LCD Low Power Mode: Measure Standby Current and Interface Leakage

An LCD low power mode that measures well on the bench can still drain a battery in the product. The meter may be in the wrong place, the interface pins may be back-powering a rail, or the touch controller may be kept awake for wake-on-touch. This guide defines the measurement node, the pin-state matrix and the wake-state report that make a standby figure reproducible and comparable.

Where should LCD low power mode current be measured?

Measure at the battery or the rail that actually stays on.

A sleep figure is meaningful only at the node that stays powered; measuring at the input includes regulators and other loads, while a rail measurement isolates the display.

The measurement point determines what the number means. At the battery or input, the current includes every load that remains powered, including regulators, the host in its lowest state and any always-on indicators. At a panel rail, the current describes only the display’s share, but it excludes the regulator’s quiescent draw and the losses in the conversion.

Both measurements are useful, but they answer different questions. The input measurement predicts battery life; the rail measurement tells you whether the display is behaving as expected. Record both, and state which one the standby specification uses.

A measurement taken with a multimeter in series can also change the sleep state, because the voltage drop may be enough to reset a low-voltage rail. The measurement should be taken after the product has reached its defined sleep state and the current has stabilized. A reading taken during the transition includes the wake or sleep sequence and is not a standby figure.

Which interface pins can feed a nominally off rail?

Pull-ups and driven pins can power an off rail.

An I2C pull-up, a DSI lane or a GPIO held high can forward-bias a protection diode and charge the panel rail, so the off rail is not at zero.

A panel or touch controller that is nominally off can still receive energy through its interface pins. A pull-up to a live rail, a host GPIO driven high, a DSI lane in a high state or a level shifter with a live supply can forward-bias the ESD diode inside the device and charge its local rail. The rail may sit at a diode drop below the driving voltage, and the device may draw current or partially wake.

A pin that is high while the device is off can also keep an internal regulator enabled, so the current may depend on the pin’s logic level, not only on the supply. The symptom is a standby current that is higher than expected and a rail that is not at zero. The cause is not always obvious, because the pin may be idle from the software’s point of view while the hardware holds it high.

The test is to measure the panel rail with the supply disabled and the interface pins in each possible state. A rail that rises when a pin is driven is being back-powered. Record the pin state and the rail voltage together.

How do instrument burden voltage and range affect the result?

Burden voltage and range can change the measured current.

A shunt inserts resistance that reduces load voltage, and a coarse range hides low currents; use a source-measure unit or calibrated shunt and record the setting.

A current meter is not invisible. A shunt or a multimeter on a current range inserts resistance in the supply path, and the resulting burden voltage can reduce the voltage at the load. If the load is sensitive to that drop, it may draw less current, reset or behave differently from the product’s real supply.

The range also matters. A meter on a high range may not resolve the microamp-level current of a sleeping display, while autoranging can introduce switching artefacts. A source-measure unit can hold the voltage constant and measure the current, which avoids the burden problem. A calibrated shunt with a low burden and a suitable amplifier is another option.

Record the instrument, the range, the burden voltage and the measurement duration. A single number without those details cannot be compared with another lab’s result.

What distinguishes panel sleep from complete power removal?

Sleep retains state; power removal does not.

A sleeping panel keeps rails and registers and wakes quickly; a panel with rails removed loses state and needs full initialization, changing current and latency.

In sleep, the panel may keep its logic rail and some registers alive while the display is blanked and the backlight is off. The current is lower than active but not zero, and the wake path can be short because the panel retains its configuration. In complete power removal, the rails are off and the panel loses state; the wake path must repeat the power sequence and initialization.

The difference is not only current: the wake path, the initialization time and the failure behaviour all change. The two states have different currents, different wake times and different failure modes. A product that claims a low standby current by removing power must also account for the wake energy and the time to first image. A product that claims fast wake by retaining power must account for the higher standby current.

Record which state the specification describes and what remains powered. The power sequencing and inrush guide covers the rail behaviour that the two states share.

