Design Note 1034: Low Power, Precision Op Amp Simplifies Driving of MUXed ADCs

Introduction

The high speed op amps required to buffer a modern 16‑/18-bit analog-to-digital converter (ADC) typically dissipate as much power as the ADC itself, often with a maximum offset spec of about 1mV, well beyond that of the ADC. If multiple multichannel ADCs are required, the power dissipation can quickly rise to unacceptable levels.

The simple buffer presented here is capable of driving the LTC2372-18 8-channel ADC and achieving near data sheet SNR, THD and offset performance with very low power dissipation if the input signals involved are in the range of DC to 1kHz.

Circuit Description

The LTC2372-18 is a low noise, 500ksps, 8-channel 18‑bit successive approximation register (SAR) ADC. Operating from a single 5V supply, the LTC2372‑18 achieves –110dB THD (typical), 100dB (fully differential)/95dB (pseudo-differential) SNR (typical) with an offset of ±11LSB (maximum) while dissipating only 27mW (typical).

The LT6016 is a dual rail-to-rail input op amp with input offset voltage less than 50μV (maximum) that draws only 315μA per amplifier (typical). It is also available as a single and a quad (LT6015/LT6017).

The circuit of Figure 1 shows the LT6016 op amp configured as a non-inverting buffer driving the analog inputs of the LTC2372-18. Typical power dissipation of each op amp is only 3.7mW. For all eight channels this is a power dissipation of only 30mW, approximately the same power dissipation as the ADC. Running the LT6016 on a single 5.25V supply and enabling the ADC’s digital gain compression mode reduces the total op amp power consumption by more than half, to 13mW, at the expense of a slight decrease in the SNR.

Figure 1. LT6016 Buffer Driving the LTC2372-18 8-Channel SAR ADC

The RC filter at the buffer output minimizes the noise contribution of the LT6016 and reduces the effect of the sampling transient caused by the MUX and the input sampling capacitor.

Circuit Performance

Figure 2 shows a 32768-point FFT of the LTC2372‑18 driven fully differentially by the circuit of Figure 1. THD is –114dB and SNR is 98.5dBFS at 400ksps, which compares well with the typical specs of the LTC2372-18.

Figure 2. 32768-Point FFT for the Circuit of Figure 1

Figure 3 shows SNR vs sampling rate with digital gain compression off and on for both pseudo-differential and fully differential modes of the LTC2372-18. With digital gain compression off, the supply voltage for the LT6016 is +8V/–3.6V. With digital gain compression on, the LT6016 runs off a single 5V supply. SNR stays fairly flat at 94dBFS (pseudo-diff)/98.5dBFS (fully diff) with digital gain compression off, and 92.1dBFS (pseudo-diff)/96.6dBFS (fully diff) with digital gain compression on, up to 500ksps for all modes.

Figure 3. SNR vs Sampling Rate for the Circuit of Figure 1 in Pseudo-Differential and Fully Differential Modes

Figure 4 shows THD vs sampling rate with digital gain compression off and on for both pseudo-differential and fully differential modes of the LTC2372-18. Here THD starts to rise above –110dB at 300ksps for pseudo-differential mode and rises above –115dB at 400ksps for fully differential mode. Digital gain compression has only a minimal effect on the THD performance. In fully differential mode, THD is never worse than –100dB up to the full 500ksps sampling rate of the LTC2372-18.

Figure 4. Pseudo-Differential, Fully Differential THD vs Sampling Rate for the Circuit of Figure 1 with and without Gain Compression

Figure 5 shows the combined offset error of the buffer and ADC vs sampling rate in pseudo-differential mode with digital gain compression off. Offset is initially less than 3LSB and does not degrade until the sampling rate reaches 400ksps.

Figure 5. Offset Error vs Sampling Rate for the Circuit of Figure 1 in Pseudo-Differential Mode

Figure 6 shows distortion vs input frequency for a 400ksps sampling rate. Above 1kHz, distortion rises for all modes.

Figure 6. Distortion vs Input Frequency for the Circuit of Figure 1

Conclusion

A simple driver for the LTC2372-18 18-bit, 500ksps, 8-channel SAR ADC—consisting of the LT6016 low power precision dual op amp configured as non-inverting buffer is demonstrated. The driver dissipates only 3.7mW per op amp (typical), and can be reduced to 1.6mW by running off a single 5V supply with the ADC in digital gain compression mode.

At sampling rates less than 300ksps, SNR is measured at 94dB (pseudo-diff)/98.5dB (fully diff) with gain compression off and 92.1dBFS (pseudo-diff)/96.6dBFS (fully diff) with digital gain compression on; THD is measured at –110dB (pseudo-diff)/–115dB (fully diff) with digital gain compression off or on. Offset measures less than 3LSB (pseudo-diff) with gain compression off. Above 300ksps, performance gradually declines up to the full 500ksps sampling rate of the LTC2372-18.

Author

Generic_Author_image

Guy Hoover

Guy Hoover is an engineer with over 30 years of experience at Linear Technology as a technician, an IC design engineer and an applications engineer.

He began his career at LTC as a technician, learning from Bob Dobkin, Bob Widlar, Carl Nelson and Tom Redfern working on a variety of products including op amps, comparators, switching regulators and ADCs. He also spent considerable time during this period writing test programs for the characterization of these parts.

The next part of his career at LTC was spent learning PSpice and designing SAR ADCs. Products designed by Guy include the LTC1197 family of 10-bit ADCs and the LTC1864 family of 12-bit and 16-bit ADCs.

Guy is currently an applications engineer in the Mixed Signal group specializing in SAR ADC applications support. This includes designing, writing Verilog code and test procedures for SAR ADC demo boards, helping customers optimize their products that contain LTC SAR ADCs, and writing hopefully useful applications articles that pass on to customers what he has learned about using these parts.

Guy graduated from DeVry Institute of Technology (Now DeVry University) with a BS in electronics engineering technology.