LT6230 Troubleshooting Guide: Fixing the Most Common Problems Before They Fix You

The most common LT6230 failure report isn't a blown part — it's an amp that sits dead in the circuit because someone wired the ENABLE pin backwards. The LT6230 is a low-noise, low-power op amp from Analog Devices with 215 MHz gain-bandwidth, 1.1 nV/√Hz input voltage noise, and a 70 V/µs slew rate, running on supplies from 3 V to 12.6 V with rail-to-rail output. It's built for low-voltage signal conditioning: ultrasound front ends, active filters, ADC drivers, transimpedance amplifiers, and single-supply instrumentation amps — anywhere you need very low noise without burning power (4.6 mA max per amplifier). This post is organized around diagnosing real problems: first what the part actually is, then how to integrate it, the failure modes to design out, how to verify it on the bench, and finally a symptom-by-symptom troubleshooting guide.

Know the part before you debug it (What is the LT6230 and what is it for?)

The LT6230 is a single bipolar-input op amp specified on 3.3 V, 5 V, and ±5 V supplies, with the LT6230-10 variant internally compensated for gains of 10 or more. The headline numbers: 215 MHz GBW at A_V ≥ 1, 1.1 nV/√Hz voltage noise, 70 V/µs slew rate (±5 V typ), 1 mA typ / 4.6 mA max supply current per amplifier, and 10 µA max in shutdown. The datasheet calls its noise-voltage-times-√supply-current product (e_n·√I_SUPPLY) "among the most noise efficient of any op amp" — that's the whole point of the part: minimum noise per milliwatt.

The datasheet's own application circuits confirm the intended roles: an APD transimpedance front end with a 1.5 kΩ gain resistor (where the amp is quieter than its own gain resistor), a gain-100 single-supply instrumentation amp, a gain-10 bandpass filter, and a 20 MHz buffer stage driving ADCs. If your job is squeezing a small signal out of a low-power system, this part is aimed at you.

One variant warning up front, because it causes real field failures: the LT6230-10 is not unity-gain stable. Its 1450 MHz GBW and 250 V/µs slew rate (5 V, A_V = −10) are only specified at closed-loop gains of 10 or more. Put a -10 in a follower and you'll get oscillation or worse — use the base LT6230 for gain-of-1 work.

Wiring the LT6230 into your design (How do I integrate the LT6230 correctly?)

Packages and pinout. The LT6230 comes in a 6-lead TSOT-23 (S6): pin 1 OUT, pin 2 V−, pin 3 +IN, pin 4 −IN, pin 5 ENABLE, pin 6 V+. The dual LT6231 comes in SO-8 or 3×3 mm DFN-8 with conventional op-amp pinouts; the quad LT6232 is a 16-lead SSOP. Note there is no MSOP in this family — if your footprint library assumes one, stop and check.

ENABLE — the pin that causes the most trouble. It's active-low: drive ENABLE low (V_L max 0.3 V) to run the amp; drive it to within 0.35 V of V+ (V_H min 4.75 V on a 5 V supply) to shut down to 10 µA max. The pin can source/sink up to 95 µA of pin current, so whatever drives it low must sink that plus any pull-up current while staying under 0.3 V. If your logic runs on a different rail than the amp's V+, use open-drain logic with a pull-up resistor to the LT6230's V+ — the datasheet recommends exactly this, because a 3.3 V logic "high" won't reach the 4.75 V shutdown threshold. The exact pull-up value isn't specified in the datasheet; 10 kΩ–100 kΩ to V+ is standard practice and works.

Decoupling. The datasheet publishes no specific bypass values. Practice: 0.1 µF ceramic at each supply pin, close as possible, plus 1–10 µF bulk nearby. The part's 85 dB min PSRR only helps if the local supply impedance is actually low.

Feedback and load resistors. This is a bipolar-input part with up to 11 µA of input bias current. Keep resistor networks in the kΩ range and match the DC resistance seen by both inputs — a 10 kΩ source impedance times 11 µA is ~110 mV of offset, which dwarfs the 500 µV max input offset voltage.

Capacitive loads. There's no published absolute-max pF limit. The datasheet's overshoot curve (5 V, A_V = 1, R_L = 50 Ω) shows worst-case peaking around 100–200 pF: at 100 pF, 10 Ω of series output resistance gives ~32% overshoot, 20 Ω gives ~20%, and 50 Ω gives ~8%. A typical ADC input capacitance (a few to tens of pF) sits in the flat ~7% region with ~48–56° phase margin — fine as-is. If you land in the 100–200 pF band, add a series isolation resistor (up to ~50 Ω) and re-check settling.

Design out the LT6230's known failure modes (What should I watch out for with the LT6230?)

These are the part-specific traps, in rough order of how often they bite:

