Choosing a TFT Interface: Five Questions Before You Freeze the PCB
The interface locks your pin budget, your cable length and your frame rate at the same time. Answer these five questions before the schematic is frozen.
An interface question sounds like a datasheet question. It is really four questions at once: how many pins can you spare, how far is the display from the board, how fast does the picture have to change, and what can your host actually drive. Get those four answers first and the interface picks itself.
SPI is the one that fits a small microcontroller. Four to six wires, no external frame buffer, and firmware you can write in an afternoon. It is enough for a 2.4-inch panel showing a menu that changes a few times a second, and not enough for full-screen video. If your UI is mostly static with partial updates, SPI is not a compromise — it is the correct choice, and it keeps the board small.
The 8080 parallel interface is the classic MCU interface: 8 or 16 data lines plus a handful of control lines. It is fast enough for animation on panels up to roughly five inches and it does not need a dedicated display controller, but it eats pins. On a 64-pin MCU it is comfortable; on a 32-pin part it is not.
RGB is where projects get into trouble. A parallel RGB panel takes 16, 18 or 24 data lines plus pixel clock, HSYNC, VSYNC and DE, and it expects a continuous stream of frames — which means your host needs a display controller and enough memory for a frame buffer. In exchange you get a simple, cheap panel and a bright, fast picture. It is the right answer for 4.3 to 10.1 inches in industrial equipment, and the wrong answer for a board without a controller.
LVDS is the same idea with differential pairs. Fewer wires for the same resolution, longer cables, better behaviour in an electrically noisy cabinet. Above roughly 10 inches, or whenever the display sits a metre away from the electronics, LVDS is usually the practical choice. It costs a little more in silicon and a lot more in layout discipline.
MIPI-DSI is what mobile SoCs speak. Two to four differential lanes, high bandwidth per wire, and a panel that can be very thin. If your product is built around an application processor — a tablet-class board, a smart terminal, a medical handheld — MIPI is often the only option with enough bandwidth. It is also the least forgiving: length matching, impedance control and lane polarity all have to be right the first time.
The five questions, in order. One: what resolution and refresh rate does the UI actually need? Two: how far is the panel from the host board, and does the cable bend? Three: how many pins are free after the rest of the design? Four: what is the electrical environment — motor drives, relays, radios? Five: what is the expected volume, because that decides whether a panel with an integrated controller is affordable.
Four mistakes we see repeatedly. Choosing RGB and discovering there is no frame buffer. Running SPI over a 30 cm flat cable and blaming the panel for corruption. Routing MIPI without controlled impedance and wondering why one board in ten fails. Forgetting that the backlight needs its own driver and that a capacitive touch panel needs an I2C bus, a reset line and an interrupt — three signals nobody budgeted for.
How we normally work. We supply the panel, the FPC and a suggested initialisation sequence, and we will confirm the interface against your host before you commit to a PCB revision. Tell us the MCU or SoC, the panel size, the cable length and the refresh requirement — those four numbers are enough for us to say which interface you should be designing around, and which ones to stop considering.
