Path: ...!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail
From: Don Y <blockedofcourse@foo.invalid>
Newsgroups: sci.electronics.design
Subject: Re: Battery (SLA) modeling
Date: Mon, 16 Dec 2024 16:20:43 -0700
Organization: A noiseless patient Spider
Lines: 113
Message-ID: <vjqcke$1bgbh$1@dont-email.me>
References: <vjgltr$394qj$1@dont-email.me>
 <vjiffm$3kqe4$1@news.eternal-september.org> <vjii6p$3ltn2$1@dont-email.me>
 <vjq1qp$19kqi$1@news.eternal-september.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Injection-Date: Tue, 17 Dec 2024 00:20:47 +0100 (CET)
Injection-Info: dont-email.me; posting-host="12fa7e020a810361f09ab830b15345b4";
	logging-data="1425777"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1+8zgL68qLI0iLQLvPsCu3j"
User-Agent: Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:102.0) Gecko/20100101
 Thunderbird/102.2.2
Cancel-Lock: sha1:a424IHRUIIU3EqrXQ20S9uYU1Vc=
Content-Language: en-US
In-Reply-To: <vjq1qp$19kqi$1@news.eternal-september.org>
Bytes: 5612

On 12/16/2024 1:16 PM, ehsjr wrote:
> On 12/13/2024 7:06 PM, Don Y wrote:
>> On 12/13/2024 4:20 PM, ehsjr wrote:
>>> On 12/13/2024 1:57 AM, Don Y wrote:
>>>> At a fixed time and temperature, is it safe to model a battery as a
>>>> fixed voltage source behind a series resistance?  Possibly, additionally,
>>>> a parallel RC to cover transients?
>>>>
>>>> In particular, does the model need to change based on whether the
>>>> battery is being charged or discharged?  (i.e., to estimate that
>>>> series resistance)
>>>
>>> For what value of "safe"?
>>>
>>> Probably better, in general, to use manufacturer's
>>> discharge curve, at least as a starting point. If
>>> your application differs appreciably from the test
>>> conditions in the manufacturers specs, you need to
>>> test it yourself, again depending on what you have
>>> in mind by "safe".
>>
>> Discharge curve depends on knowing the starting point SoC.
> 
> No. The state of charge indicates the location on the curve.

Isn't that what I said?  Knowing the shape of the curve and points
on it doesn't help me unless I can map my battery's current
state (SoC and age) to that.

> That location tells you how much time remains to reach the
> terminal charge or discharge points, under the conditions
> that were used to produce the discharge curve. The curve is
> a design tool, not something you produce dynamically with
> each charge/discharge cycle in actual use.
> 
>> For a battery in continuous, cyclic use, you need to
>> *determine* the SoC at any given time.  I.e., when to
>> STOP charging, when to stop DIScharging, how much charge
>> you can expect to have available, how long until you
>> are likely going to reach "full" charge, etc.
> 
> Exactly.
> 
>>> It's better to design the system to accommodate the
>>> battery at its best and worst SOC/capacity levels.
>>> Also the age of the battery and how it has been
>>> treated over that age is a factor, so take that
>>> into account too.
>>
>> The hope is that by continuously updating the (albeit
>> simple) model, you can reflect the effects of age IN
>> the model and, potentially, indicate when replacement
>> is required (which largely depends on the service it
>> has seen)

>  From your last paragraph, it seems that you do want to
> model dynamically, using the dynamically updated model
> as (put simply) the yes/no decider for replacement.
> Certainly you could gather data from charge/discharge
> voltage and current during same and idle time and write
> code to do the analysis.

There are two (three?) issues here, each addressable (?)
with an appropriate model.

The first is to be able to ascertain the SoC of the battery
to determine how much charge it needs to accept to be returned
to "full" charge and the best way of delivering that charge
to meet a given timeframe, etc.

The second is to be able to ascertain the SoC of the battery
to determine how much longer it can support the *current*
load and how that load may want to NOW be shaped to meet a
given support time.

The third is to be able to notice how the model has changed,
over time, to reflect aging in the battery.  Determining
when/if replacement is necessary will depend on the SoH
of the battery as well as its load and charging history;
a battery *might* appear to be "shit" yet can still be
suitable for certain smaller loads or shorter support
intervals.

> When would the analysis be "safe"?  I dunno. After enough

I'm not concerned with the "safety" of the analysis but,
rather, the safety of the *model* -- the two are distinctly
different.  I.e., one can have a perfect model and use it
poorly.

> total cycles of replacing batteries you could arrive at a
> better figure than something like "replace after 500
> charge/discharge cycles whether it needs it or not", or
> "replace every 5 years whether it needs it or not", etc.

It will never be that cut and dry.  Rather, it will be
"YOUR usage patterns suggest the battery should be replaced
soon.  Failure to do so can result in loss".

Or,

"Based on your usage patterns, you might consider changing the
battery to something larger (or smaller)"

[No reason to force people to buy more than THEY need]

> But that's using much more than fixed time and temperature
> and voltage data to arrive at whatever you have in mind by
> "safe".
> 
> Ed