The Component Selection
Once again, the MAX14921 AFE is useful here where cost reduction diminishes the accuracy of the rest of the system.
This architecture eliminates the external reference, introducing the need for an ADC with an internal reference. Here the MAX11165 ADC offersoffset error, gain error, and INL values equal to the MAX11163 used in the high-accuracy architecture (Figure 2); its internal reference provides an initial accuracy of ±4mV with a temperature coefficient of 17ppm/°C (maximum).
Again, using the Error Measurements spreadsheet shows that the maximum six-sigma error for this system is 0.178% (7.305mV) and the maximum three-sigma error is just 0.122% (5.014mV). Anecdotal evidence from the performance of the MAX14921 evaluation (EV) kit shows that this system should have much better performance than even these maximum errors indicate. Note: such anecdotal evidence serves as a framework for a possible outcome, but actual system design should rely on maximum error.
There are times when costs trump accuracy in battery-monitoring applications. In this solution, accuracy is a trade-off that designers typically must accept. Now it is advisable to eliminate any possible external components and use a microcontroller with integrated ADC (Figure 5). This architecture significantly reduces costs, and the lower component count allows precision cell-monitoring systems to be implemented in space-constrained applications. This design is a viable cost-optimized solution for lithium battery chemistries with moderate-to-steep discharge curves that do not require high precision measurement.
Figure 5. This cost-optimized architecture again features the precision MAX14921 AFE and a microcontroller with integrated ADC and reference.
There are two caveats with this design: ADC accuracy and reference voltage. First, an ADC integrated in a microcontroller typically does not have the accuracy performance of a discrete ADC.
The issues with