One of the most important characteristics of liquid-crystal displays (LCDs) is pixel response time. It is the time interval between the change of the signal in a particular cell and the change of the state of this cell. This parameter is also specified for classic cathode-ray tube displays and gas panels, but it is less crucial for them, as their response time is measured in microseconds. As for LCDs, the response time in them can reach tens of milliseconds, which may be noticeable for the human eye.
Despite some evident visual effects, such as the fuzziness of moving objects, response time cannot be more or less accurately estimated by the eye. One of the reasons is the inertia of the human eye itself (thanks to that, we don’t see the flicker of ordinary CRT displays, although every pixel of the screen glows only about 10% of the time).
In fact the problem of measuring response time looks simple at first sight: you could measure the time between the dispatch of the signal for cell state change and the actual change, tracking the pixel glow with the help of a photodiode. But it’s not that simple. The point is that you can’t pinpoint the time when signal is issued without meddling with the electronics of your display.
Another way: we could follow the vertical synchronization impulse and calculate at what time after it the data for our pixel are transferred, and adjust this time by the line synchronization impulse for the line containing that pixel… But such arithmetic only makes sense when you know exactly all time parameters of the scanning and assume that the display’s electronics doesn’t bring about any delays. The phenomena that can only be hypothesized upon, but not measured, don’t increase the reliability of the results. That’s why we put this method off, too.
After some brainstorming we chose another way of measuring pixel response time: we only need to draw the graph of pixel luminosity with a photodiode and oscilloscope and measure the time between the beginning and end of a luminosity change. There appears one question – how exactly should we measure response time for a display? Unfortunately, most manufacturers of displays and matrices don’t specify the measurement conditions and tell just the final result. Only technical documentation on certain TFT-LCD panels (not the off-the-shelf displays based on them) contains a description of the methods used to measure their response time parameter. That’s why we chose to stick to the ISO 13406-2 standard, which tells that LCD pixel response time is the total time it takes to turn a pixel on and off. Moreover, not the full time is taken, but the time from 10% to 90% of the total pixel luminance and then back to 10% (the graph is taken from the specs of Samsung Electronics’ LTM170E4 panel):
You don’t even have to watch a single pixel: it’s quite possible to light up the whole line to increase the input signal for the photosensor. In theory, as pixels light up one by one, but not simultaneously, this will blur the signal front, but such a measurement error is too tiny and can be neglected in our case. At 1024x768 resolution and 60Hz frame scan, pixels are output onto the screen at about 50MHz frequency. For example, when ten pixels come into the sensor’s window, the delay between the first and the last will be only 0.2 microseconds. It is too small considering we deal with time intervals of several milliseconds and higher.