LiFePO4 capacity test

[Delfin]

18 months into usage of a 600 ah 24vdc Lithionics bank consisting of two 300 ah batteries. Probably should have done this when new, but I just finished a capacity test, the results of which may be of interest.

1. Charged bank to 29.4 volts to trigger cell balancing, which I had never done before. Generally I stop charging around 28.2 volts.
2. Placed a 23 amp load on the bank via ship's lights
3. Next day I got tired of waiting, so increased the load to 58 amps for a couple of hours via a portable heater.
4. Verified that the draw being measured was the same on the Link 20 and Balmar SG200.
5. At 90% Depth of discharge, still measured 24.9 volts
6. Stopped discharging at 96% DoD and 24.1vdc. Pretty sure I could have taken out 600 amps if I slowed the draw a bit.

Not so sure about the accuracy of the SG200, as it indicated 100% discharge while the batteries were still at 24.5 volts. State of Health is at 89%. Both measures are clearly wrong, unless the Link 20 coulomb counter is wrong, which is unlikely as the number of hours the test took times the draw is very close to the measured amps consumed shown by the Link, plus both Link and SG200 agreed an the current draw.

These batteries really are amazing, and perhaps the SG200 will get in since with reality after a few more charge cycles. It only has about 4 to work with right now right now.

[john61ct]

rough test but all looks good

A more precise test would ignore the SoC meters, just use precise timing with a load kept at constant amps draw.

If you can get (borrow) a charger / discharger that keeps the load current constant that would be most convenient / precise / repeatable, but it is not required for a once-off.

No need with LFP to do it over 20-hours as with lead, but at least 3-4 IMO.

And yes, the ideal is a standardized fully repeatable test on each cell, run when first commissioned, and then after the breaking in period, say 50-70 cycles - capacity should increase a fair bit.

Then maybe every 2-3 years, or if easy when you check for imbalances.

If the bank is treated well you should not see much decline at all for many years.

[Delfin]

Update on the SG200. In conversation with the extremely helpful folks at Balmar, I think because of the lack of cycles, plus the size of the chargers used, the SoC reading shouldn't be relied upon quite yet until the unit has a chance to learn. I should be able to get those additional charge cycles under its belt soon, and will provide the data to Balmar in case it is helpful to them to perfect their algorithms to better reflect my use case.

[Paul L]

Originally Posted by Delfin

Update on the SG200. In conversation with the extremely helpful folks at Balmar, I think because of the lack of cycles, plus the size of the chargers used, the SoC reading shouldn't be relied upon quite yet until the unit has a chance to learn. I should be able to get those additional charge cycles under its belt soon, and will provide the data to Balmar in case it is helpful to them to perfect their algorithms to better reflect my use case.

Maybe Balmar should add one more display to their unit. QOD quality of data. It would then tell you if the other readings are useful or useless.

[john61ct]

Plus the fact that these cycles were far different from your normal cycling patterns.

But really even the most accurate BM is not the way to measure Ah drawn in a controlled load test.

Maine Sail's "adjusting BM" post has a detailed stepwise protocol, just as I said no need to use the .05C rate specifically

[Delfin]

Originally Posted by Paul L

Maybe Balmar should add one more display to their unit. QOD quality of data. It would then tell you if the other readings are useful or useless.

This instrument is very novel, and while I am sure Balmar has done an appropriate amount of testing on typical installations, banks, charging sources and chemistry, you can't test every scenario. I am quite confident in its SoC values in the normal course of usage, but in my case, I was pushing the edges of the envelop in terms of size of bank, depth of discharge and rate of charge. As a learning system, one of the unknowns (confirmed by Balmar) is how many cycles the learning will take for my size bank and usage. I gather this will vary a bit based on the particular installation.

What is very encouraging is Balmar's openness to perfecting the device, and their commitment to it as a part of a suite of products I gather they are building.

