Why Your SPI Display Manages 5 fps - and When to Leave SPI

When an SPI screen feels slow, the instinct is to blame the library. Usually the library is fine and the arithmetic was never done: a frame has a fixed size, the bus has a fixed rate, and dividing one by the other tells you the ceiling before you write a line of code. This guide does that arithmetic, shows what software can and cannot claw back, and marks the point where the correct fix is a different interface, not a faster loop.

Written by the Junxian applications engineering team. Figures quoted for “our models” are calculated from the 103 models on this site that carry a full parameter set, so they change when the catalogue changes.

The arithmetic: frame size over bus rate

A 240x320 panel at 16 bits per pixel is 153.6 kB per full frame - about 1.23 megabits. On a 40 MHz SPI clock the theoretical ceiling is roughly 32 full frames per second, and command overhead, gaps between transfers and whatever else the CPU is doing typically land you nearer 20. Move to a 480x640 panel - both dimensions doubled, so the frame is four times the size - and the same bus now caps around 8 fps theoretical, nearer 5 once overhead is counted. No driver setting changes this - it is the bus, not the code.

Run the numbers for your panel

Enter your resolution, colour depth and SPI clock - the estimate updates as you type.

What overclocking the bus buys, and what it risks

Driver ICs commonly tolerate SPI clocks above their datasheet figure, and many boards run 60 or 80 MHz successfully - but tolerance is not specification. Whether a given module survives a given clock depends on the driver IC batch, the flex length and the layout, so a margin that works on your bench can fail on a longer cable or a colder morning. Overclocking is a prototype convenience; a product should run the specified clock and get its speed elsewhere.

What DMA actually fixes

DMA does not make the bus faster - the bits still take the same time. What it does is free the CPU during the transfer, which removes the gaps between chunks and lets the application keep working, so real throughput moves closer to the theoretical ceiling and the UI stops stuttering. Worth doing, and still bounded by the same arithmetic.

Partial redraw: the honest big win

Because SPI panels hold the frame in the driver IC's own RAM, you only pay for the pixels you change. A UI that redraws a field, a needle or an icon instead of the whole screen can feel instant on a bus that manages 8 full frames per second. This is how instruments, thermostats and meters ship perfectly responsive SPI screens: they never redraw everything. Animation and video are the opposite case - every frame is a full frame, and the arithmetic wins.

Tearing, and the TE pin nobody wires

If a moving image shows a horizontal shear line, your writes are racing the panel's own refresh scan. Driver ICs expose a tearing-effect (TE) output exactly for this: it pulses when the scan is at a safe position, and synchronising writes to it removes the shear. On modules where TE is brought out to the connector it costs one GPIO; check the pinout before concluding you need a faster interface just to fix tearing.

The switch point: when SPI is the wrong answer

Full-screen animation, camera or video content, panels above roughly 3.5", or resolutions past 320x240 at real frame rates - that is RGB or MIPI territory, where the host drives pixels continuously instead of pushing changes down a serial pipe. The trade is pins and a host with a display controller, covered in our interface guide. The mix across our published models shows where each interface actually lives:

Counted across the 103 models on this site that carry a full parameter set; generic “customisable” listings are excluded so they cannot inflate a count. A model that offers several interfaces is counted in each family.

Deciding early costs nothing

The expensive version of this discovery happens after the carrier board is laid out, when the fix is a respin. If your UI plan includes animation, plan the interface for it now - several glass sizes exist in both SPI and RGB versions, and telling us the frame-rate expectation with the size gets you a module that will still be the right one at production volume.