When must the touch controller remain awake?

Touch wake requires the controller to stay powered.

If the product wakes on touch, the touch controller and interface must remain powered or scanned, which adds to standby and must be measured separately.

A wake-on-touch feature requires the touch controller to detect a contact while the rest of the product sleeps. That means the controller, its supply and usually its interface must remain at least partly powered. The current may be small, but it is part of the standby budget and can dominate if the display itself is fully off.

Some controllers support a low-power scan mode with a reduced report rate and a wake interrupt. Others require a periodic wake of the host to poll the controller. A controller that wakes the host periodically has an average current that depends on the poll rate. Each approach has a different average current and a different latency. Measure the controller’s share separately, with the display in its sleep state.

The Linux device power management documentation describes the suspend and resume ordering that determines whether the controller is still powered when the wake event occurs.

How should wake latency and lost state be reported?

Report wake time and what state was lost.

A standby figure without wake time and state description is incomplete; the product may save current by losing configuration that costs seconds to restore.

A low standby current is only useful if the product can wake quickly enough and restore what it needs. Report the time from the wake event to the first valid image and the state that was lost: panel configuration, touch calibration, backlight level or host context. If the product must reload a configuration, include that time in the wake path.

The report should state whether the wake was triggered by a button, a touch, a timer or a communication event, because each path may have a different latency. It should also state whether the panel was in sleep or fully unpowered.

The wake path should be measured with the same instrumentation and the same pin states as the standby measurement. Record the wake energy as well as the time. A fast wake that draws a large current burst can cost more average energy than a slower, gentler wake at a low duty cycle.

Which host-pin states belong in the current test matrix?

The matrix must list every host pin state.

High-impedance, driven low, high and pulled-up states produce different leakage paths, so record every interface pin state during sleep.

The host can leave an interface pin in several states: high-impedance, driven low, driven high, pulled up or pulled down. Each state creates a different path into the panel or touch controller. A pin that is high while the device is off can back-power it; a pin that is floating can pick up noise and cause the input to oscillate.

The matrix should list every pin that connects to the display subsystem: reset, enable, I2C or SPI lines, DSI lanes, TE, backlight control and any interrupt. For each, record the state during sleep and the resulting rail voltage or current.

The matrix should distinguish a pin left floating by the host from one explicitly pulled to a rail. The product’s real sleep state may be a combination, not a single setting. Test the combination that the firmware actually applies, and record it as one row.

What evidence supports a reproducible standby figure?

Record instrument, range, rail and pin states together.

A reproducible figure needs instrument and range, burden voltage, temperature, rail states, pin states and duration, not just a single microamp value.

The evidence package should let another engineer repeat the measurement and understand what it includes. Record the instrument and range, the burden voltage, the supply voltage, the temperature, the duration and the state of every rail. Add the pin states from the matrix and the firmware version that produced them.

State whether the figure includes the touch controller, the backlight driver, the regulators and any always-on indicator. A figure that includes different loads is not comparable, even if the number looks similar.

The table below is a template for the standby record. It is not a set of limits; the values come from the panel, the controller and the product requirement.

Standby current evidence record

Workload or state	Pixel or buffer format	Measured traffic or energy	Constraint	Evidence source
Display active	Native panel format	Input and rail current	Reference state	SMU log
Panel sleep, rails on	Blanked	Rail current per supply	Retained registers	SMU log
Panel power removed	No framebuffer	Leakage and pin current	Interface pin states	Pin-state matrix
Touch wake enabled	Controller low-power scan	Controller share	Wake latency	Controller log
Wake to first frame	Startup framebuffer	Integrated wake energy	Time to image	Current capture

The TPS22918 load-switch datasheet documents controlled rise and discharge for a switched rail, which affects how the rail reaches zero. A backlight driver such as the LP8556 and its automotive counterpart the LP8860-Q1 document their own standby and dimming behaviour. The ST AN4861 LTDC application note covers the display controller side of the sleep and wake path.