Advertisement

Blog

How Much Juice Is in Your Lead-Acid Battery?

There was a time when I worked for a company that sold an IC that was intended to monitor the voltage of a rechargeable lithium battery and report on the amount of usable charge remaining. Such ICs are usually referred to as “gas-gauge” ICs, although presumably gasoline is not involved (let's hope). I sometimes had customers ask if the IC could be used to monitor a lead-acid rechargeable battery. Sadly, it could not — at least not with a practical design.

You could probably make the existing IC work, but you'd need to add a few more ICs and some trimpots, so what's the point?

Lead-acid batteries are still are used extensively — in automotive applications, of course, and in UPSs (uninterruptable power sources). The UPSs range in size from the small units used in the home for a computer to the giant units used on server farms and in telephone company central offices. So you would think someone would address the needs of designers making Pb-PbSO4 -based equipment. As a quick reminder for those of you who slept through high school chemistry class, note that

Pb + H2 SO4 ⇔ PbSO4 + H2

partially describes the reversible reaction that occurs in a lead-acid cell.

As it turns out, someone has figured out how to monitor and report the charge status of these cells. Texas Instruments has released an IC, the bq34z110. TI is targeting the applications that I had in mind: UPSs, power tools, mobile radios, and small EVs (electric vehicles).

TI says the device will work “independently of battery series-cell configurations,” and that batteries “from 4V to 64V can be supported through an external voltage translation circuit that can be controlled automatically to reduce system power consumption.” The control to which TI refers is an I2 C or an HDQ port that lets the IC talk to the host computer.

The part can be used with batteries up to and above 65A-hr capacity — though I'm not sure from the data sheet how much above 65A-hr. And the part can be used in applications where charge or discharge current is up to or above 32A — though I'm not sure from the data sheet how much above 32A. To some extent, it's a function of the low-side current sense resistor's resistance and power rating. TI recommends a resistor in the range of 5mΩ to 20mΩ.

TI says the part can estimate capacity or state of charge (SOC) and compensate for temperature, age (the battery's, not yours), and for self-discharge (in effect, the battery's internal leakage). It does this via a sophisticated algorithm that TI owns the rights to. In part, the IC is measuring the open circuit voltage (OCV) or no-load voltage of the battery and comparing that with the voltage with a known load. See their application report, “Theory and Implementation of Impedance Track Battery Fuel-Gauging Algorithm” for more details.

The IC provides a simplified readout with four LEDs; if you need more precision, there is a technique via a port expander to add five or more LEDs. An Eval board is available from TI.

An IC that monitors that state of charge would not be very useful if it drew too much current in the process of making its measurements. The part draws 140μA maximum during normal operation. If you put it into (light) sleep mode, it draws 64μA (average); in full sleep mode, it draws 19μA (average). The part is available in 14-Pin TSSOP package.

Have you tried to measure the SOC in lead-acid cells? How accurate was the method that you used?

— Brad Albing, Editor-in-Chief, Planet Analog and Integration Nation Circle me on Google+

21 comments on “How Much Juice Is in Your Lead-Acid Battery?

  1. goafrit2
    October 31, 2013

    //Have you tried to measure the SOC in lead-acid cells? How accurate was the method that you used?

    The easy way of doing this is not to measure the SOC in a battery directly, rather, to have a benchmark to compare the SOCs with another system. It is like measuring open loop gain in amps where the setup affects the ability to actually do that. In this case, the current you draw affects the quality of the outcome.

    So my strategy is to benchmark that battery with another system. You can also use LEDs so that as the SOC varies, you can use the monitoring system to know where you are. That prevents you from disturbing the DUT even while geting the SOC via inferencing.

  2. Brad_Albing
    October 31, 2013

    @goafrit – as long as your model is accurate (or mostly accurate) that could work. But there are lots of variables, from one battery to another. So it's not easy to do it that way you're suggesting.

  3. eafpres
    November 1, 2013

    @Brad–in, say, an automotive application, with all the electronics and such, the battery is never in a no-load condition.  Do you know how they get that measurement?

