1

I build battery-powered circuits with MCUs that consume a few uA in sleep "stage." For total power consumption measurement (and therefore battery-lifetime), I used to use this but recently they changed its capacity range from 0.001Ah to 0.01Ah. I couldn't find a similar device with a quick search.

Is there an off-the-shelf device that could measure Ah with near (or ideally under) mAh accuracy? Does anyone have experience of such measurement with a similar device?

I am interested in an average value (for example, 0.1234 Ah needed for 1 day) type of measurement, not the instantaneous current value. Measuring each circuit "stage" with a multimeter isn't very practical as the circuits can have spikes in current consumption that the multimeter won't catch.

Maple
  • 11,755
  • 2
  • 20
  • 56
p_a321
  • 91
  • 8
  • I'm thinking about how I would personally do this with all the equipment I have access to in my professional capacity as a test engineer, and it's actually seeming quite challenging, even with fancy gear. I think you'd have to set something up with an oscilloscope and a very large storage medium to log current and voltage measurement data over a long period of time, but you'd need a *very* large storage medium unless your scope can do conditional logging, or just log averages instead of entire waveforms... – Hearth Nov 19 '21 at 14:48
  • 2
    This is good question and I think many people would be interested in an answer. Unfortunately, as soon as someone mentions "off-the-shelf" it triggers "no shopping questions" policy here. – Maple Nov 19 '21 at 14:48
  • 1
    But I imagine there's an easier way to do it involving analog integration of the current. Some battery management systems may be worth looking into; I'm not too familiar with BMSes. – Hearth Nov 19 '21 at 14:53
  • @Maple This question is walking on the razor'a edge – p_a321 Nov 19 '21 at 15:08
  • This video https://www.youtube.com/watch?v=LUB8RWzzLWc from Andreas Spies has some comparisons of a few products you might want to look into – Mat Nov 19 '21 at 15:10
  • @Hearth I imagined a device with a small mcu that reads an ADC in regular intervals, and show/transmit the result. Seems doable, that's why I asked for a ready device. The difficult part seem the mA/uA accuracy – p_a321 Nov 19 '21 at 15:11
  • 1
    "the circuits can have spikes in current consumption that the multimeter won't catch." is applicable to regular polling by the MCU as well, so high accuracy won't do you any good – Maple Nov 19 '21 at 15:14
  • 2
    @p_a321 Those intervals need to be very small if you want to catch short pulses. This is why I suggested analog integration. An oscilloscope with several GS/s might be able to catch them, but a microcontroller measuring every ms or so won't. – Hearth Nov 19 '21 at 15:20
  • 1
    Could use a capacitor array with a known dissipation factor and ESR, charge it from the battery, power the device from the array only, then measure the capacitor voltage every so-often. When it gets low, charge it back up. The measurements should give you a consumption. – rdtsc Nov 19 '21 at 15:32
  • Thanks @Mat, I was looking for a similar kind of list/comparison – p_a321 Nov 19 '21 at 16:18
  • @Hearth - The power supply decoupling capacitors reduce the need for measuring at a high rate. With enough capacitance even millisecond sampling is adequate. When using a scope to measure the current the integration function available on many scopes can be used to calculate the charge. – Kevin White Nov 19 '21 at 17:18

1 Answers1

1

(Analysis , not solutions)

To capture the average current spikes well beyond the Nyquist sampling rate poses a significant problem when you have a resolution requirement of say 60 dB from uA to mA.

The latency will be degraded using a LPF 3 decades below the Nyquist rate but will certainly capture the average spike current.

The best way is to use any Coulomb counting method to get high resolution, accuracy and low latency.

  • this generally involved using the following conversions:

    • current>voltage>frequency> accumulator counter to achieve mAh between 1h intervals or any interval scaled to this.

  • An analog equivalent is to have the following conversions:

    • Cap Current discharge Ic =CdV/dt > dV comparator> flyback Ic*dt injection to recharge Cap then choosing low low reactive current switching and count cap charge cycles in mA seconds.

This 2nd choice is basically I see now what @rdtsc has suggested in comments.

The 3rd method is to use a dual-slope integration method and or dual integrate and dump ADC method, which is what I would hope this card is using instead of an SAR ADC for better SNR results. Then the result will capture transients.

There are improvements with Kalman Filters and other methods but , may not be practical for this cheap card.

  • beyond this, it becomes a shopping question with better buying specs and limit$.
Tony Stewart EE75
  • 1
  • 3
  • 54
  • 182