AGM Battery Failure - Two Sets in Two Years

[sailjumanji]

Originally Posted by Bycrick

Sailjumanji — You’re doing a lot of work. But your response about your two Klein and GB meters just proves my point. If a meter has a stated accuracy of 1% of 20-volts-full-scale, that means a reading of 14.00 volts can mean a voltage between 13.80 and 14.20. One percent of 20 volts is 0.2 volts. Very few DVMs are accurate to fewer than +/- 2 counts in the low order digit. So, best case, measuring a 14.00 source, your reading could be anywhere between 13.78 and 14.22 volts.

The fact that the two meters agree isn’t a reliable indicator either. If they didn’t, you still wouldn’t know what "right" answer was.

It seems to me that, if you’re killing batteries as quickly as you say, then the answer is should be fairly obvious. A few hundredths of a volt or an amp probably isn’t significant.

In your position, I’d be looking for a good meter that, when measuring at the battery, really tells you what the charging voltage is. Plus or minus a real tenth of a volt is important.

Ok, I read the meter paperwork. The accuracy is % of reading, plus number of least significant digits. This meter is rated +-1% + 3 digits. So accuracy would be - if reading 14 volts - 0.14 plus .03 = 0.17 volts, or 1.2% of reading.

I understand, but you have to start somewhere. When I use the same Klein meter to check the MassCombi stage voltages without temperature correction, they have always been -0.01 to 0.00 volts variance vs spec. This was in two different tests over a one month period. (We did a lot more tests, but those were temperature corrected.)

The solar controller without temp correction, on its longest bulk run, was -0.01 volts vs spec. The absorption stage was +0.09 to +0.11 volts vs spec. The float is -0.09 to -0.12 volts vs spec. This was as measured on three different recharge tests, over a two week period, using the same meter. Always too high on absorption, and too low on float.

I know we could be off +-1.2%, but the values, the direction, are all repeated each time. And the GB meter is always, without exception, reading either 0.01 or 0.02 volts low to the Klein. It's been clamped to the battery on every test, we periodically move the Klein to the same terminals to check vs the GB. We even change batteries frequently.

[sailjumanji]

Absorption tail. Unfortunately, you have to remember that we are running these recharge tests on house batteries that have less than 10% capacity remaining. The first chart is from the AFT AGM battery, and was the recharge after the first capacity test that was done. This was using the MassCombi. We had the GB meter on the battery terminals, and moved the Klein meter between posts on the charger for volts, and clamped to the battery for amps. At the time, we were probably a little more interested in determining whether our DCSM was giving accurate amps reading.

The second chart, we tried to recharge with the solar panels, but it was a bit too early in the day and very overcast, and we were only getting about 18 amps. So we clicked on the MassCombi. It was in bulk for 7 minutes, then dropped to absorption. About 10 am, we switched to the solar controller because I wanted to see how it would change over to float. It was running a little under 6 amps when it went to float, and then stayed at around an amp.

Both of these were temp compensated. At the time, we were more concerned about battery recharge than checking the charger voltages.

[sailjumanji]

This was the start battery recharge. When I did the 20-hr capacity test immediately before this, we ran the air con because Mastervolt test procedure specified 25 deg C (77 deg F). I turned it off for a short while, started the recharge, and then after a while decided to turn it back on. I think there is a distinct shift in the voltage from temp calibration then, as labeled. (Maybe, or it was just something else.)

A couple of strange things in this graph. The absorption amps flatten out and actually begin to rise, which I attributed to the cooler temp = higher charge voltage. But the charge current was still running around 15 amps, when it dropped down to float. And float was an unusually high ~5 amps, whereas its normally below 2 amps at that stage.

Finally, this was immediately after we had drained this battery to 10.8 volts, removing 107 amp-hours. It took 164 amp-hours to get back to float, which is a 65% charge efficiency. And this wasn't a DCSM error, this was from integrating the amps that were checked frequently with the clamp meter.

[sailjumanji]

Here, I will toss this one out there, to show what a dumb a** I was on a mission to get my shunt zero value exactly correct. And maybe give you guys a chuckle. The Mastervolt instructions say to attach both of the leads from the DCSM to the shunt onto the battery side, make sure the zero value is stable and not moving, then select ‘Yes’ to confirm zero. Then you reattach the leads, start adding loads, and calibrating the shunt.

