Using Zero-Drift Amplifiers in Wider-Bandwidth Applications

Using Zero-Drift Amplifiers in Wider-Bandwidth Applications

Zero-drift amplifiers are often treated as devices for slow, dc-oriented signal chains. That view is incomplete. Chopping and auto-zero techniques do create high-frequency artifacts, but those artifacts are predictable. If the designer understands where ripple, switching spikes, intermodulation products, and alias terms occur, zero-drift amplifiers can be used successfully in significantly wider-bandwidth precision systems.

Their main benefit remains extremely low offset, drift, and 1/f noise. The additional engineering task is to control the ac side effects introduced by dynamic offset-correction techniques.

1. Chopping and Auto-Zero: Two Ways to Remove Low-Frequency Error

Chopping translates offset and low-frequency noise away from baseband. The input is modulated to the chopping frequency, amplified, and then demodulated back to baseband. The amplifier's own low-frequency error is shifted toward the chopping frequency and its harmonics, where filtering can attenuate it.

Time-domain view of the chopping process

Auto-zeroing uses a sample-and-correct approach. During one phase, the amplifier measures its own offset and low-frequency error; during the signal phase, that stored error is subtracted. Continuous operation normally requires two interleaved channels.

Basic auto-zero amplifier structure

Both techniques improve low-frequency accuracy, but they do not have identical side effects. Auto-zeroing can fold high-frequency noise back into baseband, while chopping more commonly produces discrete ripple and switching-glitch components.

2. The Two Main High-Frequency Artifacts

The first is ripple. Square-wave modulation moves residual offset and 1/f noise to the chopping frequency and odd harmonics such as fCHOP, 3fCHOP, and 5fCHOP. Because no low-pass filter has an ideal brick-wall response, some residual energy remains.

Frequency-domain translation of signal and error in a chopper amplifier

The second is glitch energy. Imperfect charge injection in the chopping switches produces transient spikes. Finite amplifier bandwidth adds another mechanism, causing additional artifacts around even multiples of the chopping frequency.

Glitches caused by charge injection and finite bandwidth

3. How Modern Zero-Drift Amplifiers Reduce Artifacts

Modern devices often combine several techniques: factory offset trimming, chopper-plus-auto-zero architectures, and local auto-correction feedback loops that sense output ripple and cancel the underlying low-frequency error.

Glitch reduction can include charge-injection trimming and multi-channel chopping. Multi-channel approaches generate more frequent but smaller pulses and move the energy to higher frequencies, where system filtering is easier.

Comparison of conventional zero-drift spikes and ADA4522 behavior

4. Find fCHOP Before Designing the Filter

The first practical step is to identify the chopping frequency. It may be listed directly in the data sheet or inferred from the noise spectral-density plots.

In the source examples, ADA4528 operates at about 200 kHz chopping frequency, while ADA4522 uses approximately 4.8 MHz chopping with an additional offset/ripple correction loop around 800 kHz. A higher chopping frequency generally moves the unwanted energy farther away from the signal band.

Chopping-related peaks in ADA4528 noise density

ADA4522 noise-density features at higher frequencies

5. Match Source Impedances at the Two Inputs

Transient input currents interact with source impedance. If the two amplifier inputs see different impedances, the same transient currents produce unequal voltage errors and create a differential artifact.

The source shows this effect clearly with ADA4522: mismatched source resistance increases chopping-related spectral peaks. Precision designs should therefore equalize the effective source impedance seen by both inputs whenever possible.

Noise spectrum with matched and mismatched source impedances

6. IMD and Aliasing Matter in Wide-Bandwidth Systems

An input tone can mix with the chopping clock and generate products such as fIN ± fCHOP, fIN ± 2fCHOP, and 2fIN ± fCHOP. These products become problematic when they fall inside the intended signal band.

When an ADC follows the amplifier, chopping ripple and glitches can also alias into the sampled spectrum. The system should therefore combine a sufficiently high fCHOP, sensible ADC sample-rate planning, and an anti-alias filter between the amplifier and converter.

Example of chopping artifacts aliasing through ADC sampling

7. Filtering Is the Main System-Level Tool

The most effective system-level treatment is filtering. Designers can add a low-pass network between the amplifier and ADC, deliberately raise closed-loop gain to reduce amplifier bandwidth, or place a capacitor across the feedback resistor to introduce local filtering.

Two practical configurations for filtering chopper artifacts

8. Practical Design Checklist

1. Define the signal bandwidth and ADC sample rate, then identify the amplifier chopping frequency.

2. Choose fCHOP well above the wanted signal band whenever possible.

3. Match the effective source impedances seen by the two amplifier inputs.

4. For differential conditioning, prefer a dual-channel device rather than two unrelated single-channel chopper amplifiers.

5. Use a low-pass or anti-alias filter without compromising required settling time.

6. Verify the final spectrum for fCHOP harmonics, IMD, and aliased products.

Conclusion

Zero-drift amplifiers are not inherently limited to dc and very low-frequency applications. Their high-frequency artifacts are structured and predictable. With frequency planning, matched source impedance, synchronized channels, appropriate ADC sampling, and filtering, designers can preserve the excellent dc behavior of zero-drift amplifiers while extending them into wider-bandwidth precision signal chains.