So I am creating a battery voltage measure, and it takes a reading both when the battery is in open circuit, and when it is loaded with a 100mA constant current. I have set it up and the reading is being taken right after the load is triggered and I've been getting some rather peculiar readings, at least, peculiar in my eyes.

I'd calculated the loaded voltage to be around 2V less that the open circuit reading (quite an old and used battery) but when I get the readings from the ADC it shows a drop of only around 1.2V.

I have found that if I leave the load on a then measure it with a DVM the output is significantly lower than the ADC reads and closer to what I am actually expecting, so I was wondering if there is some sort of delay in the battery 'adjusting' to the new load, it's all being triggered by a micro-processor so the difference in time between the load being triggered and the reading being taken is next to none as they're one line after each other, I was wondering whether it was the ADC that could have been the problem but after doing some tests I think it is more likely that there is a delay, I just wanted some confirmation of this and also some reasons as to why!
Is is a characteristic of all batteries or just old and withering ones?

Thanks in advance.

Some information on the circuit:
Battery being tested: CR123A
There is a constant current of 0.5mA always over the battery during test, this is what I am classing as 'open circuit'.
The load that is on when I am trying to measure the loaded voltage is a constant current drain, draining 100mA from the battery each time it is triggered.
For arguments sake lets say the triggering timings are as simple as: it is triggered every 25ms -> triggered on -> measurement taken -> 25ms delay -> trigger load off -> repeat.

Edit: This is part of an accelerated life test for a battery so the time before the reading is taken can't really be increased to negate this problem.

  • \$\begingroup\$ Many things happen with a time constant attached. Wouldn't the easiest thing to do be to load for a longer time then take your reading? Perhaps there are resistive, capacitive, and inductive loads involved. \$\endgroup\$ Sep 8, 2014 at 13:52
  • \$\begingroup\$ @ScottSeidman I can't really give it any longer before taking the measurement as this is part of an accelerated life test of the battery so it needs to simulate what it's target circuit does but do it quicker, otherwise the test could take years and years! \$\endgroup\$
    – MrPhooky
    Sep 8, 2014 at 14:00
  • \$\begingroup\$ Not enough data, and your life test is liable to be invalid unless you can model the mechanisms liable to be involved well enough to allow for them. | BUT what you MUST do is measure the results with a time resolution that is relevant to what you are doing to the battery. Either use an ADC to sample the voltage at a rate such that successive readings show you clearly what is happening at the transition or observe the waveform with an oscilloscope. If you cannot do either of these you do not have the facilities necessary to do the sort of thing you are trying to do. .... \$\endgroup\$
    – Russell McMahon
    Sep 8, 2014 at 14:10
  • \$\begingroup\$ .... An oscilloscope is an utterly essential tool for anyone trying to do anything serious with electronics - and you have crossed into "doing something serious". You can probably get an olde scope for GBP5 to to 10 and maybe free - or a not marvellous but useful one like this one on ebayfor 16 GBP - its from HK.OR you can use a PC audio input as a low feq scope - much on web about these. BUT .... \$\endgroup\$
    – Russell McMahon
    Sep 8, 2014 at 14:19
  • \$\begingroup\$ You have not included a schematic or part number for your ADC. There may well be a set-up time (acquisition time) that is required after you switch the load. Could you do the conversion after the 25ms delay? \$\endgroup\$
    – Tut
    Sep 8, 2014 at 14:37

2 Answers 2


This is a characteristic of any old and withering battery. As a battery nears the end of its life, it gets harder and harder to source the active charges. In the 'off' state, the battery will have a set of charges ready to go so you'll have a relatively high voltage. As you apply the load, the battery resistance will actively increase rapidly near the end of life so by the time you measure it with your DVM, the voltage will have dropped much more than in the 25ms that your ADC is working on it.

This is a phenomenon I've seen many times on NiCd's, NiMh, Alkaline's, Li-ion's so it's definitely not unique to lithium batteries.

A battery this close to the edge everyone else would have thrown away by now so manufacturers won't have any data on this.


Yes, there definitely are time constants associated with establishing new equilibrium reactant distributions when the load on a battery changes. The density of the reactants directly affects the internal impedance of the battery, and therefore its terminal voltage. These effects get larger as the battery ages and the overall density of reactants drops.

However, manufacturers don't spend much time characterizing this, so you're going to have to make your own measurements and adjust your test procedure to suit.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.