![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
For those interested, I'd like to describe what I believe is a significant issue with backlights for current monitors, and what I've learned while investigating it:
Nearly all monitors rely on pulse-width-modulation(PWM) to accomplish backlight dimming. This means the backlight is cycling on and off at a fixed rate (typically about 200Hz), and the percentage of time the light is on for each cycle can be varied. At maximum brightness the light is continuously on, while at minimum brightness the light may only be on for 10% or less of each cycle.
While this makes it easy to support a wide range of brightness, using PWM is causing visible temporal artifacts to appear. Specifically, when glancing from one part of a display to another duplicate after-images of on-screen shapes can be seen, often in multiple colors. In some cases this effect is occurring to the point where I personally find it hard to do things like read text, and is why I am trying to educate myself and others about it. I have tried to describe the effect as concisely as possible below, and hope that perhaps more people may be motivated to look into this.
Symptoms:
When glancing from one area of the monitor to another, after-images in multiple colors appear around high contrast areas. This occurs with both static and moving images, but is highly dependent on the current monitor brightness setting. A brightness of 0 is worst, while the effect disappears by the time 100% is reached. After-images tend toward yellow or blue in color, with up to 8-10 being visible at once. Multiple colors appear to be seen only when using CCFL backlights and not with LED backlights, but the after-images are visible with both. Note that this effect has nothing to do with the LCD itself, the effect is entirely due to the backlight.
This effect is much more visible to some people than others. Most people I have asked can see the effect on newer monitors when fully dimmed, but not above 50% or so. I seem to be more sensitive than average and can see it at brightness levels as high as 70% on some displays. The main effect, other than annoyance at strange after-images, is that reading text is much more difficult and takes longer. White text on black appears to be the worst case, but black on white is significantly affected as well.
Example of what the effect looks like (Left-CCFL, Right-LED):
This is made much worse by newer monitors since backlight brightness is increasing, so they must be run at lower brightness percentages to remain at the same absolute brightness (and not burn your eyes out).
Cause:
The effect seems to be caused by using PWM to control the backlight brightness. When the backlights are dimmed to their lowest value the duty cycle, or time spent lit, may only be 10% or less. This is mitigated somewhat by the phosphors in a CCFL since they tend to remain glowing after the internal gas itself has gone dark, but is also causing part of the problem. During cycling the blue phosphors appear to both turn on faster and turn off faster than the green and red phosphors. This means that for every cycle the backlight is switching in the form off-blue-neutral-yellow-off, with the amount of time spent in neutral determined by the brightness value. For the lowest brightness on many new monitors there may be no time spent in neutral as it transitions from one color directly to the other. Note that the time-averaged color remains neutral.
The other part of the effect (after-images) seems to be caused by the fact that the backlight is cycling faster than the vision adaptation rate. In the time it takes to glance from one part of the monitor to another the light has undergone several cycles. When the light is using a short duty cycle this results in several discrete after-images being visible in the exact shape of objects on the screen. All monitors I have been able to test use cycling on the order of 170-220Hz.
Measuring the effect:
I have found two methods thus far for measuring the effects of backlight cycling:
1. Use a high-speed camera. I have taken video with a Casio EX-F1 at 1200fps and found that I can get 5 or 6 frames per backlight cycle. By overlaying a few seconds worth of pixel values from the video a reasonable approximation of the intensity of all three colors can be made for the full cycle.
2. Use a slow shutter speed on a moving camera. By displaying a thin and white vertical bar on a black background and panning across the display during exposure, the light from this thin bar is spread across the frame with time. As long as the camera moves at a relatively constant velocity during a cycle (easy because it is short), the intensity values can be seen as waves on the image.
Cycles measured with technique 1:
Cycle measured with technique 2:
Findings:
A. Both methods confirm what I am seeing is due to the backlighting and not my personal vision. I do appear to be more sensitive than average (based on asking others what they see), but by no means the only one affected.
B. The effect is more pronounced in newer monitors. Most likely due to their increased maximum brightness, which means we must use a shorter duty cycle to maintain a comfortable brightness. The phosphors in some new monitors may also have a shorter period during which they continue to glow, but I have not definitively confirmed this.
C. TVs are hard to measure because their backlight brightness is often auto-adjusted depending on the content being shown. This confused me for a while when I started taking measurements, but I have confirmed it.
