Balmar SG-200 Battery Monitor

[nebster]

Originally Posted by john61ct

I suppose. But new information coming along could result in a different estimate curve.

That is the nature of a dynamical system. But, the fact that it could happen does not mean it will in practice or with regularity.

Once the sample size is large enough, if the data is sufficiently predictive, even new outlier information does not have to throw off the estimation algorithm.

Given all the hundreds of wildly varying batteries in use globally,

only a very small subset used in development of such a device even assuming unlimited resources available,

with new brands and chemistries being released all the time,

For the most part, brands and manufacturers don't matter. Chemistries are what matter, and the SG200 has a finite lookup table for those. If you are worried about hypothetical future chemistries, I guess you better hope Balmar has a nice firmware update mechanism, because fitting curves to the wrong characteristic curve is going to be tough rowing.

and the fact such algorithms will not improve actual measuring accuracy, just add complexity for the sake of an "I think I'm done learning now" indicator,
seems to me an unrealistic expectation.

You've stated that, and a number of us disagree. There's no technical obstacle to computing and displaying a confidence indication.

The key is to remember, the actual phenomena being measured are not precise in the first place, so there is no point other than marketing, in encouraging the customer to believe in precision indicators beyond the 3-5% range.

To the contrary: I am confident a capacity estimation could easily converge to within 1%. I know that because if you show me a curve from a real LFP battery, I can in short order tell you its capacity pretty closely. And that's just using my eyeball heuristics and with a single sample curve.

Now, whether 1% versus 3% is that important in the grand scheme of things, that we could debate about. I guess the consumer can make that decision. If I'm delfin, who wants to see how his capacity changes over time, I think I'd like 1% granularity, since 3% or 5% might take years to*become observable.

[john61ct]

Originally Posted by Maine Sail

The original Smartgauge tells you SoC very accurately

Not sure why you would use the adjective "original"?

Did Merlin release distinct versions, other than maybe firmware updates?

That syntax, used in the context of a discussion about Balmar's SG200, IMO implies too great a connection between the two very different devices?

[Paul L]

Originally Posted by john61ct

That is the only practical measure IRL, and only applies to **that** battery in **those** conditions.

BS. Then why even have an SG200 if it isn't a practical measure?
You want to play both sides of the game, claiming the device is accurate and a must on a cruising boat, and that only a complex test that very few actual cruisers ever do is required.
I guess when its all just theoretical to you and you have no cruising electrical experience, mimicking what you read online as if its gospel is the best alternative.

[Maine Sail]

Originally Posted by john61ct

Not sure why you would use the adjective "original"?

The pre Balmar version is "original Smartgauge"....

[john61ct]

Originally Posted by Maine Sail

Lets keep this simple, the SG200 can take the guesswork out of it for most owners.

I completely agree that simpler is better, here wrt the SG200 display.

I think having the SoC% display a number other than 100% when the bank is Full as it can get without overcharging does not make things simpler, but more complicated.

Is that indeed how it works at this point?

[john61ct]

And is it the case that there no possibility to manually reset the gauge at the charge-termination point?

Eliminating the **need** to do it is indeed an admirable goal, but IMO it's a very simple and easy calibration, so removing the **ability** to do so seems potentially a step backward for owners who find the SG200 is not as accurate with their specific bank as they'd hoped.

[john61ct]

Originally Posted by Maine Sail

How do "end amps" work when the battery starts to sulfate?

As you know, it starts taking longer to get to that 100% Full point.

The below is for other members, but do please correct any specific errors.

If the owner has successfully been getting to that point anyway most cycles, not letting the bank sit for long below that point, then likely that is a good indicator the bank is reaching EoL.

There may come a time that the endAmps spec needs to be adjusted, if .005C at first maybe to .01C, or progressively higher to .03C if AHT becomes impractical given the type of charge sources available.

Deka's spec of

charge current rate taper starts to change less than 0.1A per hour

is a good fallback after that point, if the owner feels other information indicates the bank is still sound enough to keep using.

[Maine Sail]

Originally Posted by john61ct

Exactly.

The 100% SoC point is IMO best defined as the point at which that battery, at that time is Full, as indicated by holding Absorb voltage until current tapers down to the endAmps condition, ideally both spec'd in the mfg datasheets.