  4. Davidled
    November 1, 2013

    When benchmarking battery, chemical reaction (Ah versus Temperature) according to temperature might be reviewed.  I think that each manufacture configuration would be slightly different.

  5. rdwjr
    November 2, 2013

    Compared to normal operation, a vehicle sitting in a parking lot has such low current consumption that the battery enters a “relaxed state”. Once that state is detected, a very accurate voltage measurement can be used to correlate with the actual state of charge.

  6. rdwjr
    November 2, 2013

    You are correct. The temperature relationship to state of charge and battery resistance is critical in “Gas Gauging” The modeling is done separately for every variation in battery chemistry, not just every manufacturer. Fortunately, for lead acid, there are more similarities than in the LiIon world.

  7. Brad_Albing
    November 4, 2013

    @eafpres – the answer from @rdwjr is good — go with what he said.

  8. samicksha
    November 5, 2013

    I guess, SoC by voltage is the simplest method, but again it can be inaccurate, considering the fact that cell types have dissimilar chemical compositions that deliver varied voltage profiles, Lead acid has diverse plate compositions that must be considered when measuring SoC by voltage.

  9. goafrit2
    November 6, 2013

    >> When benchmarking battery, chemical reaction (Ah versus Temperature) according to temperature might be reviewed.

    Those can also be modelled. There are chemists that have spent their entire lives to provide those models. The key is finding a way to measure battery without draining the juice in the system. I do not like in-vivo measurement as it takes a lot of resources on the device on duty.

  10. goafrit2
    November 6, 2013

    >> Once that state is detected, a very accurate voltage measurement can be used to correlate with the actual state of charge.

    The probem is that if you do not plan the measurement very well, you can take that battery out of its normal relaxation state in the bid to measure it. In other words, how can you measure the battery without waking it up?

  11. fasmicro
    November 6, 2013

    @Samicksha, that is a valid concern. The varying nature of battery cells is the key reason why people find it hard to develop one solution solves all in this sector. Yet, if you have the core SOC right, small modifications can help get the profiles of different cells right.

  12. Clyde
    November 6, 2013

    Brad,

     

    Was it Benchmarq (acquired by TI, as I remember) that you used to work for?

  13. Clyde
    November 6, 2013

    Brad,

     

    Was it Benchmarq (acquired by TI, as I remember) that you used to work for?

  14. rhayashi
    November 6, 2013

    There are too many variables, lead antimoney, lead calcium, pure lead, sealed/vented, special additives, different acid density? low resistance intercell straps? age, and temperature off the top of my head. Some folks tried impedance to guestimate charge too, didn't work. The rule of thumb of 2.0V @ C/20 (or was it 10) rate is about the best time to cutout the battery.

  15. RedDerek
    November 6, 2013

    Look like a good IC for charge-discharge monitoring of Pb-acid batteries in applications such as power conversion (solar charge the bank during the day and convert during the night by usage). Now to see if they can create a gauge for the car battery – it should accurately tell the driver that the battery is going dead and one has xx days left before it starts to become a problem starting the car.

  16. Davidled
    November 6, 2013

    Instead of adding Solar Charge Bank near engine compartment, solar panel would be installed in the top of roof of vehicle. Energy will be stored directly to battery or secondary battery supply juice to Electronic module by using DC-DC or AC-DC converter.

  17. samicksha
    November 7, 2013

    Yes fasmicro, but the error of voltage-based SoC occurs when disturbing the battery with a charge or discharge.

  18. Brad_Albing
    November 17, 2013

    @Clyde – Nope – Intersil.

  19. Brad_Albing
    November 17, 2013

    @rhayashi – Admitedly, not my area of expertise, so I will defer to the guys from TI.

  20. Brad_Albing
    November 17, 2013

    @RedDerek – for some of the POS cars I've driven (especially in the frozen tundra of Buffalo in the days of my youth) that would have been darned useful.

  21. Brad_Albing
    November 17, 2013

    @DaeJ – not a bad idea – I think maybe someone has done that already.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.