnt that complicated either, and in our field the complexity on a scale from 0 to 100 is about 10. This means you will have to wire up one device that looks like a transistor (or a chip) and maybe two other transistors, a few resistors, a pot, and a small cap or two. This has the advantage of being easy to adjust and also quite stable.

What have you built in the past that was electronic and involved transistors?
Have you ever used an op amp?

Also, did you say that you know what resistance will load the bus down enough to prevent over voltage, or could you try a few experiments to find out some information about your system?

The load that immediatly reduces battery V by about 1.5 volts is 1800W so about 0.08 ohms. A 750 watt load (nearly 2 ohms) reduces V by about 0.4V which is likely enough. Especially as these values are typically when the panels are producing 500W or more themselves so battery load is less this 500W and I THINK the tripping doesn't occur until this 500W panel output switches off/ramps down momentarily.

If this IS the case then I may need less dummy load.

I have soldered IC's etc before (repairing circuit boards) but a long time ago. My biggest concern would be making a PCB.

To clarify; I think 750W- 500W or 250W will suffice. This of course is (at 12V nominal) about 0.57 ohms.

Before I meant 0.2ohms not 2 ohms by the way.

Hi again,


Oh ok, so you are dealing with some decent amount of power there then, and some significant current too.
Unfortunately, in order to suppress a voltage of 1.5v for 1/2 second at 42 amps (500Watt figure) you'd need about 14 Farads of capacitance. I dont think you'd want to do that, unless you get lucky and this figure is so nonlinear that a smaller capacitor actually does just enough, but that's something to test for.
So you'd be talking about some 0.2 to 0.3 ohms like you said, and at 500 watts that equals a good ATX power supply tester that can handle the power for at least a second or two, or whatever you need. You'd have to build a resistor bank if you wanted to do it any other way, perhaps using water cooled resistors.
Now that depends on how often this has to function, once per day, twice per day, etc. Resistors rated for 100 watts immersed in distilled water would handle the load pretty well, unless it had to work every two minutes
If you could find 10 three ohm resistors each rated for 50 watts you'd probably have yourself a good resistor bank. I would suggest in this case switching each of the ten banks with their own power transistor.

Well if you havent worked with electronic parts for quite some time you might find this project a little bit of a challenge. You might want to try to get someone to build it for you. Unfortunately Florida is a bit far from me so a field trip down there would take too much time. It's too bad i'd really like to see your setup first hand. Maybe you can take some nice hi res photos and post them so we can see exactly what we are dealing with.

Remember that I have an existing 1800w water cooled load all set up with a 12V 0.8A coil relay which activates it.
This is a home built shower water boost heater (plus 8 to plus 24F depending on flow) which works great although on a normal day I don't use it, even when I do use it it is never in the day time when the inverters night trip.

I still think this is the perfect dummy load as it is all in place

If the caps don't work I will try the water heater route. Surely activating a 10 watt coil must be straight forward ? I HOPE!! LOL.

PS for the second day the inverters did not trip; Very Strange