Bring-Up Debugging: No Display, Wrong Colours, Flicker and Everything in Between

The first power-on of a new display design produces a small set of symptoms, and each one has a short list of likely causes. Working through them in a defined order turns a frustrating afternoon into a checklist. The same sequence applies later, in the field, when a unit that worked for a year develops an intermittent fault.

Part of the Interfaces, Touch & Signal Integrity guide — 39 articles on RGB, SPI, LVDS, MIPI, eDP and the touch controller.

Start with power and timing before signals

Most “no display” faults are not video faults.

Before examining the video path, confirm every supply rail is present, at the correct level, and within the module’s tolerance, and that they come up in the specified order – the safety requirements for that supply and protection design are commonly written against IEC 62368-1. Then confirm the pixel clock and the data enable or equivalent signal are running at the expected rates, because a panel with correct power but no clock shows nothing at all while appearing healthy to a multimeter. Checking the backlight separately – with a fixed current or a bench source – isolates the panel from its illumination.

This ordering matters because a fault in the power or timing domain mimics a video fault, and vice versa. CDTech’s notes on panel bring-up, wrong timing and flicker follow the same sequence.

No display at all

Four candidates, checked in order.

• A missing or incorrect supply rail, including the panel’s high rail and the driver board’s logic rail.
• No clock, or a clock present but with a frequency outside the panel’s specification.
• A reset or enable signal that is not released, or released at the wrong time relative to the supplies.
• Incorrect configuration of the host’s display controller for the panel’s timing.

Measuring the rails and the clock at the panel connector, rather than at the host, eliminates the cable and connector as variables in the same step.

Image present but wrong colours

Colour faults are mapping faults.

The classic causes are a swapped colour order in the host’s configuration, an incorrect bit-depth setting, a panel configured for a different colour mapping, or an inverted pixel format. Where the image is recognisable but the colours are wrong in a consistent way – red appearing blue, for example – the cause is almost always a configuration mapping rather than a hardware fault. Where the colours are wrong in an inconsistent way, such as a gradient appearing as bands, the bit depth or the interface’s data format is the place to look.

Drawing a test pattern with known values on each channel, then comparing the displayed result to the expected, resolves the mapping in a few minutes and produces a record for the design file.

Flicker, instability and rolling images

These symptoms point at timing margins.

A rolling or unstable image usually means the pixel clock or refresh rate is outside the panel’s supported range, or the vertical timing parameters are incorrect. Flicker that appears only at certain brightness levels points at the backlight driver rather than the panel. Flicker that appears only at certain content or refresh rates suggests the interface is at its limit. Distinguishing between these requires knowing the brightness setting and content when the symptom occurs, which is why observing the behaviour while changing one variable at a time is more informative than static measurement.

Where the flicker is a modulation of the backlight rather than an image artefact, it will be visible on a screen showing a solid colour, which separates the two causes immediately.

Sparkle, noise and intermittent dropout

Intermittent symptoms are almost always marginal links.

Sparkle – isolated bright or wrong pixels appearing randomly – indicates a marginal differential link: a pair mismatch, an impedance discontinuity, or a timing margin that is consumed at some temperatures. Dropouts that correlate with vibration indicate a connector or cable. Dropouts that correlate with a motor starting indicate a supply or common-mode event. Each has a different fix, and the correlation with a physical event is the fastest way to distinguish them.

Where the symptom appears only in the assembled machine, the cable route and the return path are the first things to examine, as described in CDTech’s notes on long display cables and signal integrity.

Partial images and shifted pictures

Geometry faults come from timing parameters.

An image that is present but shifted, wrapped, or missing a band at one edge usually has incorrect porches, sync polarities or total counts. Because these parameters are recorded in the panel’s timing table, the check is a comparison rather than an investigation. Where the panel’s table is unavailable, the designer is reduced to experiment, which is why obtaining it early saves time later.

A black border on one side with the rest of the image intact is often a scaling artefact rather than a timing error, so checking whether a scaler is in the path is a useful early step.

Touch faults during bring-up

Touch faults often follow the backlight.

A touch panel that misbehaves during first bring-up usually has a layout or grounding issue rather than a broken sensor: the touch FPC running close to the backlight driver, a reference that is not established, or a controller configured for a different glass geometry. Testing with the backlight off and then on separates an interference problem from a configuration problem, and testing with the display in the intended enclosure reveals whether the mechanical assembly changes the behaviour.

Because tuning parameters can mask a layout problem, recording raw signal data during bring-up is more useful than adjusting gains until the symptoms are tolerable.

A diagnostic sequence that works

Nine checks in order.

1. Confirm supplies and their sequencing at the panel connector.
2. Confirm the clock and the data-enable or equivalent signal.
3. Confirm the reset and enable signals release in the correct order.
4. Verify the host’s controller configuration against the panel timing table.
5. Test the backlight independently with a fixed current.
6. Display a known test pattern and verify colour mapping.
7. Check the interface electrical behaviour with an eye diagram if sparkle appears.
8. Test touch with the backlight off and on, and with the enclosure assembled.
9. Repeat the critical checks at the temperature extremes.

Following this order means each step is performed once and its result recorded, which is what makes the second unit faster than the first.

Recording what was found

A bring-up log saves weeks later.

The useful record includes the panel’s revision, the host configuration, the measured timing parameters that worked, and any symptoms and their causes. When the same design is produced in a second factory, or when a field failure needs analysis, that record answers questions that would otherwise require rebuilding the setup. It also documents the difference between what the datasheet specified and what the working configuration actually was.

Where an issue is resolved by a change to the host configuration, note both the previous and the new values. That record prevents the change being lost in a later firmware merge.