What I am trying to do about it:
First, to inform others about what could be affecting them. This seems to be particularly troublesome for some newer laptops using LED backlighting (see the new Macbook pro (original version) or the Lenovo T400). Using a short duty cycle and a slow cycle rate (90Hz in some cases) is causing many people problems. There may be a lot of people out there that, like me, did not know what was, or is still, affecting them.
Second, to see if anyone currently makes or will make a display that shows neither the after-image nor the colored after-image effects. In the case of CCFL most of the effect could likely be mitigated by using cycling on the order of 500Hz or more, but for LEDs would likely need to be significantly faster because of their almost instant on-off.
Third, to buy a monitor I can use. Yes, the effect is bad enough that I would not want a new display with this problem, but I seem to be in the minority of having that opinion for the moment. I would personally rate the effect as worse than AG coating, but not as bad as TN brightness shifting.
How to fix the issue (none of which are currently available or very attractive):
1. Buy a monitor with faster backlight cycling. At a fast enough rate the after-images should no longer appear discrete and will blur together as if from a continuous source. For CCFLs this is probably somewhere around 1000Hz, and may be even higher for LEDs. As yet I do not know of anyone who sells such a monitor, though it is physically possible.
2. Buy a monitor with a dim backlight. I have my current display calibrated to 140 cd/m2 and never use it above this value. Many new displays have a max brightness of 400 cd/m2 or higher, which means to meet my current level they would have to be used at 30% brightness and would definitely exhibit problems. If a light could be bought with a max brightness of, say, 180 cd/m2 it would not need to be dimmed so far.
3. Replace the backlight driver circuit to handle faster cycling. Not easy, but possible for the electrically inclined (not me).
4. Replace the backlight with a dimmer light. Has lots of potential problems.
5. Apply a neutral density (ND) filter to the rear of the LCD, but in front of the backlight. This would allow the monitor to be run at full brightness to prevent artifacts, but keep it to a reasonable brightness level. Of course this would be an incredible waste of power, but it's also probably the easiest method. Maybe a polarizer could be added to the back to act like an A-TW filter also? Anyone know more about this?
Nearly all monitors rely on pulse-width-modulation(PWM) to accomplish backlight dimming. This means the backlight is cycling on and off at a fixed rate (typically about 200Hz), and the percentage of time the light is on for each cycle can be varied. At maximum brightness the light is continuously on, while at minimum brightness the light may only be on for 10% or less of each cycle.
While this makes it easy to support a wide range of brightness, using PWM is causing visible temporal artifacts to appear. Specifically, when glancing from one part of a display to another duplicate after-images of on-screen shapes can be seen, often in multiple colors. In some cases this effect is occurring to the point where I personally find it hard to do things like read text, and is why I am trying to educate myself and others about it. I have tried to describe the effect as concisely as possible below, and hope that perhaps more people may be motivated to look into this.
Symptoms:
When glancing from one area of the monitor to another, after-images in multiple colors appear around high contrast areas. This occurs with both static and moving images, but is highly dependent on the current monitor brightness setting. A brightness of 0 is worst, while the effect disappears by the time 100% is reached. After-images tend toward yellow or blue in color, with up to 8-10 being visible at once. Multiple colors appear to be seen only when using CCFL backlights and not with LED backlights, but the after-images are visible with both. Note that this effect has nothing to do with the LCD itself, the effect is entirely due to the backlight.
This effect is much more visible to some people than others. Most people I have asked can see the effect on newer monitors when fully dimmed, but not above 50% or so. I seem to be more sensitive than average and can see it at brightness levels as high as 70% on some displays. The main effect, other than annoyance at strange after-images, is that reading text is much more difficult and takes longer. White text on black appears to be the worst case, but black on white is significantly affected as well.
Example of what the effect looks like (Left-CCFL, Right-LED):
This is made much worse by newer monitors since backlight brightness is increasing, so they must be run at lower brightness percentages to remain at the same absolute brightness (and not burn your eyes out).
