This report compiles bench measurements of the OPA397’s input-referred noise and shows how closed‑loop gain and bandwidth choices change the noise outcome. Measurements compare multiple gain settings and source impedances to separate amplifier voltage noise, amplifier current noise interacting with source resistance, and resistor thermal noise. Readers will learn practical design tradeoffs between gain, bandwidth and integrated input noise.

Scope: tested parts are OPA397 devices in standard SOIC packages, gains from unity to 100 (unity, 5, 10, 50, 100), source impedances of 50 Ω, 1 kΩ and 10 kΩ, and measurement bandwidths of 0.1 Hz–10 kHz and 10 Hz–100 kHz. One clear takeaway: raising closed‑loop gain reduces bandwidth and can both increase or decrease integrated noise depending on source impedance and feedback resistor choices.

Background: OPA397 essentials & noise fundamentals

Device snapshot — what to pull from the datasheet

For quick reference extract: input voltage noise density (nV/√Hz), input current noise (A/√Hz), gain‑bandwidth product (GBW), slew rate, input offset, recommended supply voltages, and any suggested external components or compensation notes. Below is a concise spec table (not a datasheet copy) to guide theoretical noise calculations and expected behavior.

SEO note: this section includes OPA397 and the long‑tail phrase OPA397 noise specs to help readers locate the key parameters for theoretical noise predictions and measurement planning.

Noise basics every reader needs

Define input‑referred voltage noise density as e_n(f) in nV/√Hz and integrated (RMS) noise as the square root of the noise power across a bandwidth. Key formulas: integrated noise = sqrt(∫ e_n(f)^2 df). Resistor Johnson noise = sqrt(4 k T R B). When source impedance is non‑negligible, amplifier current noise i_n produces voltage noise i_n·R_source that adds in quadrature to e_n. Closed‑loop gain and feedback resistor choices set bandwidth and resistor noise contributions.

Measurement setup & methodology (data-driven)

Test circuit & gain configurations to document

Describe non‑inverting and inverting topologies with explicit resistor values for each gain tested: unity (gain=1), 5 (e.g., Rf=40k, Rg=10k), 10 (Rf=90k, Rg=10k), 50 (Rf=490k, Rg=10k), 100 (Rf=990k, Rg=10k) — adapt component values to standard series while keeping feedback resistor magnitudes practical. Use terminations of 50 Ω, 1 kΩ and 10 kΩ to reveal current‑noise effects. Layout notes: short input traces, guard rings for high‑impedance nodes, solid ground plane and decoupling next to the device significantly reduce parasitic and measurement variability.

Instrumentation, bandwidth and correction

Use a high‑resolution FFT analyzer or spectrum analyzer with a low‑noise preamp if needed; typical settings: sample rate / span to cover DC–100 kHz, linear averaging (16–64 traces), RMS detector. Measure and record the instrument noise floor by terminating the analyzer input with the same source impedance and subtract it in quadrature from measured spectra. Report both noise density plots and integrated RMS noise over 0.1 Hz–10 kHz and 10 Hz–100 kHz. A reproducible measurement log should capture supply voltage, temperature, gain, resistor values, input termination, instrument model and settings.

Measured input noise vs gain — results & interpretation

Presenting the data: tables & plots to include

Include noise density vs frequency plots for each gain and source impedance, an integrated RMS noise vs gain plot, and a table listing noise density at 1 Hz, 10 Hz and 1 kHz. Figure captions should state measurement bandwidth, circuit topology and source impedance; use long‑tail captions such as "OPA397 measured input noise at gain 10, non‑inverting, 1 kΩ source" for clarity and SEO value. Provide alt text for any images reflecting these phrases.

Interpreting the tradeoffs: where noise comes from

Break measured noise into amplifier voltage noise, amplifier current noise × source impedance, and resistor thermal noise. Compute theoretical noise using datasheet e_n and i_n values and resistor Johnson noise, then compare to measurements. Discrepancies often point to measurement floor, board layout parasitics or extra capacitance in feedback. For high source impedances, current‑noise terms dominate; for low source impedances, amplifier voltage noise and resistor noise govern.

Gain, bandwidth & stability tradeoffs

Closed‑loop bandwidth, GBW behavior and phase margin effects

Measure closed‑loop −3 dB bandwidth vs gain and track phase margin at representative gains. Because GBW is approximately constant, increasing closed‑loop gain reduces bandwidth and can raise integrated noise if the reduced bandwidth still includes the dominant low‑frequency noise. Report unity‑gain behavior and note any phase‑margin degradation as feedback resistor magnitude increases or as capacitive loads are introduced.

