How Can LCD Light Output Be Increased for Outdoor Visibility?

[Jonpro03]

Hello,

My name is Jon Proietti and I am an intern at a digital signage company. We've been looking into the possibility of using full color LCD displays outdoors. We want to increase the light output through the LCD so that it will be more easily visible in direct sunlight.

We purchased a large LCD television to play around with. Here is my understanding of how it's set up; please correct me if I am wrong.

At the very back is a series of white LEDs for the back light. 2" in front of them is a diffuser followed by two sheets of what I would call prism sheets. Their function seems to be directing the diffused light directly forward. There is a half inch of space before the LCD module, which I assume has the polarizer(s?) built in.

The original backlight put out roughly 6000 nits (cd/m2) and an all white message on the display allowed 280 nits through. Roughly 5%.

We built a jig and replaced the backlight with our own design, capable of 16000 nits. But somewhere in the process I lost a lot of throughput.

INPUT   OUTPUT
NITS     NITS

1694	28.6    1.69%
3442	57.8    1.68%
5131	81.63	1.59%
6780	106.8	1.58%
8390	131.2	1.56%
11040	165.3	1.50%
12320	188.3	1.53%
14070	212.1	1.51%
15250	236.5	1.55%
16750	259.4	1.55%

I'm not sure where I went wrong. Maybe it's that the jig is not enclosed and light is escaping. I know that some light is bouncing off of the diffuser. Maybe I need to enclose the jig with some reflective white surface to reflect the light back?

I've also considered that maybe I couldn't get an accurate reading of the original backlight LEDs. They are spaced about 3" apart, which makes getting a good reading with my nits gun hard to do (LS-100).

How much light can I expect to go through an LCD module?

 

[M Quack]

There are several steps in the process, and at each one you will loose light.

From the source trought the polarizer you will loose 50% unless your light source is naturally polarized - without taking diffusers etc into account that are there to make the back lighting homogeneous and "flat".

Then there is the liquid crystal itself. Ideally it would be able to rotate the polarization by 90 deg. In reality this is bound to be much less. You might gain here with a thicker LC (higher electric field for switching required) or even several LC layers.

If this rotation angle is called theta, then the transmission through the second polarizer should then be sin^2(theta) = 100% for theta=90 deg, but only 50% for theta=45 deg,
25% for theta=30 deg and so on.

Finally, remember that the white message is composed of coloured pixels. Assume that each color filter transmits 1/3 of the spectrum (I am afraid that this is generous!).

50% for unpolarized light source
25% for theta=30 deg rotating angle
33% for color filter

makes a bit more than 4% in total.

If I were you, I'd check if the original LEDs are polarized, and if your nit-gun is sensitive to polarization. A simple photo filter (check both directions through the filter to also test circular polarization!) or polarized sunglasses will do the trick.

The order of the LC will probably depend on the temperature. That may explain why the transmission drops at large power. That is something to check, too.

 

[Jonpro03]

I read online that I can expect about 8%. That would be fantastic if I could hit that, though I feel that number may be too generous as you described.

I have a theory about our significant losses.

The backlight we are using is composed of red green and blue LEDs. I have a feeling that the LCD color filter has a slightly different wavelength than the LEDs we are using. Therefore the peak wavelength of the LED is being blocked by the filter.

I plan on using a spectroradiometer tomorrow to measure a pure red message on the LCD using the original backlight, and so on for green and blue. That should give me a good understanding of what wavelengths the filter is, and what wavelength I should use for my backlighting.

 

[Jonpro03]

So my hypothesis was correct.

The LEDs we're using have peak wavelengths of:

R: 634nm
G: 523nm
B: 464nm

While the color filter has peak wavelengths of:

R: 611nm
G: 544nm
B: 448nm

Here is the Red:

The right border is set to 634nm, the peak of the LEDs we've been using. You can see that about 50% attenuation can be expected.



Green:

Here the left border is set to 523nm, with attenuation at about 68%.



Blue:

The right border is set to 468nm, attenuation about 28%. OUCH.

So I have some work to do to either find some LEDs with the right wavelength, or an LCD module with a broad spectrum color filter.