The reason I purchased the SG200, right after buying the SmartGuage, was to get the State of Health measure. My hope is that this will help identify whether my particular charging and usage regime is good, bad or indifferent. If good, I'll see a very, very slow reduction in SoH. If bad, I hope to be able to get that indication without going through a capacity test. Time will tell.

[Paul L]

Delfin
I'm not being facetious on saying there should be quality of data indicator. If you are going to call tech support to just be told the unit hasn't cycled enough, well the unit knows how much it has cycled and could turn on a nice yellow led to indicate unreliable data. In your case you believe, if I understand correctly, that drawing a medium load over a long period for testing and then high charging is out of the range of capability for the device. Well the device knows all of this information and could easily turn on the yellow light.
A standard battery monitor is accurate, except when it isn't. It takes intuitive experience to know if the unit is offering up bogus data. And here we have a more complex battery monitor that offers more complex, qualitative data and it takes even more intuitive experience to determine if the data is useful or bogus.

Quality of data indicator should be a design criteria for a device like this that is intended for the broad boat market. It also should have a stated accuracy spec.
Adding new monitors to a techy boat is fun. Easily using the data on Mom and Pops cruising boat I'm not sure about.

[Delfin]

Originally Posted by Paul L

A standard battery monitor is accurate, except when it isn't. It takes intuitive experience to know if the unit is offering up bogus data. And here we have a more complex battery monitor that offers more complex, qualitative data and it takes even more intuitive experience to determine if the data is useful or bogus.

Not so much. The design purpose of the device is to eliminate what you refer to as "intuitive experience". State of charge is easy enough to calculate as an empirical fact, if you know capacity. You can figure out capacity as I did, which is a pain, or you could use a device like the SG200 to calculate it based on impedance, CAR and I am sure other factors. What makes the SG200 unique is that it can calculate SoC with lithium chemistry, and adds a State of Health measure that reflects aging of the battery. Intuition isn't going to be very useful for that purpose.

Originally Posted by Paul L

Quality of data indicator should be a design criteria for a device like this that is intended for the broad boat market. It also should have a stated accuracy spec.

It does have an accuracy spec. It's published in the manual. Your complaint is that you feel it should have an indicator of when the system has learned enough to be accurate. Nice idea, but this is new technology so it doesn't seem unreasonable that just like every other piece of new technology that has come down the pike over the last few millennia, some time may pass before it passes immediate muster with all use cases. Adding additional expense and complexity for a feature that becomes useless after a few cycles isn't something I'd be keen on paying for.

Originally Posted by Paul L

Adding new monitors to a techy boat is fun. Easily using the data on Mom and Pops cruising boat I'm not sure about.

Then it probably isn't for you.

[Paul L]

Originally Posted by Delfin

Not so much. The design purpose of the device is to eliminate what you refer to as "intuitive experience". State of charge is easy enough to calculate as an empirical fact, if you know capacity. You can figure out capacity as I did, which is a pain, or you could use a device like the SG200 to calculate it based on impedance, CAR and I am sure other factors. What makes the SG200 unique is that it can calculate SoC with lithium chemistry, and adds a State of Health measure that reflects aging of the battery. Intuition isn't going to be very useful for that purpose.

It does have an accuracy spec. It's published in the manual. Your complaint is that you feel it should have an indicator of when the system has learned enough to be accurate. Nice idea, but this is new technology so it doesn't seem unreasonable that just like every other piece of new technology that has come down the pike over the last few millennia, some time may pass before it passes immediate muster with all use cases. Adding additional expense and complexity for a feature that becomes useless after a few cycles isn't something I'd be keen on paying for.

Then it probably isn't for you.

The extra led is just an example. It is zero cost to have another soft display that indicates the quality of the data. It might even save tech support cost.

Hopefully the device will work out well for you and you will trust the SOH reading in practice in your use.

[Paul L]

Originally Posted by Delfin

.......

It does have an accuracy spec. It's published in the manual. Your complaint is that you feel it should have an indicator of when the system has learned enough to be accurate. Nice idea, but this is new technology so it doesn't seem unreasonable that just like every other piece of new technology that has come down the pike over the last few millennia, some time may pass before it passes immediate muster with all use cases. Adding additional expense and complexity for a feature that becomes useless after a few cycles isn't something I'd be keen on paying for.

