Using a delta-sigma ADC in high-precision multisensor systems - Embedded.com

Using a delta-sigma ADC in high-precision multisensor systems

Multi-sensor configurations for oil, gas, and petroleum measurements continually acquire sensitive pressure and thermal data. However, capturing the various combinations of physical temperature and pressure entities requires a very concise, high-resolution system.

This is challenging for designers, as the high-resolution sensor circuitry required spreads across wide ranges of temperature and pressure. In most cases, multi-sensor electronics are too large for factory application, and discrete analog signal conditioning are not precise or rugged enough (Figure 1).

Figure 1: A pressure safety valve protects piping systems from over-pressure. (Source: Shutterstock)

The solution to multi-sensor electronics lies with the high-resolution ADC. The combination of thermocouple, resistance temperature detector (RTD), pressure sensors, and analog-to-digital converter (ADC) are attainable with a precise, high-speed, multi-channel delta-sigma (ΔΣ) ADC, where the key specification is RMS noise. This creates a high-precision, robust multi-sensor system for oil, gas, and petroleum electronics.

This article briefly discusses the issues associated with achieving a precise temperature and pressure cell interface using a ΔΣ ADC.

Pressure sensing

The pressure measurement devices are those that require electrical excitation and those where pressure is the only source of power. The mechanical style devices include bellows, diaphragms, bourdons, tubes, and manometers. With these devices, a change in pressure initiates a mechanical reaction, such as a change in the position of mechanical arm or a tube’s liquid level.

Electrically-excited pressure sensors are synergistic with ΔΣ ADCs and microcontrollers. These sensors can be capacitive sensors, linear variable differential transformers (LVDT), or piezoresistive sensors. Typically, the piezoresistive sensor is the device of choice (Figure 2).


Figure 2: A piezoresistive pressure sensor (a) is typically the device of choice. The high side of the piezoresistive bridge model (b) requires a voltage or current excitation. (Source: Maxim Integrated)

In Figure 2a, the fabricated sensor’s top side is a resistive material and the bottom is a diaphragm. The high side of the piezoresistive bridge model (Figure 2b) requires a voltage or current excitation. The magnitude of excitation affects the dynamic range of the output of the sensor, the maximum difference between VOUT + and VOUT – in a 3.3 V system, which generally ranges from tens of millivolts to several hundred millivolts. Electronics, which follow the bridge sensor using amplifiers and an ADC, change the differential output signal to digital representation.

Importance of temperature sensing

There are numerous types of temperature sensors that are appropriate to any application in terms of temperature range, linearity, accuracy, ruggedness, and ease of use. The temperature sensors in this application monitor the pressure sensor’s temperature to ensure that reliable pressure readings occur. To perform this temperature measurement, this application uses a K-type thermocouple and RTD (Figure 3).

Figure 3: A two-lead TYPE-K thermocouple requires a second temperature measurement with the RTD for cold-junction compensation (CJC). (Source: Maxim Integrated)

In Figure 3, the rugged thermocouple temperature sensor can sense high temperatures up to +1260°C, while the RTD measures the temperature at the thermocouple/copper junctions.

High-resolution ADCs

With ADCs, there is a very strong tradeoff between resolution and speed. Of the fastest converters, the pipeline ADC can produce data rates in speeds of tens of giga-samples-per-second (Gsps), while producing respectable resolutions of up to 12 bits.

The middle-of-the-road ADC is the successive-approximation-register (SAR) converter. This converter produces samples at an output slower than the pipeline converter operating at around 10 Ksps to 10 Msps and at an increase in resolution of up to 18 bits. The SAR converter is a good industry workhorse, if the acceptable input voltage least-significant-bit (LSB) sizes are in microvolts (μV). However, if the application needs conversions of LSB sizes in the nanovolts (nV) region, the only feasible alternative is a ΔΣ ADC (Figure 4).

Figure 4: The basic ΔΣ ADC converts the input voltage into a ΔΣ modulator. (Source: Maxim Integrated)

The ΔΣ ADC in Figure 4 converts the input voltage into a ΔΣ modulator. The modulator creates a one-bit, noise-shaped, pulse train that represents the analog input voltage. The converter then accumulates the one-bit pulse train and through over-sampling, performs a variety of digital filtering on the signal. With time, the filter rejects higher-frequency noise and produces multi-bit results as high as 24 bits. The converter sends these results to the output terminal of an external microcontroller.