• Output inversion under heavy overdrive at gain ≥ 2. The datasheet warns that in gains of A_V ≥ 2, "the output can invert with very heavy overdrive." The fix it prescribes: limit input overdrive to 0.5 V beyond the supply rails. In a follower the behavior is different — diode D2's feedback path clamps the output about a diode drop below the input instead of fully inverting. If your gain stage ever "flips" the wrong way, this is why. Clamp the input externally; don't rely on the amp.
• No internal input series resistors — by design. A 100 Ω series resistor on each input would add 1.8 nV/√Hz, raising total noise from 1.1 to 2.1 nV/√Hz. The part was optimized to run without them, so don't casually add input resistance while chasing the noise spec. The trade: input current must be limited to ±40 mA absolute max, and the datasheet's sizing rule is 25 Ω of series protection resistance per volt of overdrive beyond ±0.7 V. The input diodes tolerate slew-rate overdrive and clipping transients on their own — add series R only for sustained large overdrive.
• ESD diodes on all inputs AND outputs conduct "unlimited" current past the rails. Only a transient limited to ≤100 mA is guaranteed harmless. A back-driven output — inductive kick, charged capacitor, another supply domain pulling the node — can destroy the part in steady state. Protect externally.
• The shutdown leakage trap. When disabled, output leakage is ~1 µA typ, which is fine — but current can flow into input-protection diodes D1/D2 if the output exceeds the input by a diode drop while the amp is off. If an external load pulls the output high while ENABLE is high and the input sits low, you'll draw current through the input diodes. Check your off-state bias network.
• Common-mode range tops out below the rail. On ±5 V the input range is −3 V to 4 V; on 3.3 V it's 1.15 V to 2.65 V (guaranteed by CMRR). The output swings rail-to-rail, but the input does not — keep common-mode inputs within ~1 V of the top rail or behavior degrades.
• Large-signal bandwidth is slew-limited. Full-power bandwidth is FPBW = SR/(2π·V_P). The 215 MHz GBW is a small-signal number; the guaranteed full-power bandwidth is 4.66 MHz min at 3 V_P-P. Don't budget large sine swings at the small-signal bandwidth.
• Thermal budget. Operating range is −40 °C to 85 °C with a 150 °C junction limit — but the DFN-8 package caps at 125 °C junction. The TSOT-23's 250 °C/W θJA is the hottest package; give it copper pour if it drives 20 mA loads. Output short-circuit is rated for indefinite duration, so a faulted output won't die instantly — it will just run hot.

Verifying the LT6230 on the bench (How do I test the LT6230?)

The datasheet has no production test script, so this is a bench plan built from its own EC-table limits and conditions. Reference conditions: V_S = ±5 V (or 5 V/3.3 V single), V_CM = V_OUT = mid-supply, ENABLE low.

1. Supply current first. Measure per-amp supply current: typ 1 mA, max 4.6 mA. Anything well above 4.6 mA usually means oscillation or a mis-terminated load. Toggle ENABLE and confirm shutdown drops to 10 µA max; verify output leakage ≤1 µA typ while disabled.
2. ENABLE timing. Turn-on is 300 ns typ; turn-off is 85 µs typ. If any downstream logic assumes fast shutdown, this is where it fails.
3. DC offset. Measure V_OS in unity gain (max 500 µV @ 25 °C, 600 µV over temp for the LT6230; drift 0.5 typ / 3 max µV/°C). If measured offset far exceeds your R×I_B estimate, investigate — bias current is typ 5 µA, max 10 µA @ 25 °C, 11 µA over temp, with offset current max 0.7 µA.
4. Output swing under your real load. No load: 50 mV (V_OL) / 60 mV (V_OH) max off the rails. At 5 mA: 200/215 mV. At 20 mA: 200/215 mV. At 20 mA: 200/215 mV. Short-circuit current is rated for indefinite duration, so a faulted output won't die instantly — it will just run hot.

Symptom-by-symptom troubleshooting (Why is my LT6230 circuit misbehaving?)

• Amp shows no output at all: Check ENABLE polarity first. The part runs only with ENABLE ≤ 0.3 V; shutdown requires ≥ 4.75 V on a 5 V rail. A 3.3 V logic "high" leaves the pin in the gray zone between 0.3 V and 4.75 V — neither fully on nor off. Fix with open-drain drive and a pull-up to the amp's V+.
• Draws far more current than expected / runs hot: Spec is 1 mA typ, 4.6 mA max per amp. Excess current almost always means output oscillation or a heavy/mis-terminated load. Check for ringing with a scope at the output pin, and verify the load impedance.
• Unexplained DC offset at the output: Bipolar-input bias current is the prime suspect — up to 11 µA through high-value resistors (10 kΩ × 11 µA ≈ 110 mV). Match the DC resistance at both inputs and lower network impedance. Then check against the 500 µV max offset voltage.
• Output inverts or latches on overdrive: The documented A_V ≥ 2 inversion mode. Clamp the input to within 0.5 V of the rails and size series protection at 25 Ω per volt of overdrive beyond ±0.7 V.
• Amp looks dead after an ESD hit or rail event: Over-rail current through the I/O ESD diodes — only ≤100 mA transients are guaranteed survivable. Look for external circuitry back-driving the output past a rail in steady state.
• Disabled amp still leaking current: Check the off-state bias. Output leakage itself is ≤1 µA typ, but if the externally-pulled output sits more than a diode drop above the input while ENABLE is high, current flows through input diodes D1/D2. Red bias the output node, not the amp.
• Bandwidth or slew rate far below spec: Confirm you have the right variant — LT6230 is 215 MHz at A_V ≥ 1; the LT6230-10 is 1450 MHz but only at A_V ≥ 10 and will misbehave badly at low gain. Then check your supply: slew rate min is 35 V/µs at 3.3 V/5 V single-supply over temperature, and full-power bandwidth is only 4.66 MHz min at 3 V_P-P.
• Input driven near the top rail: The input common-mode range tops out at 4 V on ±5 V and 2.65 V on 3.3 V — the input is not rail-to-rail even though the output is. Pull the input back below the limit.

Triage order: measure supply current and output swing at your real load first — those two catch most oscillation and loading problems. Then verify V_CM stays under the non-rail-to-rail top limit, and that ENABLE actually reaches its thresholds. The two LT6230-specific killers remain overdrive inversion at gain ≥ 2 and the active-low ENABLE with its high shutdown threshold — design both out on paper before the board comes back.

Known limitation: a standalone 3.3 V typical slew-rate value was not found in the datasheet tables — only the 35 V/µs over-temperature minimum covering both 3.3 V and 5 V single-supply. Treat any "3.3 V typ" slew figure as unverified.

Part page: LT6230.