Key considerations in powering an ADC

Article By : Don Dingee

Knowing more about powering an ADC can improve data collection outcomes and avoid noise problems.

Earlier in this series, we looked at ways to improve analog-to-digital converter (ADC) accuracy with better grounding techniques and enhanced external voltage references. Now, it’s time to look at the power supply itself.

Power integrity is a complex field, and for very high-performance designs, electromagnetic (EM) simulation becomes an important tool, but one the most makers won’t be able to access due to costs. As usual in this series, we’re not going to the complex theory. Instead, let’s discuss the basics makers should know about powering an ADC.

Analog circuits prefer clean power

After plugging in a USB cable or a power “brick,” one might not give much more thought to powering a small digital system. The lights come on, the prompts come up, and all seems well. It’s the magic of digital circuitry. Until one gets into very high frequencies, digital chips switch crisply between their zero and one states. There is power supply noise, but it must grow very large, perhaps in the hundreds of millivolts, before a digital circuit notices any ill effects.

Analog circuits, on the other hand, are more demanding in terms of power integrity. Noise tends to find a way from the power supply rail into the analog signal. Cutting out that noise is one of the motivations, and it must be done as close to a sensor as possible. If a higher resolution ADC is used, step width—the voltage representing a one-bit digital output change—may be only a few millivolts, making the ADC much more susceptible to noise.

Power supply noise shows up in three ways: ripple, spikes, and thermal noise. Ripple is a time domain effect seen clearly on an oscilloscope, often fed through from the AC power source or a switching artifact. Spikes or transients arise when the power supply controller sees a load change. Thermal noise is common in all electronic circuits, usually at a background level.

Linear vs. switched regulation

Back in the day, analog design purists were sticklers for linear regulated power supplies, usually something like ±12 VDC or ±15 VDC. Linear regulation starts with heavy stuff: large transformers, big capacitors, and the like. Using these, one can make a glassy smooth power supply rail with imperceptible ripple and no spikes. There are also linear regulators designed to step down from one DC voltage to a lower one. Besides being large, linear supplies and regulators are inefficient, losing much of the power input in the conversion process.

Gaining efficiency meant moving to switched mode power supplies (SMPS) or switching regulators. As the name implies, these chop incoming power at a frequency, then smooth the output, which can be higher or lower than the input voltage. For example, a switching regulator could turn the 3.3 VDC found on many maker modules into 12 VDC suitable for analog circuits.

The DC683A reference design is built around Analog Devices’ LT1935 switching regulator. Source: Arrow.com

So, if converting to a higher voltage is easy, why do many modules try to run their integrated ADC on a lower voltage? It’s a cost thing; saving a switching regulator saves space and money. Most integrated ADCs in microcontrollers run at low resolution, like 8-bit. With a high ADC step width and a low voltage swing on sensor inputs, one might run the ADC on a switched digital supply rail without seeing too much noise in the digital output.

Power depends on the application

But maybe not. Suppose the switching regulator behind the digital power supply has a frequency of 30 kHz, and the application samples audio at 96 kHz. Power supply ripple at that 30 kHz switching frequency could be in the samples, making it nearly impossible to filter out. In this example, the best solution would be to set up a separate analog power supply regulator and push it to a higher switching frequency out of the sampling bandwidth, 200 kHz or more.

Increasing the power supply range to analog circuitry has another benefit: improving the signal-to-noise ratio (SNR). An external ADC driver can amplify an input signal using a broader analog power supply. Then, a higher resolution ADC can run more bits at a step width above the noise level, getting cleaner results.

Since the current demand for just an analog driver and an ADC is probably low, a small switching regulator can create the analog power rail for most applications. Ultimately, adding a switching regulator for analog circuitry and improving its noise performance may be more straightforward than cleaning up a digital power rail used for most of the module’s components.

For extremely high-precision applications, this simplified approach may not work. Much of the application literature talks about using a low dropout (LDO) linear regulator for creating analog power rails, presumably after a switching regulator boosts onboard voltages to a high enough level for an LDO to operate.

If one had those more powerful EM simulation tools, it would be possible to hunt down the sources of power supply noise, electromagnetic interference, and other issues. In most cases where makers want to keep things simple, separating the analog power rails on a dedicated regulator is a better choice.

Knowing more about powering an ADC can improve data collection outcomes and avoid noise problems.

 

This article was originally published on Planet Analog.

After spending a decade in missile guidance systems at General Dynamics, Don Dingee became an evangelist for VMEbus and single-board computer technology at Motorola. He writes about sensors, ADCs/DACs, and signal processing for Planet Analog.

 

Expo booth:

New products & solutions, whitepaper downloads, reference designs, videos

Conference sessions:
  • Internet of Things (IoT)
  • Supply Chain
  • Automotive Electronics
  • Wave of Wireless
3 Rounds Lucky Draw:

Register, join the conference, and visit the booths for a chance to win great prizes.

Leave a comment