How to Save Power in High-Precision Measurements

Introduction

In precision signal conditioning and measurement applications, a delta-sigma ADC has often been preferred over a SAR ADC on account of its high-resolution output and highly integrated internal blocks such as a programmable gain amplifiers (PGAs) or general-purpose input-output (GPIO) voltage references. However, in some circumstances, the delta-sigma ADC's high resolution is not needed or cannot be achieved, and its high-power consumption becomes a drawback.

Let's consider a strain gauge sensor as an example, which can be used in a Wheatstone bridge converting the signal of interest (pressure, load, force, etc.) to a differential bridge output voltage in a range of a few millivolts. In this case, an amplifier with large gain will be required as an analog front-end (AFE). For instance, to connect a strain gauge sensor with 10mV full-scale output to a delta-sigma ADC with 2.5V full-scale voltage, a PGA needs to be configured to have a 250 gain. This high-gain PGA adds significant noise to the AFE, reducing the 24-bit delta-sigma ADC's effective resolution to 12 bits. Besides the reduction of resolution, a fixed, high oversampling rate determines the current consumption of the delta-sigma ADC, which can be quite high and cause significant reduction in battery life. Delta-sigma ADCs also bear substantial costs due to internal circuit complexity.

An alternative to this approach is a SAR ADC combined with a high-gain INA that can consume much less power with similar resolution at a reduced solution cost. Every SAR ADC implements a two-phase track and conversion structure. In the track (data acquisition) phase, the SAR ADC consumes very little power, and most of the power dissipation happens during the conversion phase. This allows the power consumption of the SAR ADC to scale down with the sampling rate. In a strain gauge sensing circuit, the data acquisition rate can be as low as 1000sps. At such a low sampling rate, some SAR ADCs consume nanowatts of power. Together, with a low-power INA as the AFE, the SAR ADC can save more than 50% power as compared to the delta-sigma ADC with integrated PGA. Like the delta-sigma ADC, the performance of the SAR ADC is limited by the INA. With a modern high-performance, low-noise INA, the SAR ADC can measure strain gauge sensing signals with effective 14- to 16-bit resolution, which has similar or even better performance than the 24-bit delta-sigma ADC.

In summary, the SAR ADC+INA structure is a better choice for battery-powered strain gauge sensor AFEs by offering similar accuracy with reduced power consumption and cost. In applications powered by a battery, power consumption is always a crucial concern, a system design engineer should consider using a low-power, low-noise, high-precision INA with a SAR ADC instead of a traditional delta-sigma ADC.

Detailed Description of Circuit

The structure of the SAR ADC+INA is shown in Figure 1. A Wheatstone resistor bridge is used to represent a strain gauge sensing circuit; an RC filter is placed between the INA and SAR ADC. The INA's differential input can be connected to bridge the output directly. The SAR ADC input should be a single-ended or differential input with negative input connected to ground.

As we already discussed, the INA’s noise performance limits effective resolution of the SAR ADC. For example, for an INA with noise density of 100nV/ and 3kHz bandwidth, the total bandwidth noise at the INA input is 5.48µV_RMS. When the INA is configured with a gain of 100 for an input range of 3VADC, the signal-to-noise ratio (SNR) is 65.7dB. If we choose an INA with a noise density of 40nV/ , the SNR is 73.5dB. So, the noise density around 40nV\ is good enough to achieve 11- to 12-bit resolution.

To save more power, pay attention to both the INA and SAR ADC. For an INA, low quiescent current is important. An INA with quiescent current under 100µA is desirable. As we have already discussed, the SAR ADC consumes less power at a slower sampling rate. There are many SAR ADCs that have this feature. We are targeting to use a SAR ADC with a supply current at range of 10µA (including AVDD and OVDD). So, the total current consumption of the SAR ADC+INA will be approximately 110µA. As a comparison, all current delta-sigma ADCs with PGAS on the market consume more than 300µA current.

It may sound easy to find the right INA based on the requirements we have listed for this application, however, when you do the research, you will find there are few options. Power consumption is the bottleneck. With similar noise performance, most of the INA consumes current at the 1mA range. Fortunately, with our newly released MAX41400 low-power, precision instrumentation amplifier, which features a 41nV/ noise density at a PGA gain of 100 and consumes only 65µA current, you can build a circuit like this. Another specialty of the MAX41400 is its programmable gain with guaranteed gain error over temperature. Traditional INAs often use single external resistors with two internal resistors to set the gain. While the user can adjust the gain using different values of external resistors, external components always add extra gain error and gain drift. Using internal resistors provides the best accuracy and gain drift over temperature. The MAX41400 has gain error as low as 0.05% for all gain options and 5ppm/°C temperature drift. It also offers up to 8 different gain settings and can be changed on-the-fly.

Measurement from Bench

As mentioned earlier, a SAR ADC+INA is a great fit for strain gauge-sensing applications with coin-cell batteries such as bike power meters. The strain gauge sensor is attached on the crank arm to measure its torque. The existing solution uses a 24-bit delta-sigma ADC with an internal 128 PGA gain. The total effective resolution is 12 bits and power consumption is 305µA. We replaced the delta-sigma ADC with a SAR+MAX41400 combination as the AFE. Experiments were conducted and completed on the bench to compare performance.

We use a 3.2V lithium-ion battery here as the power supply to provide power to the MAX41400, SAR ADC, voltage reference, and microcontroller. We use a signal generator to provide a differential signal to represent a strain gauge sensor output signal. The microcontroller sends a command to the SAR ADC to initiate conversion, collect data from SAR ADC in sampling rate of 1000sps. Each section's power consumption is listed in Table 1.

AC performance is tested with a 11Hz sine wave at an amplitude of 27mV provided at IN+ of the MAX41400 PGA gain configured to 100. ENOB is calculated at a different sampling rate.

We tested two SAR ADCs with different resolutions and summarized all three solutions in Table 3. Current consumption and effective resolution are compared. The MAX41400 with SAR ADC can achieve more effective resolution and has much less current consumption than a delta-sigma ADC.

Conclusion

A delta-sigma ADC is known for precision measurement; however, a high-gain PGA often limits its performance. An INA, coupled with a SAR ADC, is a better choice in cases where power consumption is crucial. Care needs to be taken when choosing the right INA. With low power, low noise, and programmable gain, the MAX41400 is a great fit for front-end sensing.