I would like to use the ADA4528 opamp or similar opamp to monitor current consumption across a sense resistor for an IoT device but with a Caveat. I want to measure the entire range with one setup/setting

Project: To create a current sensing instrument that I can use to plot the entire Transmission window of a battery power IoT device. To this far iv used different tools on the different current setting and then I add it all up afterward. This is time-consuming and often think small energy savings made in code is not visible due to set up at different current levels.

Parameters: Range required is from 1uA to 200mA (I would like to get nA). System to be powered via USB. so max supply for OpAMP will be 5v unless boosted but would like to avoid if possible. Only DC currents to be measured.

The first thought and one of the questions. if A low tolerance thermal stable resistor is used like a 0.1R is it even practically possible to measure nA at a current of 1uA over that resistor or would other factors dominate.

Second thought, if my first thought is unfeasible then how about a compensation system. Most IoT devices run at 3.6V or lower. So using a 5v LDO or switching setup to step down as the supply. Will it be possible to use a higher value sense resistor like 1R and when the current gets close to the 100mA to 200mA mark the feedback pin of the LDO would tell the PSU to compensate for the voltage drop across the resistor to keep the system voltage stable.

If this could work it will allow the output to be fed into a scope or ADC and record the entire event as a single test. Maybe an entire Soc like the ADS8681 could be a nice solution

Thanx in advance

  • \$\begingroup\$ Show some diagram, how would you interface the opamp, load and shunt resistor. \$\endgroup\$ – Marko Buršič May 31 '20 at 11:02

You run out of dynamic range too easily, there is a reason this is a hard problem to solve.

Lets say you use a transimpedance amplifier to measure your currents, this way there is no voltage drop to affect your reading, and you get a direct voltage output (would need other supporting components to be able to sink 200mA, but this is to show the issue)

lets say you scale things so 200mA = 5V, and 0 = 0V, 1uA then ends up as 250uV, which while measurable needs quite a lot of effort as your at 18 effective bits of resolution minimum, to get nanoamps your at 28 bits, this gets even harder again as you start hitting against noise limits, as you probably want to sample these currents at KHz - MHz

TLDR, your in the ballpark of 5.5-7.5 digit DAQ/multimeters, its not easy or cheap. and just because that audio ADC says its 24 bit, does not mean the result is accurate past the first 14,

  • \$\begingroup\$ Hello, Thank you for your input, I understand that we really playing in the lab test equipment world here and would require upwards of $750-7500 pA meters. but I'm curious as to how accurate a device could be made using what is available for the sort of 75-100$ BOM cost. Iv been through a lot of datasheets and some are claiming capable of 23.5 effective bits. The ADS1283 is a 32bit device with effective 24 it seems. The question is are we in the sort of 1-2% Error margin, 2-5% 5-10% in your opinion? \$\endgroup\$ – Rustie0125 May 31 '20 at 9:41
  • 1
    \$\begingroup\$ Hard part is, I have laid out 7ish digit multimeter PCB's (28-35 bits), so while I can say you can pull it off with a sub $100 BOM, you trade it with a very high time cost, your issues will be heat and noise, each of those will reduce what the ADC claims for its effective number, increase the sample rate to KHz + and that may be lower still, if you aim for the target of 18 bits, then its a possible task \$\endgroup\$ – Reroute May 31 '20 at 9:48
  • 1
    \$\begingroup\$ The transimpedance amplifier approach is still one of the better ones, you would break this into 2 op amps minimum, 1 that stays cooler that senses the offset, and 1 that gets warmer by controlling a pass element to some negative voltage rail, you then have some protection circuitry and feed that into your ADC with as clean as stable a power supply you can, you need to lay it out like a thermocouple amplifier, heat gradients should be minimised where it can cause offsets, if there is a temperature gradient across the ADC or op amp pins it may cause a larger voltage than your signal \$\endgroup\$ – Reroute May 31 '20 at 9:51
  • \$\begingroup\$ Thank you for that into it is helpful, Just a clarification on the above. You noted that a system scaled for 0-5v output will give 250uV, it seems to be 25uV for 1uA(25R). Is that what you meant by saying you will not have the massive voltage drop using a TransIMP AMP because it can detect smaller currents? At 200mA I would still need to dynamically compensate for 500mv drop unless its 2.5R then the drop is tolerable but the TransIMP AMP would have to be able to read 2.5uV at 1uA. \$\endgroup\$ – Rustie0125 May 31 '20 at 11:01
  • 1
    \$\begingroup\$ A transimpedance amplifier keeps its input at "ground", but sinking current and having a negative output voltage that you would then measure, it would be a 25 ohm resistance you would be measuring across, and it would have up to 1W of heat under full 200mA current, you can scale things around a bit, but you probably want as much signal to begin with as possible. your ADC is what needs to measure that tiny voltage, to this end you may need to move that "ground" node a little to keep in the common mode range of your ADC (assuming this circuit has an isolated supply) \$\endgroup\$ – Reroute May 31 '20 at 11:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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