Jon's Imaginarium – Unique Low-Cost Battery Test Method for Microcontrollers

[Visitor]

Via EDN, Adem Kaya's article, Check Battery Capacity With a Simple Test Method, looks like it should detect low batteries well before failure is imminent. It's simple enough to implement on any micro with an ADC.

Measuring open-circuit battery voltage is nearly meaningless. This method with an impressive name – electrochemical dynamic response (EDR) – uses a resistor to momentarily apply maximum load while measuring the dip in battery voltage and the recovery time to full voltage to provide a more reliable battery test. With some experience, you should be able to estimate remaining battery life. The curves below show this characteristic.

The schematic is pretty simple, primarily adding a load resistor and mosfet. The "booster" is to provide a stable reference for the ADC. Using something like a PIC18F-series (and a multitude of others) having a built in reference and the ability of measuring Vcc without any external connections, the booster and voltage voltage divider can be eliminated, making this circuit is almost trivial to add to a battery-operated project.

While I don't think "electrochemical dynamic response" is an earth-shaking idea (more like common knowledge), using it to more-accurately estimate battery life is a great idea. Check out the full article linked above.

 

[MrDEB]

ran across this and got the old mind thinking, using this concept for 36volt and 48 volt e-bike batteries?
integrate an LCD to display battery condition.
batteries are Li-po and NImih.
wonder what your thought on subject?

 

[Visitor]

I would say this method is impractical for testing e-bike batteries, because of the high current drain in this application. Say the maximum current draw is 10 amps (but I suspect much more*). To use this method, the battery is loaded to the maximum level for a short period of time, and the response measured. For a 48 volt battery, you would need a load resistor that will dissipate 480 watts.

If you want to pursue this idea, please start your own thread on the topic.

* One battery I looked at has a C(charge rate) of 7 amps. Discharge rates of 5C or higher look common, which would be 35+amps, and 1680+watts.

 

[crutschow]

I find that the open-circuit voltage of an alkaline battery is a fairly good indicator of its state of charge.
For example 1.3V seems to indicate about 50% of the capacity is left and 1.0V indicates it is essentially dead.

And the open circuit voltage is also a good indication of a Lead-Acid battery's state of charge.

But that doesn't work for other battery types such as NiMH and Lithium.

 

[Nigel Goodwin]

But that doesn't work for other battery types such as NiMH and Lithium.

As part of my Li-Ion design routines I monitor both voltage and current during both charging and discharging - I then plot graphs for the two processes and then select suitable points on the graphs (such as every 10% of time) and use those figures to display battery life and charging status. You can't use just voltage, or just current, you need to use both - but you should have them anyway, as part of the charging routines.

 

[unclejed613]

the open circuit voltage is also a good indication of a Lead-Acid battery's state of charge.

yes, but not of it's overall condition... you can have good open circuit voltage, but if it has a sulphated cell and you put a load on it you'll see you have a problem...

 

[ccurtis]

This looks like a great method. The trick is determining what recovery time corresponds to good, half, and drained for a given battery type and size. Without that data, you can't do anything with it, but guess.

 

[Buk]

I would say this method is impractical for testing e-bike batteries, because of the high current drain in this application. Say the maximum current draw is 10 amps (but I suspect much more*).

E-bike batteries are arranged as nSerial-mParallel sets, and have sensing wires attached to each of the parallel sets. This test method could be applied at the parallel set level where the voltage is only 3.2...4.2v depnding upon the chemistry.

But ultimately, it would need to be applied at the individual cell level; as the most common problem is that one cell in a P-set go off and thus prevents that P-set from charging properly. And that would require a mechanism -- either mechanical or electronic -- to isolate each individual cell in a pack; which would be complex and expensive.

But very useful if it was implemented.

See this for more info on how it is used by Li-ion manaufacturers to grade their output.