SLA Battery State of Charge

ry based on voltage. The data seems to be all over the map. There is some v ariation from the specifics of the battery type, flooded SLA, AGM and gel b attery. But even then there are numbers on the web that are all over the ma p.

pty" depend on how much you wish to baby the battery and how much capacity you wish to obtain from it.

charge status. The other is to sound levels of alarms for running low and nearly out of remaining run time.

Very little. The theory is simple enough - the output voltage is proportion al to the logarithm of of the concentration of the charged element over the concentration of the discharged element.

Unfortunately the constant of proportionality varies with temperature, so you have to know the temperature of the relevant bit of the battery to work it out.

Charging or discharging a battery makes it warmer, and the heat takes time to diffuse away from the area between the plates where it gets generated.

In theory you could put a thermistor inside the battery, but nobody does.

Pumping in or taking out a little charge, and measuring how far the battery voltage moves per quantum of charge could be more informative, and doing i t for a bit in alternation would mean that you were always dumping in the s ame amount of heat while it was going on, and have the chance the measure a steady temperature rise at the surface of the battery.

It's still pretty horrid.

There are some smart devices that measure internal cell resistance, and from that and a variation of Peukert's law get an indication of residual charge.

The usual method is to check terminal voltage and make an educated guess.

Most alarms are set by a terminal voltage of less that 11V, or perhaps a little higher. Obviously dependent on discharge rate and Peukert's law!

tery based on voltage. The data seems to be all over the map. There is some variation from the specifics of the battery type, flooded SLA, AGM and gel battery. But even then there are numbers on the web that are all over the map.

empty" depend on how much you wish to baby the battery and how much capacit y you wish to obtain from it.

he charge status. The other is to sound levels of alarms for running low an d nearly out of remaining run time.

Thanks for the suggestion, but I don't follow. Peukert's law relates disch arge rate to usable capacity in the battery. I don't see anything relating to remaining capacity.

Our design will drive intermittent high currents so if the voltage differen tial can say something about the remaining capacity, we might be able to us e that.

Remaining capacity is a function of the rate of discharge so it affects the remaining capacity - although this may be irrelevant if the most important thing is to be able to drive a moderately high current.

Monitoring that drop in voltage under maximum load and an empirical determination might be the simplest way of doing it. You probably want a two level warning of

It's long been my impression the the exact voltage of a lead-acid battery is an artifact of construction. For example, lead-calcium batteries seem to display slightly higher voltage than lead-antimony.

It does seem you could extract some hint of SOC from the ratio of loaded to unloaded voltage. Useless in a float situation, unfortunately. In any case, I believe you'll have to know something about the individual battery, or at least the battery lot/type/chemistry.

hth,

bob prohaska

The

erential can say something about the remaining capacity, we might be able t o use that.

y,

Maybe I'm wrong, but I thought I mentioned it is an AGM type, no? Even so, I find different voltage data and the voltage also depends on temperature and I can't get the power board designer to acknowledge putting a temperatu

ient) is useless.

I didn't catch it was AGM. That probably rules out lead-antimony.

How much will be known about the battery deployed? If nothing more than it's a lead-acid of N volts and P amp-hours I think it'll be infeasible to fashion a better than very crude "gas gauge". If the designer will specify a particular product, it might be useful to get a few samples and characterize them. No help at all as things age, unfortunately. Any chance of an integrating ammeter?

Good luck,

bob prohaska

That should have been done from the start. I wonder whether his spec specifies how often battery replacement should be carried out.

e:

ht

eem

ded

any

?C

be

The integrating charge measurement could easily be done as the FPGA will mo nitor the current. But that's like a parachute jump with a meter telling y ou your distance from the airplane. How close you are to the ground depend s on how high the plane was when you jumped. Over temperature and age that distance will change. So the crude voltage measurement will have to suffi ce.

It keeps coming back to the team lead thinking his opinion should be the gu iding light even when he knows little about a topic. I'll be honest and ad mit I don't know a lot about batteries, but I can read and have read everyt hing I can find. That's why it is apparent to me that there is not a level of consistency that will allow a very accurate gas gauge on the remaining time for the battery. I would not worry with it myself... but I'm responsi ble for the alarms and once the power fails the battery has three states de termining the level of alarm. We discussed this a bit today and will need to wing it. The lead pointed out that we don't have to make a decision tod ay and I pointed out that we won't know any more about the matter tomorrow. Putting off many things, such as creating requirements and analyzing err ors in measurements is the sort of thing that needs to be done up front, no t after someone has spent weeks or months working on something.

Virtually every part of this project has been done the way kids might devel op some gadget based on an Arduino. In fact, that was the original design. Then they cobbled together some bits to control a motor and connected a p ressure sensor. Adding bit by bit they got something that looks like it wo rks, but it doesn't come anywhere near working in a real manner.

I do get a laugh from time to time. Every time we need to add some electro nics I make a mental bet with myself that the guy doing the power module wi ll find a Linear Technology part to use... so far I've always won that bet. There is nothing in the LT catalog he doesn't love. $$$ but he doesn't seem to care. Heck, he had one part on the board that was pretty much supe rfluous. It was a input power monitor that cut the power if out of range o r reversed. But everything else was LT and already surviving to nearly the same voltages and reverse.

He still has not produced an off state power budget. The LT parts might do a good job, but they are going to drain the battery faster than we would l ike. We also are going to see current surges on the large motor caps that they are not rated for. I'm going to have to bring that up again in the de sign review.

However, it is important to keep in mind that we are not designing a produc t. What we are creating is more of a reference design. It will be up to a manufacturer to modify the design to suit their requirements. I just know I'm not going to design in anything that has a chance of harming a patient . This battery issue is sounding more and more like one of those things t hat can not be done adequately. So the alarm will just be one state, the h ighest since the potential harm to the patient is severe if the machine shu ts off. If they want the design to measure potentially inaccurate battery levels and lower states of alarm, they will need to have someone else do it .

That's a relief. In that case I will try to stop my brain coming up with parallels to the 737 MAX design process.