Using bandwidth limiting / filtering to lower integrated noise

Simple RC low‑pass filters in the feedback or at the input can reduce integrated noise; choose corner f_c ≈ 2–5× the signal bandwidth to avoid signal distortion. For example, adding a 1 nF across a 90 kΩ feedback resistor yields f_c ≈ 1.77 kHz. Beware that added capacitance at the feedback node can reduce phase margin; mitigate with a series compensation resistor in series with the feedback capacitor or a small snubber network to maintain stability.

Practical design examples & recommended OPA397 configurations

Low‑noise sensor front end (recommended parts and values)

Example A — low‑gain, wideband preamp for dynamic signals: non‑inverting gain=5, Rf=40k, Rg=10k, expected bandwidth ≈ GBW/5, predicted integrated noise dominated by amplifier e_n and resistor Johnson noise. Example B — high‑gain, low‑bandwidth preamp for DC sensors: gain=100, Rf=990k, Rg=10k, add input RC (1 kΩ ‑ 0.1 µF) for 1.6 kHz corner; expect increased resistor noise from large Rf but reduced bandwidth yields lower integrated wideband noise for low‑frequency measurements. Include predicted noise calculations from datasheet values next to measured placeholders.

Layout, resistor selection and PCB practices that matter

Checklist: use low‑noise metal film resistors with low tempco, minimize feedback resistor values where possible to reduce Johnson noise, keep input traces short, place bypass capacitors close to supply pins, and use a solid ground plane. Don'ts: avoid long input runs, avoid large stray capacitance on feedback nodes, and avoid single‑point grounds crossing noisy digital returns. Quick fixes include adding guard rings and relocating sensitive traces away from switching regulators.

Quick checklist, measurement templates & SEO pointers (actionable)

Measurement & design checklist for readers

• Verify supply decoupling and quiet reference; log supply and temperature for each run.
• Confirm instrument noise floor with matching termination and subtract in quadrature.
• Record source impedance and test both polarities; run thermal drift and long‑time averaging checks.
• Troubleshoot ground loops, power‑supply ripple and microphonics if measured noise exceeds theory.

SEO and publication guidance for this article

Recommended long‑tail keywords: "OPA397 input noise measurement", "OPA397 noise vs gain tradeoff", "measure OPA397 noise density". Place OPA397 in the intro and in at least two H2 titles (Background and Measured input noise vs gain) and use "input noise" and "gain" in H3 captions and figure alt text. Use descriptive file names such as opa397-noise-vs-gain.png for figures.

Summary

• Measured trends show that closed‑loop gain choices on the OPA397 strongly affect bandwidth and integrated input noise; for low source impedances, raising gain often lowers integrated noise by reducing bandwidth, while for high source impedances current‑noise terms can dominate.
• The dominant noise contributors are amplifier voltage noise at low source impedances, current‑noise×R at high source impedances, and resistor Johnson noise; compare datasheet OPA397 noise specs to measurements to diagnose discrepancies.
• Practical recommendation: for most low‑noise sensor front ends start with moderate gain (5–10) and a modest feedback resistor (tens of kΩ), then tune a gentle RC filter to limit bandwidth without compromising phase margin.

Notes for the writer (brief)

Maintain a data‑driven, instructional tone and place measurement conditions directly next to each figure and table. Ensure every H2 above contains the specified H3 subsections and include figure/table templates and a measurement log as appendices if possible. Keep language US‑centric, use SI units for resistance and frequency, and avoid referencing external websites in the body.

What is the OPA397 input noise at gain 10?

Answer: Typical measured input noise density at gain 10 will approximate the datasheet voltage noise density above the amplifier's low‑frequency corner; integrated RMS noise depends on bandwidth and source impedance. Report both density at 1 Hz/10 Hz/1 kHz and integrated RMS over your chosen bands to give a complete picture (50–100 words).

How does gain affect measured input noise for OPA397?

Answer: Increasing closed‑loop gain reduces closed‑loop bandwidth (GBW constraint), which can reduce integrated broadband noise if voltage noise dominates; however, higher feedback resistor magnitudes can increase resistor thermal noise and magnify current‑noise effects, so optimal gain balances these competing terms.

Which gain/bandwidth approach should designers try first?

Answer: For general low‑noise sensor applications start with a moderate gain of 5–10 and feedback resistors in the tens of kiloohms, then limit bandwidth slightly above your signal band with an RC filter; this typically yields a favorable tradeoff between noise, bandwidth and stability.