Then it probably isn't for you.

I just went looked at the Balmar website and skimmed the sg200 manual. I don't see any accuracy specifications listed. Can you point me to them?

[Delfin]

Originally Posted by Paul L

I just went looked at the Balmar website and skimmed the sg200 manual. I don't see any accuracy specifications listed. Can you point me to them?

You're correct Paul - I looked at the manual and the specifications just list the usual stuff with respect to amperages, voltages, etc. They state they calibrated the voltage and amperage readings "at the factory", and I have checked those against a Blue Seas digital voltmeter, a Link20 volt meter and amp counter, and a clip on ammeter and voltmeter. All measures agree with one another, so if your concern is that these values are inaccurate, based on my measures, they are correct.

Regarding accuracy of SoH and SoC, I can validate SoC with comparisons to the Link20 amp counter and known capacity of my bank. I can't say the same thing about the SoH, since it is a novel calculation going on behind a curtain, but am comfortable relying on Rod Collins testing where he states:

"The new Balmar SG200 reports SOC and SOH (state of health), amperage, voltage etc.. The fact that the new SG200 can track SOH is a complete game changer and it is accurate on LiFePO4 as well.

I currently use the new Balmar SG200 on my own boat and just recently switched from a Xantrex LinkPro, which was a realy PITA for tracking LiFePO4. The SG200 has proven to be extremely accurate on LiFePO4."

We'll see how these measures play out over time in my install. My interest in the device is with the SoH value. Over time, I can only determine if this is accurate with an annual capacity test, which I will be performing.

[rgleason]

Delfin, thanks for the report. I'll very interested in your next one. I have sg200 on 2 FLA T105 and they've cycled just a couple of times, but it seems pretty accurate.

[Paul L]

Originally Posted by Delfin

You're correct Paul - I looked at the manual and the specifications just list the usual stuff with respect to amperages, voltages, etc. They state they calibrated the voltage and amperage readings "at the factory", and I have checked those against a Blue Seas digital voltmeter, a Link20 volt meter and amp counter, and a clip on ammeter and voltmeter. All measures agree with one another, so if your concern is that these values are inaccurate, based on my measures, they are correct.

Regarding accuracy of SoH and SoC, I can validate SoC with comparisons to the Link20 amp counter and known capacity of my bank. I can't say the same thing about the SoH, since it is a novel calculation going on behind a curtain, but am comfortable relying on Rod Collins testing where he states:

"The new Balmar SG200 reports SOC and SOH (state of health), amperage, voltage etc.. The fact that the new SG200 can track SOH is a complete game changer and it is accurate on LiFePO4 as well.

I currently use the new Balmar SG200 on my own boat and just recently switched from a Xantrex LinkPro, which was a realy PITA for tracking LiFePO4. The SG200 has proven to be extremely accurate on LiFePO4."

We'll see how these measures play out over time in my install. My interest in the device is with the SoH value. Over time, I can only determine if this is accurate with an annual capacity test, which I will be performing.

Pretty lax engineering to not specify accuracy specs for the main criteria that this unit is marketed for. We offer State of Health and State of Charge that is spec'd as just better than anyone else. Only in recreational marine electronics.

[transmitterdan]

SoC and SoH don’t have industry agreed tolerances or even agreed units of measure for that matter. The SoC value can be defined numerous ways. It’s a topic about as inflammatory as anchor choice.

[Paul L]

Originally Posted by transmitterdan

SoC and SoH don’t have industry agreed tolerances or even agreed units of measure for that matter. The SoC value can be defined numerous ways. It’s a topic about as inflammatory as anchor choice.

Certainly no industry standards, but the unit offers these numbers up as a percentage. Given that, it isn't inflammatory at all to ask for a plus or minus accuracy specification. Is it 84% +/- 10% or 84% +/- 25% or is it accurate exception when its not?