A Year-Long Experiment In OLED Burn-In

If you need to add a small display to your project, you’re not going to do much better than a tiny OLED display. These tiny display are black and white, usually found in resolutions of 128×64 or some other divisible-by-two value, they’re driven over I2C, the libraries are readily available, and they’re cheap. You can’t do much better for displaying a few numbers and text than an I2C OLED. There’s a problem, though: OLEDs burn out, or burn in, depending on how you define it. What’s the lifetime of these OLEDs? That’s exactly what [Electronics In Focus] is testing (YouTube, in Russian, so click the closed captioning button).

The experimental setup for this is eleven OLED displays with 128×64 pixels with an SSD1306 controller, all driven by an STM32 over I2C. Everything’s on a breadboard, and the actual display is sixteen blocks, each lit one after another with a one-second display in between. This is to test gradually increasing levels of burnout, and from a surface-level analysis, this is a pretty good way to see if OLED pixels burn out.

After 378 days of testing, this test was stopped after there were no failed displays. This comes with a caveat: after a year of endurance testing, there were a few burnt out pixels. correlating with how often these pixels were on. The solution to this problem would be to occasionally ‘jiggle’ the displayed text around the screen, turn the display off when no one is looking at it, or alternatively write a screen saver for OLEDs. That last bit has already been done, and here are the flying toasters to prove it. This is an interesting experiment, and although that weird project you’re working on probably won’t ping an OLED for a year of continuous operation, it’s still something to think about. Video below.

22 thoughts on “A Year-Long Experiment In OLED Burn-In

  1. Very nice test! I would leave one display totally unconnected as sample and then test single pixel luminosity levels with any equipment other than eye to compare relative levels (just for % of wear). But for a first such test this one is really good!

  2. “If you need to add a small display to your project, you’re not going to do much better than a tiny OLED display”. Actually there are plenty of options. My favorite is the 1.44″ 128×128 displays.
    I think the two reasons why the OLEDs are popular is that they are monochrome (which requires way less resources and fiddling with colors and fonts) and the I2C interface which uses only 2 pins, the other ones being generally SPI.

    On another hand, this test is done very well and has answered the endurance question that most of us had about the tiny OLEDS.

  3. We have these kinds of displays running for 5 years in our Ultimaker2 printer now. So I have a lot of examples of pixels burning out. I’ve yet to see a pixel go fully dark, but they lose brightness over time as this experiment also shows. In our case, it’s generally the “Remove print” text that is visible on fully filled screens.

    Next to a screen saver, lowering the brightness of the screen also greatly improves the overall lifetime of the pixels, without impacting the information flow. In the Ultimaker3 we implemented that the screen goes to minimal brightness after 30 seconds, and goes back to full brightness on user interaction. Most people hardly notice this, and I haven’t seen any screen showing signs of wear in that.

  4. That’s almost the same test that OLED manufacturers perform. The standard is a checkerboard pattern and they test the brightness every hundred hours or so to find the half-life of the panel. If you look up the datasheets, they’ll have a number in the 10000-40000 hours range. That’s where the panel has a brightness drop of 50%. As soon as you light it up, it starts breaking down.

    The screen saver is just a wear leveling technique, but it helps if you’ve got some UI elements that come and go so that you’d see the ghost of another icon or something. Avoid stuff like bounding boxes if you can. To extend the real lifetime, you have to reduce the brightness or turn it off.

    1. Another trick you can implement with monochrome displays is to invert them periodically. I was set to use some of these OLEDs like this for home information, but gave up because they are too small. However, with a light sensor I would toggle day/night mode. In day I keep the background white and text black, on night the reverse.
      This is why these small OLEDs are useful for devices that hardly use the display or are not in use all the time.

    2. “Avoid stuff like bounding boxes”

      I did a web search on “bounding boxes”, but I am not sure which definition applies to this.
      Would you please elaborate a bit for me?

      1. In the video, you see a grid of empty boxes being filled up gradually. The grid itself, a empty rectangle around each large square lit area is a “bounding box” as its on 100 percent of the time. Basically a box around text/graphics thats on. The “Border”.

          1. Just for general information it comes from boundary, a line that marks the limit of an area. So the bounding box is the box which shows the boundary.

    1. Yes, regular LED’s wear out too, especially if they’re constantly driven close to (or exactly at/over) manufacturer recommended current level.
      Like other electronic components, you can prolong their lifetime by overspeccing or cooling them.

    2. The wear out mechanism is likely a chemical process towards entropy (especially as organic materials tend to break down faster than inorganic ones like those used in regular LEDs). So unfortunately the best you can do is try to stretch out mtbf in use or manufacture but you’ll only ever delay the inevitable.

    3. I have an ASUS G1 laptop (circa 2006) that has a small OLED above the keyboard.

      All the pixels that display the time on that display are so dim you can barely see them in the dark.

      The other pixels are super bright on startup, when the display shows the ASUS logo when booting.

      They definitely do get dimmer over time when used.

    4. We have an iHome OLED clock/radio with an Apple charging port (~8 years old).
      The clock digits fade to unreadable every year or so, and unplugging the clock for several days partially refreshes them.

  5. “and although that weird project you’re working on probably won’t ping an OLED for a year of continuous operation”

    I do actually have one project where the OLED-display is on 24/7, namely a LoRa-to-WiFi receiver I made with an ESP32. One of the two cores handles receiving and transmitting LoRa- and WiFi-stuff, the other one asynchronously draws useful details on the display whenever it receives a LoRa-packet for easy diagnostics. It also draws one icon for WiFi-connectivity and another for MQTT-connectivity, so if one of the two icons is missing, I’ll know why data isn’t moving where it’s supposed to.

    Now, I could write some sort of a screensaver for the OLED or move things about at some intervals, but..ehh, in a project like this, I just don’t care about burn-in as long as I can see what’s on it.

    1. Same technology for oled TV’s and for oled Phones. Some manufacturers claim to have better durability than others, but a quick internet search will pull up pictures people posted of their expensive oled TV or Phone with screen burn in. For color displays it’s a particular challenge because the colors each age at slightly different rates.

  6. and yet … because of oled displays few otherwise good dive computers went to sh*t …maybe these are just not best application for oled … (xdeep – but ok they do still service and replace them for free when they burn out, liquivision – becomes junk)

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.