Easily and objectively verified by any owner with an ammeter.

Which I guess you are here calling "chemical 100% SoC"?

I hate to break it to you John but you can have a battery that meets your "end amps" definition yet not be at 100% SOC. You'd know this if you had practical experience with batteries. This is due to sulfation and the time it takes for the energy to slowly absorb into the plates.

[john61ct]

Originally Posted by Maine Sail

Lets not confuse SoH and SoC

I am pretty sure nothing I've stated indicates any such confusion.

I would greatly appreciate your correcting (or asking about) other statements you find wrong or where my intended meaning is not clear. I do try to be precise in my communications, but do concede the wording is sometimes hard to follow. If you're willing to go to the trouble, please quote specifically where I've gone wrong, such general criticism is less helpful.

[john61ct]

Originally Posted by Maine Sail

I hate to break it to you John but you can have a battery that meets your "end amps" definition yet not be at 100% SOC. You'd know this if you had practical experience with batteries. This is due to sulfation and the time it takes for the energy to slowly absorb into the plates.

From a practical point of view, what should the owner of such a bank then do, assuming on shore power and the bank is still worth keeping?

Stop charging or press on past say 12 hours AHT?

Whatever your answer, **that stop-charge point** is IMO the point every BM afaik should be showing 100% if it is working as well as it can.

If the SG200 cannot be set up to work that way (is that the case?)

then can you point to another BM for which that is also true?

[john61ct]

Originally Posted by Paul L

You want to play both sides of the game, claiming the device is accurate and a must on a cruising boat, and that only a complex test that very few actual cruisers ever do is required.

I do not feel that way in either case and am pretty sure I've never stated either position in that way.

There are types of owners and use cases, expressed desired goals where such recommendations are indicated, but of course both are irrelevant to a very high percentage of cruising boat owners.

[john61ct]

Originally Posted by nebster

For the most part, brands and manufacturers don't matter. Chemistries are what matter, and the SG200 has a finite lookup table for those

A 12V $25 flooded big box special has the same chemistry as a 200lb Rolls Surrette 2V cell. They have very different wear patterns.

Originally Posted by nebster

I guess you better hope Balmar has a nice firmware update mechanism, because fitting curves to the wrong characteristic curve is going to be tough rowing.

Well I would sure expect they do, can you name a quality modern BM that doesn't?

The usual complaint I see is not giving the user enough control over when updates get auto-installed over the Internet, possibly burning up MB on a very expensive mobile data plan.

[Delfin]

Originally Posted by Maine Sail

We have tested batteries that can barely must 30% of their Ah rating yet can achieve a chemical 100% SoC based on both open circuit voltage and specific gravity. The owner of one bank, and an original Smartgauge, knew his SoC was plummeting way too fast (time @ X load) and this is what gave him insight into his failed bank. So yes his battery was being charged to 100% SoC but it had an actual Ah capacity of less than 30% of rated or less than 30% SoH but it could still achieve a chemical 100% SoC. The SG200 removes the mental time equation by adding SoH to the mix..

Just because a battery can charge to 100% SoC this does not tell you anything about the batteries SoH in relation to Ah capacity unless you examine other variables.

True dat. Case in point - I have the Smartguage wired to my LA starter bank and the SG200 to the Lithium. The starter bank shows 100% SoC, but they are getting pretty well sulfated, so what started out as 210 Ah @24vdc is probably around 100 Ah or less. Hydrometer readings confirm this even after repeated and sequential charge cycles with 29 volts bulk and 5 hours of absorb. So, I'll soon be replacing the starter bank that will no doubt be at 100% SoC when I recycle them.

Originally Posted by Maine Sail

John why do you enter discussions acting as an expert, chronically using absolute statements, when you have little knowledge or practical experience of what you are actually discussing? What you just said above is about as gobbledygook as I have read in a while, it does not even make sense..



Lets keep this simple, the SG200 can take the guesswork out of it for most owners. Program your rated capacity, go use the bank, pretty soon it will be telling you an SoH that is much more accurate than any device we've seen thus far. It won't do this immediately or with 5% discharge cycles but it is darn simple.

