Re: Battery (SLA) modeling

The current carrying elements in a battery are ions moving through a soluvent, and they do look capacitative, but they are Warburg impedances and the capacitance varies with frequency.

Probably not that much if the battery isn't getting close to full charge or total discharge.

Fully charged nickel cadmium batteries generate molecular oxygen at one electrode which diffuses back to the other electrode and oxidises metal there (which is why you can trickle charge Ni/Cd cells for ever if you keep the trickle charge current low enough). That might mess up the impedances a bit.

Except for a few special cases, electrode-electrolyte interfaces are very non-linear and you may find that the potential of the inferred voltage source is different for charge and discharge conditions. John

Design for the worst possible series resistance and then some. I have been caught out by that a couple of times. Screaming and apparently untraceable audio instability in the middle of doing the P.A. for a public event is highly embarassing. ("Yes, we put in a new battery - oh no, someone must have put the old one back in the box with the others - terribly sorry!")

Also, whatever the manufacturers clain for the battery capacity, design for half that amount. Batteries don't last forever but your customers will assume they do and won't want to change them until they are well below half capacity.

I've seen kit intended to be powered by lead acid accumulators that has three diodes in series with the supply for when the battery is fresh off the charger and nearer to 14v than 12 with a switch to short them out when the battery starts to get low. It avoid the annoyance you had.

I think most regular users of SLA's are aware of their weaknesses.

I love the way they swell up spectacularly inside certain brands of UPS making them almost impossible to remove. Cruel environment for any SLA.

I've wondered about electro-chemical time lags, ions drifting around in liquids, as in what would the Spice model of some battery be? And on a slower time scale, the morphology of a battery changes with time. Nearly discharged batteries sure go Hi-Z.

What we need here is a good physical chemistry guy.

Yes I have noticed SLA can exhibit an initial voltage sag far longer than can be explained just by internal resistance- I assumed some kind of stagnant electrolyte layering effect?

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".

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.

Ed

That's what Warburg impedances are about.

I've got a Ph.D. in physical chemistry, but batteries are electrochemistry - I know a guy who has just completed a Ph.D. with a thesis on a particular lithium ion battery chemistry. It's a specialised subject.

You can usually model them (voltage across real chemical battery systems) adequately with 2 or 3 different exponential decay time constants and amplitudes if you must.

The first is for transient behaviour ions already close to electrodes, the next for reactions already in progress in bulk and the very slowest for diffusion limited chemistry as the thing equilibrates into its new state. The latter can be quite long but usually only a small correction.

No. The state of charge indicates the location on the curve. 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.

Exactly.

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.

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

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

I guess not many universities offer a PhD in non-physical chemistry these days. :(

The choices in my day at Melbourne University in Australia were Organic Chemistry, Inorganic Chemistry and Physical Chemistry. Win Hill started a Ph.D. in Chemical Physics at MIT at much the same time. It's more a theological distinction than anything with any real world significance, but John Larkin never paid much attention to his undergraduate chemistry lectures. I completed a master's degree in Inorganic Chemistry on the way to getting my Ph.D.

The clever thing to have done at the time would have been physical organic chemistry, and the Melbourne professor of Physical Organic Chemistry moved to America while I was getting my Ph.D, and took a couple of his graduate students with him.

I've been thinking about doing a mid-life master's degree, though some of my buddies in comp sci tell me there's not much point to a "terminal master's degree" these days and should just go for a PhD if one's going to bother putting the money down. Biomedical engineering is my interest at this time. I hear there are some great programs at Canadian schools, never been a better time to see the world I think...

I don't know about Biomedical Engineering, but in Biology, the working level is PhD; anything less is a technician.

I'd troll through listings of available positions and see what the require, or at least desire.

Joe Gwinn