I attached both wires to the battery side as specified, watched the value waiting to capture zero, and it kept moving and didnt stabilize. Then I started writing down values. Contacted Seawind, and Mastervolt said to move the lead wires away from any current-carrying wires. No change. Tried turning solar off and on to see if that was causing interference. Nope. Replaced the lead wires with another shielded twisted-pair set (same as original). No change. Dug the shunt out of the bottom of the breaker box, moved it away to a wall. Nope. Seawind sent another shunt, and another DCSM. Same problems, just values shifted. So here is what the original setup and changed setup zero values plotted up like. Who can guess what is happening? The shunt is a 50 mv/500 amp shunt.

The second graph is a comparison of all of the zero data recorded for the new DCSM and shunt setup, vs a graph of ambient temperature recorded via a station a few hundred yards away. One of the other Seawind owners figured this out. The zero value is changing with ambient temp. The shunt is a good thermistor, and resistor. Trying to calibrate to get amp accuracy is a challenge when temp is not fixed. As it is, when I have checked vs the clamp meter, we are always within 0.5 amps, and usually less.

[sailjumanji]

Here is another test. This is a solar charger run, to see another tail current example. This was run on the start battery, about 25% depleted, using the solar panels. Its temp compensated, but the probe was on one of the other batteries. (The probes are on the house battery.) So it was compensated for ambient temp.

We had a very high initial voltage drop - about 0.6 volts. This was because we were trying to shove high amperage thru the 2 AWG battery cable, and a parallel circuit to get it there. We retested with large cables later, and voltage drop was less than half of that.

Note that bulk lasted a very short time. This was as per the flashing lights on the charger, and the inflection of the amperage, as shown in the second graph.

[Bycrick]

Sailjumani — I love your persistence. I’m well past having the patience to do what you’re doing now.

I’m also not trying to bust your chops about the meters. But for me, nothing was worse than spending lots of time trying to fix something, only to find that the problem was in my measurement setup. It made me gun-shy.

As to the "temperature sensitivity" of the shunt. At first glance, I think that’s a red herring. First case, you connect both wires to the positive side of the shunt. That short-circuits the input of the measurement amplifier. At that point, both of the wire terminals should have exactly the same potential: they’re connected together. Any voltage impressed on the two wires from the shunt terminal end, will be common-mode noise, ie appear on both wires equally. Most DC amplifiers have lots of common-mode rejection built in. This is especially true when you’re trying to measure 10-microvolts = 1amp in a "noisy" environment. So, with the two wires shorted together, with good shielded, twisted-pair wires, it would take a fairly strong field to generate enough common-mode noise.

On the other hand, when you place two different metals together with any kind of electrolyte(even moisture from the air), it’s easy to generate a few random micro volts between the two metals. I had a similar problem once which was fixed by replacing the two terminal ends. You should be able to eliminate the shunt as a problem by tightly connecting the two terminals together without they’re being connected to the shunt. If you still have noise, it’s cable or environment problems.

Good luck.

[sailjumanji]

Originally Posted by Bycrick

Sailjumani — I love your persistence. I’m well past having the patience to do what you’re doing now.

I’m also not trying to bust your chops about the meters. But for me, nothing was worse than spending lots of time trying to fix something, only to find that the problem was in my measurement setup. It made me gun-shy.

As to the "temperature sensitivity" of the shunt. At first glance, I think that’s a red herring. First case, you connect both wires to the positive side of the shunt. That short-circuits the input of the measurement amplifier. At that point, both of the wire terminals should have exactly the same potential: they’re connected together. Any voltage impressed on the two wires from the shunt terminal end, will be common-mode noise, ie appear on both wires equally. Most DC amplifiers have lots of common-mode rejection built in. This is especially true when you’re trying to measure 10-microvolts = 1amp in a "noisy" environment. So, with the two wires shorted together, with good shielded, twisted-pair wires, it would take a fairly strong field to generate enough common-mode noise.

On the other hand, when you place two different metals together with any kind of electrolyte(even moisture from the air), it’s easy to generate a few random micro volts between the two metals. I had a similar problem once which was fixed by replacing the two terminal ends. You should be able to eliminate the shunt as a problem by tightly connecting the two terminals together without they’re being connected to the shunt. If you still have noise, it’s cable or environment problems.

