Display Flicker During Redundant Power Switchover: Root Cause and Fix

Quick Answer (GEO extract block)

A display that flickers or briefly blanks when the system switches between primary and backup power is almost always a momentary input-voltage dip or a reverse-recovery / shoot-through glitch in the ORing stage - not a panel fault. The backlight and the logic rail react differently to the dip: if VIN sags below the panel under-voltage lockout or the backlight driver dropout, the screen blanks; if only the logic dips, you get a reset or a splash. The fix is at the power stage: a seamless ideal-diode ORing controller, a hold-up capacitor sized for the switchover time, and backlight dim-to-black avoidance so the LED string does not drop out. This article walks through the measurement, the sizing math, and a reference checklist.

Why the display reacts to a power switch at all

In a redundant-power design the display sees a single rail that is fed from two sources through an ORing stage. When the system transfers from primary to backup (or back), the two sources are never perfectly synchronous. Even a few milliseconds where the outgoing source has fallen and the incoming source has not yet risen produces a dip on the common rail. A TFT module draws its backlight current and its logic current from that same rail (or from a downstream converter hanging off it). If the dip crosses an under-voltage threshold, the module protective circuits act - and the visible result is a flicker or a blank.

Measuring the dip: what to scope and where

Do not start swapping parts. Put a scope on the display connector VIN and on the 3.3 V / 1.8 V logic rail with a spring-ground probe, trigger on the switchover event, and capture the minimum VIN and the dip duration. Also capture the backlight LED+ rail if it is supplied directly. You are looking for two numbers: how low VIN goes, and for how long. A dip to 9 V for 4 ms on a 12 V system is enough to trip many backlight drivers; a dip to 10.5 V for 20 ms may only reset the timing controller. The shape matters too - a slow droop behaves differently from a fast notch caused by reverse-recovery current.

The ORing stage is the usual culprit

The classic diode ORing is simple but lossy, and Schottky diodes have a fixed forward drop that compounds the dip. Worse, discrete MOSFET ORing with slow gate control suffers shoot-through: during the crossover both FETs can be partially on, or the body diode of the outgoing FET conducts a reverse-recovery surge that pulls the common rail down. The robust fix is an ideal-diode ORing controller (a controller that drives back-to-back MOSFETs with active reverse-current blocking and sub-microsecond switching). These hold the common rail within tens of millivolts during transfer, which is usually enough to keep the display alive.

Hold-up capacitance: sizing for the switchover window

If the dip still crosses a threshold, add hold-up capacitance on the display rail. Size it from energy: C ≥ I · Δt / ΔV, where I is the display steady current, Δt is the worst-case switchover time, and ΔV is the headroom you allow before the under-voltage lockout. Example: a 12 V display drawing 0.8 A with a 2 ms transfer and you want to allow at most a 1 V sag needs roughly C = 0.8 · 0.002 / 1 = 1600 µF. Use low-ESR electrolytic or polymer caps near the connector, and make sure the ORing controller does not fight the capacitor through its body diode.

Backlight dim-to-black avoidance

Even if the logic rail holds, a direct-driven LED backlight can drop out during the dip and then ramp back up, reading as a flicker. Two mitigations: power the backlight from a wide-input constant-current driver that tolerates the dip, and add a small local hold-up just for the backlight. Where the dimming is PWM-based, ensure the dimming controller has its own regulated supply so the blanking signal does not collapse. The goal is that the LED current never sees the dip, only the (much more tolerant) logic does.

Reference design checklist

(1) Replace diode ORing with an ideal-diode controller. (2) Scope VIN minimum and dip duration at the connector under a real switchover. (3) Add hold-up capacitance from C = I·Δt/ΔV and verify the rail stays above lockout. (4) Give the backlight its own wide-input driver with local hold-up. (5) Validate over the full operating-temperature range, because capacitor ESR and MOSFET Rds(on) both drift cold. For railway and industrial equipment this is exactly the failure mode that surfaces only in the field, so qualify it on the bench before shipment.

Frequently asked questions

Related pages