ΔΣ modulator

The ΔΣ modulator starts the ADC’s noise reduction process. Close examination of this modulator quickly reveals where the ΔΣ label comes from (Figure 5).

Figure 5: The second-order ΔΣ modulator comprises a feedback system containing a front-end Δ function followed by two integrators (Σ function). (Source: Maxim Integrated)

In Figure 5, after the two integrators, the signal converts through a 1-bit ADC with a sample rate equaling sampling frequency (FS) and then feeds back through a 1-bit DAC with the same sample rate to the inputs of the two integrators. In this system, there is an injection of quantization noise (ei) with the 1-bit ADC. Per the formula at the bottom of Figure 5, the noise appears at the output along with noise from previous conversions.

The modulator generates a noise-shaping effect on the accumulation of the signal at the modulator’s output. This noise-shaping effect shapes the 1-bit conversion quantization noise into higher frequencies (Figure 6).

Figure 6: The noise at the output of the modulator creates a noise-shaped response. (Source: Maxim Integrated)

In Figure 6, the Nyquist frequency for the system is the modulator’s sampling frequency, FS. The order of the modulator determines the level of the quantization noise over frequency (Figure 7).

graph shows the noise-shaping capability of 3 modulators

Figure 7: This graph shows the noise-shaping capability of first-order, second-order, and third-order modulators. (Source: Maxim Integrated)

In Figure 7, the quantization noise of the lower-order modulators is higher near DC and lower at high frequency. The ΔΣ ADC collects or over-samples the modulators’ 1-bit output stream and exercises lowpass digital filtering.

Digital/decimation filter

With the ΔΣ ADC core, there are two actions that occur to reduce system noise. The modulator successfully shapes its quantization noise to higher frequencies and the digital/decimation filter attenuates the high-frequency noise.

The ADC’s output data rate, as dictated by the following digital lowpass filter cutoff frequency, is FD. The frequency response of the digital/decimation filter (dashed line in Figure 4) successfully attenuates the higher frequency noise.

Complete ΔΣ ADC picture

A complete working ΔΣ ADC at the core requires a ΔΣ modulator and Sinc and finite impulse response (FIR) digital filters (Figure 8).

Figure 8:  This diagram shows a complete working ΔΣ ADC with pressure sensor and temperature sensor inputs. (Source: Maxim Integrated)

In the core ΔΣ ADC block diagram (Figure 4), there is a digital/decimation filter. The actual ΔΣ ADC in Figure 8 has the common Sinc and FIR digital filters, which complete the converter’s low-noise picture.

The Sinc digital filter performs a lowpass filter function. A first-order filter design settles in one data-word period. The fourth-order Sinc filter or Sinc4 settles in four data-word periods. The frequency domain filter shape appears with dips over frequency (Figure 9).

graphs show the frequency response of a third-order Sinc filter
Figure 9:  These graphs show the frequency response of a third-order Sinc filter (Sinc3 ). (Source: Maxim Integrated)

In Figure 9, the lowest attenuation can be programmed to match convenient frequencies such as multiples of 50 Hz or 60 Hz. The device in Figure 8 implements a Sinc4 digital filter.

The rounded characteristics of Sinc digital filters make them one of the simplest digital filters to implement, so they are very useful in mixed-signal applications. However, there are applications where sharper corners are preferable. The FIR filter offers sharper corners with an added benefit of stability. The ΔΣ ADC in Figure 8 has a 50 Hz/60 Hz filter that provides more than 90-dB rejection at 50 Hz and 60 Hz at a data rate of 16 samples per second.

The complete ΔΣ ADC (Figure 8) has additional auxiliary functions such as an input multiplexer, programmable gain amplifier (PGA), complex digital filter, clock generator, and reference matrix. With a PT100 RDT, a 160 μA current source, and a PGA gain of 128, the MAX11410 ΔΣ ADC features input range of 1.234 V to 2.837 V. With this 24-bit converter in a Sinc4 configuration, the voltage LSB size is 0.039 μVRMS. The temperature accuracy is across the ±100°C range and RTD accuracy is ~4.7 μ°C/bit.

This article has presented issues associated with achieving a precise temperature and pressure cell interface with a ΔΣ ADC for oil, gas, and petroleum electronics. The application circuit uses a pressure, thermocouple and an RTC sensor to achieve a single-device conversion, where the key specifications are noise, an input multiplexer, and bill-of-materials (BOM) cost.

>> This article was originally published on our sister site, EDN.


Bonnie Baker is a principal writer at Maxim Integrated.

 

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.