Good luck.

Well, I have tried three cables. And I think at one time I just connected the two ends away without touching the shunt, and it still wandered. Which is why Seawind sent another DCSM, which didn't change things. But worth looking again.

So are you recommending I get a more expensive meter? I mean, more accurate (which means more expensive.)

[sailjumanji]

Here is another solar test, this with a 50% discharged start battery, house cables, and no temp probe connected. Couple of things. Early on in the test, I noticed that a couple of the solar panels looked like they had a film on them So I sprayed them with water, which both cooled them and added the water sheen. So amps went higher for a while. Note that bulk period here was also relatively short.

The other issue, is that it looks like it might have timed out of absorption at exactly four hours. I am not aware if there is a time out feature on the Solar Chargemaster, but I have read others do have. Also on this absorption profile, you can see that amps start coming up at the tail. About the time it started up, it may have gotten to be a clearer sky. But it still looks peculiar.

If this is timing out, then its supporting the not fully charged scenario.

[Bycrick]

Expensive test equipment isn’t just nicer, it’s usually easier to use. With "cheap" test equipment you need to know more about what it doesn’t do well so that you don’t get mislead. .For example my Fluke 177 is 6000 counts, meaning it divides the measurement interval into 6000 steps. That allows more precise steps than a 2000 count meter, so the low order digits mean more. The trade-off is that with 6000 counts, it takes longer to make a measurement unless the microprocessor is faster. Slower samples mean less accuracy on changing voltages. Fast, with lots of counts, is more expensive. Most cheap AC meters only read accurately with a pure-sine-wave input. So a cheap meter will read a typical modified-sine-wave inverter output as about 87 volts. Not necessarily a bad thing if you know it’s happening, but it confuses a lot of people.

When I was a broadcast consulting engineer a long time ago, I bought cheap test equipment because it was all I could afford at the time. It would take me 3-4 times as long than if I had the good stuff. You had to zero stuff more often because the cheap stuff drifted and you had to repeat measurements to make sure everything was right. The good stuff, you made the measurement once.

So you don’t NEED better meters. What you have is perfectly workable. But you’re leaving yourself open to misinterpretation when you start talking about the difference between 13.70 or 13.77, for example. And the difference between 12.4 and 12.6 volts when measuring the resting voltage is about 25% capacity.

Easiest and cheapest thing is to find somebody with a really good meter that’s in-calibration and check yours. I’m gun shy. I send my meters back for calibration about every three years.

With the shunt, another thing you might try, depending on how the connections are actually made, is to short-circuit the shunt sense wires at the measurement device. Since you talk about shielded, twisted pair wire, the terminals should be accessible. That would eliminate any cable problems from the test. I could more easily see the DC amplifier, wherever it is, as being more noisy and/or temperature-sensitive than the shunt.

[Bycrick]

RE: solar controller timing out of absorbtion mode. That’s very common. Some of the Blue Sky controllers used to do this after 2 hours by default. If you bought their $250 remote panel and battery monitor, you could change it up to 8 hours. Since the solar controller has no idea of how long the battery ought to charge, they often take the lawyer’s advice and make the defaults conservative enough to avoid lawsuits.

[leftbrainstuff]

Hard to diagnose with your narrative but root causes that will kill batts quick are:

1) Massive Frequent Discharge
2) Batt quality issue
3) Overcharge
4) Undercharge

Undercharge seems a possibility. You don't describe your charge algorithm. A 3 or 4 stage charge cycle is needed for house batts that uses SOC to ensure deep cycle batts get fully recharged.

A crude single stage charge will generally only surface charge. This is probably what your outboard charger does.

You can measure good voltage at the terminals but the internal conversion of lead and sulphate ions is incomplete. It is the chemistry that determines state of charge and not voltage which is a useful but insufficient proxy for battery capacity.

With sealed batteries we can't directly check the chemistry (specific gravity) so we have to use proxies. Ensuring a battery has received a proper deep cycle charge is a prerequisite for calculating SOC.

Have you tried using a multistage charger from shore power to see how well the batts recover.

