Understanding Integrated vs. Standalone Analog-to-Digital Converter Tradeoffs
The importance of Analog-to-Digital Converters (ADCs) is growing – due to the need for systems to acquire distributed data and provide localized monitoring and diagnostics. That leaves the designer to evaluate whether they should choose a microcontroller with an integrated ADC to meet these requirements – or if they should use a standalone ADC.
Keep reading to learn more about how to evaluate the use cases of integrated ADCs vs. standalone ADCs.
First, the most used ADC architecture for data acquisition and monitoring is the Successive Approximation Register (SAR) ADC. The SAR converter works by acquiring an analog signal at a specific point in time. The converter then executes a series of comparisons of the acquired analog input voltage against a reference voltage from the most significant bit to the least significant bit. Each bit comparison takes one clock cycle, so for a 12-bit ADC, you need 12 cycles to complete the conversion. Benefits of the SAR ADC are that it can readily be integrated into a microcontroller. It provides both excellent AC and DC performance at low power and meets the analog performance requirements for a large range of applications.
When it comes to choosing an integrated ADC vs. a standalone ADC, the tradeoff comes down to complexity, size and price vs. performance.
The main advantages of utilizing an integrated ADC are complexity, size and price. Complexity will be lower with an integrated ADC since there won’t be a need to develop software to interface to an external ADC – nor will there be a need to account for the placement and routing of analog and digital signals to and from the ADC. The integration of the ADC with the microcontroller means the overall board footprint will also be smaller.
Therefore, if size is a key critical requirement then an integrated ADC offers a significant advantage. Price of a microcontroller with an integrated ADC is typically lower than the combined price of a microcontroller and a standalone ADC. In fact, a web search shows that in some cases the price of a microcontroller with an integrated ADC is the same price as the microcontroller without the integrated ADC! That’s not uncommon. Looking at how the integrated ADC is designed, specified and tested illustrates why you can get an integrated ADC for “free.” Typically, an integrated ADC is not designed for robustness over manufacturing variations or for performance over temperature. In terms of validating the performance of the ADC, most integrated ADCs are either guaranteed by design or characterized at room temperature only. Any manufacturing process variation or temperature variation greatly degrades the performance of the ADC.
In fact, it’s difficult to find an integrated ADC that specifies the effects of temperature drift on ADC performance. Since the design is not robust and there is no production screening of the ADC performance, there is no yield loss due to ADC performance anomalies. Therefore, an integrated ADC can be offered at no additional cost since there is no relevant additional cost to produce it. Essentially, an integrated ADC specified in this way only provides the function of converting an analog input to a digital output with little correlation to the accuracy or precision of the conversion.
What performance can you expect from an integrated ADC?
Make no mistake, it’s not easy to integrate a SAR ADC with a microcontroller. The clocks, switching and noise from the microcontroller circuits in proximity with sensitive analog circuitry greatly degrade performance. The designer has a decision to make: Do I try to make my design robust so noise coupling and temperature effects are minimized? Or, do I live with degraded ADC performance?
Out of necessity, the answer is to degrade the ADC performance. Because of this, it’s difficult to quantify the performance of an integrated ADC since they do not typically specify performance over temperature. However, there are examples where ADC performance is specified over temperature. This provides a glimpse into the trade-offs that were made. In fact, one insight from temperature drift data is that ADC accuracy is not deterministic over temperature.
For example, a search on the web for a newly released microcontroller with an integrated 12-bit ADC shows that it specifies the ADC’s gain error as ±2.5%. That means for a device with a +2.5% gain error the ADC output code will indicate a value 2.5% higher than the analog input voltage, whereas another ADC with -2.5% gain error will indicate an output code 2.5% lower than the analog input voltage. The accuracy and precision of the measurement is not anywhere near what would be expected from a standalone 12-bit ADC.
Let’s look deeper into this device.
In terms of ADC use, most signals measured are DC. The important ADC specifications impacting DC accuracy are integral non-linearity, differential non-linearity, offset and gain. Evaluating the integrated ADC that does specify DC performance over temperature (from -40C to 85C) as mentioned above shows a total DC error of 108.1-bits or 2.6%, which leaves us with only 5.2-bits of resolution. Keep in mind this error can be positive for one device and negative for another, so the difference in measurement between devices can be twice the error, or in this case 5.2%. What happened to our 12-bit ADC?
In this case, the large gain error of the ADC over temperature reduced the DC performance. The device performance improves considerably using an external voltage reference to a total DC error of 32.7-bits or 0.8%, but using an external reference takes away the integrated ADC advantages of less complexity, smaller size and lower price.
So, DC performance of integrated ADCs is not very good – but what about AC performance? Looking at the AC specification for the integrated ADC, the ENOB is 11.4-bits and the SNR is 70dB, which is good performance. However, the definition of ENOB and SNR are measures of ADC performance relative to full scale, therefore the DC error of 2.6% limits the device from being able to take advantage of a full-scale analog input for AC measurement.
There is still a benefit to having high ENOB and SNR even with large DC errors. High SNR and ENOB indicate the precision, or repeatability, of the measurement is high.
Next, consider a standalone ADC. Looking at the DC specifications of the MCP33111-10, a 12-bit, 1MSPS ADC, the total DC error equates to 1.04-bits or 0.025%, but that is only at 25C. Temperature drift for offset and gain can significantly impact the ADC accuracy, but because the device was designed and tested for precision performance over temperature the impact is minimal. Taking drift into account over the entire -40C to 125C temperature range, the total DC error equates to only 1.13-bits or 0.028%. That is 93x better accuracy than using the internal reference on the integrated ADC and 29x times better than using an external reference on the integrated ADC.
In terms of AC specifications, the integrated ADC has an SNR of 73.2dB and an ENOB of 11.9-bits, so the AC performance and precision of the standalone ADC is slightly better than the integrated ADC.
For designers who want low complexity, small size and low cost, and do not need accuracy or precision, the integrated ADC is the best choice. Designers who want precision or repeatability from measurement to measurement with the same device or are only looking for deviations in a measurement from that device will find that an integrated ADC with high SNR or ENOB is a good choice. For consistency or repeatability of measurement from device to device or system to system –or better than 5% accuracy – then a standalone ADC is the best choice.