An LED matrix can look perfectly acceptable on a workbench and still flicker on camera, show uneven brightness, ghost between rows, or lose detail when brightness is reduced. Three parameters explain much of that behavior: refresh rate, scan ratio, and LED current. They are connected, so changing one without understanding the others often creates a new problem.
This guide explains how those parameters interact in multiplexed LED displays and gives you a measurement-first workflow for improving display quality. If you are still choosing the drive architecture, first read our MAX7219 vs HUB75 vs direct multiplexing comparison.
Refresh rate is how often the complete LED display is scanned
In a multiplexed matrix, all LEDs are not continuously driven at the same time. The controller activates one row or group of rows, drives the required column data, blanks the outputs during the transition if necessary, and then advances to the next group. After every scan group has been serviced, one complete display refresh has occurred.
For LED signage, this refresh rate can be much higher than the video frame rate. Texas Instruments notes that LED-display refresh rates commonly reach the kilohertz range because low refresh rates can produce dark or uneven scan lines when a camera captures the display. That is why a display that looks stable to your eyes may still show bands in a phone video.
A useful simplified relationship is:
refresh rate ≈ row-scan frequency ÷ number of scan groups
The exact relationship depends on the driver architecture, PWM method, blanking intervals, data-loading time, and whether grayscale data is shifted while another row is active.
What scan ratio actually means
A 1/8-scan display divides its pixels into eight time-multiplexed groups; a 1/16-scan display uses sixteen groups; a 1/32-scan display uses thirty-two. A higher multiplex ratio reduces the amount of drive hardware required for a given number of LEDs, but each group receives a smaller share of the available cycle.
This is an engineering trade-off, not a simple “higher is better” setting. TI’s TLC5958, for example, is specifically designed for multiplexed displays and supports duty ratios up to 1/32. TI also uses it in a 64×64 RGB reference design with 1/32 time multiplexing. Dedicated display drivers include memory, timing hardware, constant-current outputs, and other features because maintaining image quality at high multiplex ratios is much harder than toggling GPIO pins in a loop.
Peak LED current is not the same as average LED current
With multiplexing, an LED is active only during its assigned scan interval. That means you must distinguish the instantaneous current while the LED is on from its time-averaged current. PWM dimming adds another duty-cycle term.
For a simple system, the average LED current can be approximated as:
IAVG ≈ IPEAK × scan duty × PWM duty
Do not use that equation as permission to multiply LED current blindly. The safe peak current depends on the LED’s pulsed-current ratings, pulse duration, thermal conditions, driver limits, wiring, PCB copper, power supply, and manufacturer recommendations. A display driver may also regulate current differently from a resistor-driven GPIO design.
Dedicated constant-current drivers are preferable when brightness uniformity matters. The MAX7219, for example, includes multiplex scanning, segment/digit drivers, digital and analog brightness control, and an external resistor used to set segment current. That is fundamentally different from driving a large matrix directly from MCU GPIO.
Why increasing the scan ratio can hurt image quality
Suppose the time available for one complete display cycle stays fixed. If you increase the number of scan groups, each group gets less time. If instead you keep each group’s on-time unchanged, the complete refresh takes longer and the refresh rate falls. Real display controllers balance these constraints with faster clocks, dedicated grayscale engines, internal memory, optimized blanking, and parallel outputs.
High multiplex ratios can expose four practical problems:
- Visible or camera flicker: the complete refresh frequency is too low for the viewing or recording conditions.
- Brightness loss: each LED has less effective on-time and the drive current or PWM settings do not compensate safely.
- Ghosting: row selection or column data changes while outputs are still enabled.
- Uneven grayscale: data-transfer and PWM timing consume too much of the available scan interval.
Refresh rate and frame rate are different
The frame rate describes how often the displayed image data changes. The LED refresh rate describes how often the multiplexing system completes the physical scan needed to reproduce that image. A static image can therefore have a frame rate near zero while the panel continues refreshing its LEDs thousands of times per second.
Confusing these terms leads to bad debugging. If text is static but a camera shows horizontal bands, increasing how often your application sends the same frame may accomplish nothing. The problem can be the panel scan/PWM timing rather than the application frame rate.
A practical measurement workflow
- Measure a row-select signal. Use an oscilloscope or logic analyzer to determine the row dwell time and complete scan period.
- Calculate the real refresh rate. Do not rely only on a library setting. Measure from one activation of a row to its next activation.
- Check blanking. Verify that outputs are disabled while row/address signals and column data transition. Poor sequencing is a common source of ghost pixels.
- Measure current correctly. Determine peak LED/channel current and estimate or measure average supply current. Confirm both against the driver, LED, connector, PCB, and supply limits.
- Test multiple brightness levels. Low PWM duty cycles can reveal timing limitations that are hidden at full brightness.
- Test with a camera. Record at the actual shutter/frame-rate conditions relevant to the application. Camera banding is often a stricter test than visual inspection.
- Run a worst-case pattern. Full white, checkerboards, single-pixel patterns, and rapidly changing content expose different current and timing failures.
If you need to inspect the bus and timing signals themselves, see how to debug UART, I2C, SPI, and digital timing with a logic analyzer.
Do not size the power supply from “average-looking” content
A display showing a few characters may draw far less current than a full-brightness test pattern. Design the power path for the credible worst case: supply capacity, connector current, trace width, voltage drop, driver dissipation, and local decoupling all matter.
Voltage drop can also create a misleading brightness problem. If pixels far from the supply entry point are dimmer, measure the panel voltage under a heavy pattern before changing PWM or current settings. Increasing LED current to compensate for a weak power-distribution network can make the electrical and thermal problem worse.
How to choose a better configuration
| Symptom | Measure first | Likely area |
|---|---|---|
| Looks fine to eye, bands on camera | Complete scan/refresh period | Refresh rate, PWM timing |
| Dim at high scan ratio | Peak current and effective duty cycle | Current setting, scan ratio |
| Ghost pixels between rows | OE/blanking versus row/data transitions | Sequencing and settling time |
| Brightness varies across panel | Supply voltage at multiple panel points | Power distribution, current regulation |
| Low brightness loses grayscale quality | PWM pulse width and timing | Driver/PWM resolution |
What a professional-looking LED display requires
There is no universal refresh-rate or current value that makes every panel “professional.” The correct values depend on the panel, LED and driver datasheets, scan architecture, PWM depth, viewing environment, camera requirements, thermal design, and power system.
The reliable approach is to treat refresh rate, scan ratio, PWM, and current as one timing-and-power system. Measure the scan waveform, confirm the current limits from primary documentation, verify blanking and power integrity, then test the display with both worst-case patterns and the cameras that will actually be used. That produces a defensible design instead of a display that only happens to look good on the bench.
Primary technical references
- Texas Instruments, Common LED Display Challenges, Root Causes and Workarounds in Narrow Pixel Pitch Matrix LED Display (SLVAFM7).
- Texas Instruments, TLC5958 48-channel constant-current LED driver documentation and TIDA-00161 64×64 RGB LED matrix reference design.
- Analog Devices, MAX7219/MAX7221 serially interfaced LED display driver datasheet.