Cause:
The effect seems to be caused by using PWM to control the backlight brightness. When the backlights are dimmed to their lowest value the duty cycle, or time spent lit, may only be 10% or less. This is mitigated somewhat by the phosphors in a CCFL since they tend to remain glowing after the internal gas itself has gone dark, but is also causing part of the problem. During cycling the blue phosphors appear to both turn on faster and turn off faster than the green and red phosphors. This means that for every cycle the backlight is switching in the form off-blue-neutral-yellow-off, with the amount of time spent in neutral determined by the brightness value. For the lowest brightness on many new monitors there may be no time spent in neutral as it transitions from one color directly to the other. Note that the time-averaged color remains neutral.
The other part of the effect (after-images) seems to be caused by the fact that the backlight is cycling faster than the vision adaptation rate. In the time it takes to glance from one part of the monitor to another the light has undergone several cycles. When the light is using a short duty cycle this results in several discrete after-images being visible in the exact shape of objects on the screen. All monitors I have been able to test use cycling on the order of 170-220Hz.
Measuring the effect:
I have found two methods thus far for measuring the effects of backlight cycling:
1. Use a high-speed camera. I have taken video with a Casio EX-F1 at 1200fps and found that I can get 5 or 6 frames per backlight cycle. By overlaying a few seconds worth of pixel values from the video a reasonable approximation of the intensity of all three colors can be made for the full cycle.
2. Use a slow shutter speed on a moving camera. By displaying a thin and white vertical bar on a black background and panning across the display during exposure, the light from this thin bar is spread across the frame with time. As long as the camera moves at a relatively constant velocity during a cycle (easy because it is short), the intensity values can be seen as waves on the image.
Cycles measured with technique 1:
Cycle measured with technique 2:
Findings:
A. Both methods confirm what I am seeing is due to the backlighting and not my personal vision. I do appear to be more sensitive than average (based on asking others what they see), but by no means the only one affected.
B. The effect is more pronounced in newer monitors. Most likely due to their increased maximum brightness, which means we must use a shorter duty cycle to maintain a comfortable brightness. The phosphors in some new monitors may also have a shorter period during which they continue to glow, but I have not definitively confirmed this.
C. TVs are hard to measure because their backlight brightness is often auto-adjusted depending on the content being shown. This confused me for a while when I started taking measurements, but I have confirmed it.
What I am trying to do about it:
First, to inform others about what could be affecting them. This seems to be particularly troublesome for some newer laptops using LED backlighting (see the new Macbook pro (original version) or the Lenovo T400). Using a short duty cycle and a slow cycle rate (90Hz in some cases) is causing many people problems. There may be a lot of people out there that, like me, did not know what was, or is still, affecting them.
Second, to see if anyone currently makes or will make a display that shows neither the after-image nor the colored after-image effects. In the case of CCFL most of the effect could likely be mitigated by using cycling on the order of 500Hz or more, but for LEDs would likely need to be significantly faster because of their almost instant on-off.
Third, to buy a monitor I can use. Yes, the effect is bad enough that I would not want a new display with this problem, but I seem to be in the minority of having that opinion for the moment. I would personally rate the effect as worse than AG coating, but not as bad as TN brightness shifting.
How to fix the issue (none of which are currently available or very attractive):
1. Buy a monitor with faster backlight cycling. At a fast enough rate the after-images should no longer appear discrete and will blur together as if from a continuous source. For CCFLs this is probably somewhere around 1000Hz, and may be even higher for LEDs. As yet I do not know of anyone who sells such a monitor, though it is physically possible.
2. Buy a monitor with a dim backlight. I have my current display calibrated to 140 cd/m2 and never use it above this value. Many new displays have a max brightness of 400 cd/m2 or higher, which means to meet my current level they would have to be used at 30% brightness and would definitely exhibit problems. If a light could be bought with a max brightness of, say, 180 cd/m2 it would not need to be dimmed so far.
3. Replace the backlight driver circuit to handle faster cycling. Not easy, but possible for the electrically inclined (not me).
4. Replace the backlight with a dimmer light. Has lots of potential problems.
5. Apply a neutral density (ND) filter to the rear of the LCD, but in front of the backlight. This would allow the monitor to be run at full brightness to prevent artifacts, but keep it to a reasonable brightness level. Of course this would be an incredible waste of power, but it's also probably the easiest method. Maybe a polarizer could be added to the back to act like an A-TW filter also? Anyone know more about this?
