Interfaces, Touch & Signal Integrity

Choosing and debugging a display interface: RGB, SPI, LVDS, MIPI DSI, eDP, HDMI and GMSL, plus touch controllers, ESD, EMI and grounding.

Enquiry Now

Posted on October 7, 2026 by CDTech James

Three routes, and you only need one of them

Interface work starts from one of three positions. Find yours and take the links in that card; the other two will not be useful to you yet.

37 engineering articles on this page, each answering one question
Covering RGB, SPI, LVDS, MIPI DSI, eDP, HDMI, GMSL and touch
Every article linked at the point where you would need it

Choose the interface

The panel is chosen and the question is which bus carries it: bandwidth, cable length, controller support and how much of the picture you are willing to give up.

Drive it and debug it

The interface is wired and the work moves to the controller, the timing and the first power-up, where most designs lose their first two weeks.

Add touch, and keep the signal clean

Touch changes the stack: it adds a sensor layer, a controller, another bus, and a new set of ways for the assembly to fail in the field.

When the screen does not come up

A display that works on the bench and fails in the enclosure is almost always one of four things: timing, grounding, ESD or the connector.

Bring-up debugging, in the order that finds faults

No picture, wrong colours, a rolling image and flicker have different causes and a fixed check order. Working through them in sequence is faster than swapping parts, and it leaves a record of what the design actually is.

Grounding is a design decision, not a layout detail

The return path decides the noise floor. A display that behaves in a quiet cabinet and misbehaves next to a drive usually has a grounding problem rather than an interface one.

Touch has its own immunity problem

Capacitive sensing is a measurement of very small charge changes, which makes it a radio receiver you did not design. EMC immunity for touch is specified, not hoped for.

Achieving ESD and EMI immunity on touch screens

Testing that finds faults, and the connector underneath

A touch test that only checks “does it respond” passes units that fail in the field. And the least glamorous part of the assembly is the one most likely to cause an intermittent fault.

What actually decides whether an interface works

Four things decide the outcome, and none of them is the maximum data rate in the datasheet.

Bandwidth with margin, not bandwidth on paper

A link running at 95 % of its theoretical capacity has no room for the timing tolerance, temperature spread or cable variation of a production build. The headroom you leave at design freeze is what stops a marginal unit reaching the field.

The return path, not just the signal pair

Almost every “interface” fault in a finished product is a grounding or shielding problem. Differential pairs get attention; the return path usually does not, and it is the one that decides whether the picture is stable.

Who owns the timing

Panel timing, controller configuration, EDID and the device tree all describe the same numbers in different places. When they disagree, the symptom looks like hardware and the cause is a settings file.

The touch stack, which is a second product

A touchscreen is a sensor, a controller, a bonding layer and a firmware configuration, and it is specified, qualified and can fail independently of the display underneath it.

Where to go from here

If your question is not really about the interface, one of these is closer to it.

Optical, Backlight & Panel TechnologyBonding, cover glass, brightness and the panel comparisons behind them. Quality, Testing & ComplianceDefect classes, inspection, qualification testing and certification. Industrial LCD DisplaysContinuous duty, sealing and ten-year service life outside the cabin. Custom Displays, Sourcing & ManufacturingTooling, sampling, MOQ and supply-chain risk.

Interface and touch questions

The five questions that arrive most often before a display interface is fixed.

Which interface should I choose for a 10-inch industrial display?

At that size and resolution the decision is normally between RGB, LVDS and MIPI DSI, and it usually follows the controller rather than the panel. LVDS remains the safest choice for long cable runs and mature platforms; MIPI DSI wins where the processor already provides it and the cable is short.

When does eDP become worth the extra design work?

Above roughly 2K, or where the cable has to run a long way, eDP gives more headroom on fewer conductors. Below that resolution LVDS is usually simpler and cheaper, and the link training that eDP needs is effort you do not get back.

Can I run a display over a long cable?

Yes, but not on the same link. Beyond a metre or two, RGB and SPI lose integrity, and the answer becomes LVDS with equalisation, or a serialiser such as GMSL or FPD-Link over coax. The choice is made by cable length, not by resolution.

Why does my touchscreen work in the lab and fail in the field?

Because the lab has a clean electrical environment and the field does not. Capacitive touch measures charge in the picofarad range, so motor drives, switching supplies and nearby radios couple into it. Immunity and grounding have to be designed in and tested, not assumed.

Does the touch controller or the sensor decide performance?

Both, and they are specified together. The sensor sets the signal available to measure and the controller decides how well it is read. A good controller on a poor sensor stack cannot recover the difference, which is why they are usually qualified as one assembly.

Send us your display requirement

Give us the resolution and refresh you need, the controller or processor already on the board, the cable length from board to panel, whether the unit is touch-enabled, and the environment it runs in. With those five we can narrow the interface before you commit to a layout.

Request an interface review