• Mobile App For Android Now Online!

    Download it here. The app is searchable in the Google Play Store under Sailboat Owners.

    Sorry iPhone/iPad users, we are still waiting on Apple. :(

    Click the X in the upper right corner to make this go away

20 hour battery capacity test done

May 17, 2004
3,433
Beneteau Oceanis 37 LE Havre de Grace
Every couple years I try to do a full 20 hour capacity test of my house batteries to see how they’re holding up. The last couple times I’ve used a shunt, Arduinos, and Raspberry Pi to log the discharge data for comparison. I figured maybe the charts and data will be helpful to someone.

1616235315362.png


The batteries (Deka AGM’s) are 7 years old now, and still reading 100 and 103 amp hours, versus their rating of 105. Can’t complain about that. It is fair to say they live a sheltered life. We do a lot of daysails but that usually only pulls them down to about 80-90% SOC. Maybe a couple times per year we stay out and bring them closer to 50%. Whenever we’re not out we’re on shore power with them charging on a 20A Xantrex Trucharge2.

Here’s the capacity test setup -
1616235818775.jpeg


The battery powers a cheap inverter. Into the inverter I plug a lamp dimmer (the kind with a regular plug on each end), which dims a work light with an incandescent bulb. I can use the dimmer to keep the current draw tuned to about 0.05C (5.1 amps). The shunt is in the ground wire from the inverter. One Arduino reads battery voltage and one reads (with an amplifier) the voltage across the shunt. They send the data back to a Pi that logs it. The whole shunt/Arduino/Pi setup is usually on the boat in summers so I can keep an eye on the batteries there. By using the inverter to power the lamp I find I don’t need to tune the dimmer too often even as battery voltage changes. Every few hours i double check and just tweak it 0.1A or so if it drifts. The only problem with the inverter is that is has a fan that cycles, varying the amperage by about 0.2A intermittently, but I try to keep the average at 5A, and the Pi logs the exact draw each second so I can use the average over time to calculate true AH.
 

dLj

.
Mar 23, 2017
1,777
Belliure 41 Snug Harbor, Lake Champlain
So my question for you: Wouldn't it be better to only do this test, say, halfway? It you only ran it to about 50% discharge, haven't you captured enough data that you could predict the rest? If the slope of the curve it tight enough there, wouldn't that give you confidence your batteries are still fine? Why do you need to actually go the the deep discharge in this test? Doesn't this test in itself damage you batteries?

Just curious and I don't know anything about this kind of testing so I'm really just asking for gaining knowledge.

dj
 
Mar 29, 2017
573
Hunter 30t 9805 littlecreek
I do a real life test at dock every spring with my refrigeration and volt gauge 20 hr is about right for a weekend outing
 
May 17, 2004
3,433
Beneteau Oceanis 37 LE Havre de Grace
So my question for you: Wouldn't it be better to only do this test, say, halfway? It you only ran it to about 50% discharge, haven't you captured enough data that you could predict the rest? If the slope of the curve it tight enough there, wouldn't that give you confidence your batteries are still fine? Why do you need to actually go the the deep discharge in this test? Doesn't this test in itself damage you batteries?

Just curious and I don't know anything about this kind of testing so I'm really just asking for gaining knowledge.

dj
Good question. I’m not sure the slope at 50% would really be steep enough to distinguish. Looking at the data from the 4 tests it looks like there’s a spread of about 0.02 V at the 50 aH mark (with the highest 50aH reading being the lowest final total, interestingly). That’s about the same amount of variation as between 48.3 and 50.77 percent SOC on any one battery. So if I tried to extrapolate out from there I guess I’d be within about 5% of total capacity, which isn’t bad.

As these batteries age I’ll keep charting the data every couple years. When their SOH gets significantly worse it should be easier to see whether the 10 hour mark really distinguishes between good and bad, or if the curves don’t really diverge until lower SOC’s.

I don’t think doing the test every two years is really doing any measurable damage. One 100% discharge every 2 years is not like a cruising boat going to 50% 300 times per year. I’m also sure to put the batteries on a charger immediately when I finish the test, so they’re never just sitting at that lower SOC. Maine Sail has recommended the 20 hour test procedure in the past and said it’s not a significant harm to perform infrequently.
 

dLj

.
Mar 23, 2017
1,777
Belliure 41 Snug Harbor, Lake Champlain
Good question. I’m not sure the slope at 50% would really be steep enough to distinguish. Looking at the data from the 4 tests it looks like there’s a spread of about 0.02 V at the 50 aH mark (with the highest 50aH reading being the lowest final total, interestingly). That’s about the same amount of variation as between 48.3 and 50.77 percent SOC on any one battery. So if I tried to extrapolate out from there I guess I’d be within about 5% of total capacity, which isn’t bad.

As these batteries age I’ll keep charting the data every couple years. When their SOH gets significantly worse it should be easier to see whether the 10 hour mark really distinguishes between good and bad, or if the curves don’t really diverge until lower SOC’s.

I don’t think doing the test every two years is really doing any measurable damage. One 100% discharge every 2 years is not like a cruising boat going to 50% 300 times per year. I’m also sure to put the batteries on a charger immediately when I finish the test, so they’re never just sitting at that lower SOC. Maine Sail has recommended the 20 hour test procedure in the past and said it’s not a significant harm to perform infrequently.
I very much support your testing and reporting out the data! :)

I also completely agree that what you're doing isn't doing any "significant" harm. But if I were to be able to get essentially (as you say within about 5%) the same data without that, first, it would make the test time much shorter, and second, I'd be happier with that test... Although perhaps as you go along, you'll find that the 50% mark just doesn't cut it... I really just asked as looking at your graphs, it sure seemed like you had almost all the data you needed by about 50% discharge, but I've no knowledge really... Seeing how this data changes as the batteries age will indeed be very interesting.

Great set-up you have by the way!

dj