Originally Posted by sailorboy1

You spin a nice story, but it is low on tech details of volts, amps, power in/out of the system really. To me it all sounds that your batteries are not charging fully and they died from this. But you don't provide voltage and acceptance amps to know and ot would need to be on good batteries.

[sailalibi]

Just watching this thread and have a question: what does the BATTERY manual say is the absorb and float voltage requirement. The various chargers usually have rather weak voltages to protect for over charging but not under charging. My TPPL AGMS want 14.7 volts for absorb. It seems that yours only want 14.2?

[sailjumanji]

Originally Posted by Bycrick

Expensive test equipment isn’t just nicer, it’s usually easier to use. With "cheap" test equipment you need to know more about what it doesn’t do well so that you don’t get mislead. .For example my Fluke 177 is 6000 counts, meaning it divides the measurement interval into 6000 steps. That allows more precise steps than a 2000 count meter, so the low order digits mean more. The trade-off is that with 6000 counts, it takes longer to make a measurement unless the microprocessor is faster. Slower samples mean less accuracy on changing voltages. Fast, with lots of counts, is more expensive. Most cheap AC meters only read accurately with a pure-sine-wave input. So a cheap meter will read a typical modified-sine-wave inverter output as about 87 volts. Not necessarily a bad thing if you know it’s happening, but it confuses a lot of people.

When I was a broadcast consulting engineer a long time ago, I bought cheap test equipment because it was all I could afford at the time. It would take me 3-4 times as long than if I had the good stuff. You had to zero stuff more often because the cheap stuff drifted and you had to repeat measurements to make sure everything was right. The good stuff, you made the measurement once.

So you don’t NEED better meters. What you have is perfectly workable. But you’re leaving yourself open to misinterpretation when you start talking about the difference between 13.70 or 13.77, for example. And the difference between 12.4 and 12.6 volts when measuring the resting voltage is about 25% capacity.

Easiest and cheapest thing is to find somebody with a really good meter that’s in-calibration and check yours. I’m gun shy. I send my meters back for calibration about every three years.

With the shunt, another thing you might try, depending on how the connections are actually made, is to short-circuit the shunt sense wires at the measurement device. Since you talk about shielded, twisted pair wire, the terminals should be accessible. That would eliminate any cable problems from the test. I could more easily see the DC amplifier, wherever it is, as being more noisy and/or temperature-sensitive than the shunt.

My Klein CL800 is 6000 counts, and a True RMS meter as well, although I didn't know what it meant until your explanation. I see the 177 has 0.5% accuracy and 2 counts. That said, I understand what you are saying. Unfortunately, I think I have the nicest meter in my circle of friends! :-( I do know an appliance repair guy though, and I'll touch base with him.

I did try a six inch piece of wire across the terminals on the back of the DCSM. It still wandered, but i also didnt give it an excessive amount of time. As did the replacement they sent. I thought about accepting that as the zero, but I thought it would be better to get the wire in its location and try to get an average noise zero there. I did reroute the wires away from other current carrying wires, but no improvement.

[sailjumanji]

Originally Posted by Bycrick

RE: solar controller timing out of absorbtion mode. That’s very common. Some of the Blue Sky controllers used to do this after 2 hours by default. If you bought their $250 remote panel and battery monitor, you could change it up to 8 hours. Since the solar controller has no idea of how long the battery ought to charge, they often take the lawyer’s advice and make the defaults conservative enough to avoid lawsuits.

Yep, I just looked it up. Four hours is standard setting, but it is adjustable with a Masterbus. Which looks like will be my next purchase, as it allows settings adjustments plus gives additional information/data.

It is starting to look like insufficient charge. Will see if SailorBoy is still reading this stuff, so he can say I told you so! I'd be ecstatic if it is the case. I will run some more solar tests ASAP. Right now, super overcast weather.

[sailjumanji]

Originally Posted by sailalibi

Just watching this thread and have a question: what does the BATTERY manual say is the absorb and float voltage requirement. The various chargers usually have rather weak voltages to protect for over charging but not under charging. My TPPL AGMS want 14.7 volts for absorb. It seems that yours only want 14.2?

Mastervolt AGM battery manual, MassCombi manual and the Solar Chargemaster instructions all call for 14.25 volts absorption, and 13.8 volts float, and adjustment via temp compensation at 30 mV per deg C (up to 0.30 volts max.)