I gather from Balmar that larger Lithium banks at >12vdc may require some tweaking to the launch algorithm. I hope to be able to do some decent measures following multiple charge cycles on a trip up to Alaska that will start in May. I'll start with a measured capacity for the LFP bank, and see how many cycles from an untrained SG200 it takes for the SoH to be reflective of the measured capacity. I should be able to get at least 100+ cycles so what will be interesting (to me anyway) will be how long it takes for the device to learn my bank, how well SoC maps to the coulomb counter as measured by a Link20, and whether end SoH can be confirmed with a final capacity test.

Not being an expert, and also a fan of Godel, I'm unable to make absolute statements about much of anything, but I hope I can contribute something with this testing.

[nebster]

Originally Posted by john61ct

A 12V $25 flooded big box special has the same chemistry as a 200lb Rolls Surrette 2V cell. They have very different wear patterns.

If they're the same chemistry, then within some charge envelope that's sufficiently limited ("reasonable") for each battery, the charge curve should approximate the characteristic shape for that chemistry.

If the curve doesn't roughly match the baseline shape, the computer is not going to be able to fit very well, or it may have to do piecewise training in different behavior regimes, or simply not do any fancy learning/fitting except in one portion where the curve matches.

I think this is why you must tell SG200 which chemistry you are using. I would expect its estimator to work very poorly (and possibly just fall back to straight coulomb-counting) if it cannot achieve a reasonable match.

In which case, in the scenario that you assert is unachievable but that I think is entirely feasible, the SG200 or a similar device could then continuously indicate that it is not able to achieve high confidence in its capacity estimate.

Well I would sure expect they do, can you name a quality modern BM that doesn't?

The comment was made tongue-in-cheek: Balmar has likely designed the computer in a way that it can handle a wide range of real battery curves, but only when they approximate the characteristic shape for a small number of common chemistries.

If your assertion then they would need to constantly be updating the firmware and adding new battery types and styles and brands and sizes. There's no way they're doing that, and the software very likely doesn't work that way.

[nebster]

Originally Posted by john61ct

Which is known to be inaccurate, unless the result of a recent load test.

You seem very confused about what "SOH" means on an SG200.

SOH is just {SG200-computed-capacity} divided by {a capacity someone typed in when they set up the SG200}.

It will always only be as helpful as its denominator, regardless of any individual's accuracy.

The concept is that you punch in your nameplate specified capacity (for the denominator) and the computer estimates the real capacity over time (for the numerator). The computer then displays that ratio.

Why is this so complicated?

> Let B = the capacity the SG200 currently calculates from its fitting algorithm.

Which accuracy is unknown, unless compared to A.

No, that sentence is incorrect.

The calculated capacity (the numerator above) is as accurate as SG200's black-box learning algorithm can be, which is to say, we have no idea without just observing how it behaves in real life. It might be very accurate, or wildly inaccurate, or somewhere in between.

But comparing it to "A" (the denominator) doesn't change that accuracy. It doesn't get more accurate when compared to A. It just becomes a unitless value.

> You wrote, above, that SOH should "adjust" to 100% over time, even if it comes up lower in early tests.

The **only** two SoC% points **known** to be accurate for sure, are

0% (10.5V under a light load, nominally .05C), and

100% (point tapering current hits "endAmps Full")

You seem to be mixing SOH and SOC when you are either reading or responding, or somewhere in your head in the interim. You might want to go back and review some of your latest posts in this thread, because I think you are confused -- and then you confuse us, because we can't figure out what you're writing.

> SOH is displayed as B / A.

Wut?

only if A is a fixed number intended to indicate either vendor rating (which we know rarely accurate), or the "maximum ever" capacity determined by a benchmark load test after the commissioning / breaking in period.


The SoC% number displayed should IMO be a relative "countdown", with the 100% mark matching the definition above.

When the SOC shows 100%, the SG200 has determined that the stored energy equals the calculated capacity.

SOC = {calculated stored energy} / {calculated capacity}
SOH = {calculated capacity} / {stated capacity}

The SG200 is measuring and computing the first two variables.

This is pretty straightforward, so another thing you might try is (a) downloading the manual and reading it, or (b) buying one and hooking it up and trying it. We'd all be the better for it.