The problem on inspect LEDs on a large PCBA.

[yangyang-rexxar]

Hi all

I have a big PCBA whose dimension is 44cm X 28cm, and about 50 LEDs intersperse on it. I have a 2046 X 1536 pixel camera to do that.

The problem is, if I use a 12mm focal length to capture it, the object length should be more than 50cm, so the details on the board is not so clear such as the PCBA label(I am not so sure if the resolution is important for LED color and intensity measurement), but if I use a short focal length lens, the distances and point of view from lens to every LED will have a big difference, and the distance and view point will take effect on intensity measurement, so the LEDs in the center of board will be more light and on the edge ones will be dim.

So, any solution for this point, can I use formula to compensate it? Or probably I need to use several cameras?

[BruceAmmons]

Using a longer focal length and placing the camera a large distance from the PCBA will reduce your distortion.  It shouldn't affect things like detail clarity as long as it is focused.  The main problem at long distances is vibration, which can blur the image in extreme cases.

Bruce

[yangyang-rexxar]

Hi Bruce

From your idea the object length is more than 100cm in my application, this made the difference for every led distance less, in this case, do I still need a formular to compensate the distance variation, I am not so sure whether the detail clarify of picture take large effect on the intensity measure. My target is to find out the dim ones in several dozens of LEDs and to check if color is correct. Do I need more than one cameras?

In previous system they use a kind of sensor which is called SmartFin for every led, but the test result is not so good.

[BruceAmmons]

Always position the PCBA so it fills the field of view of the camera.  With a shorter focal length, it will be closer.  With a longer focal length, it will be further away.

From the size of your PCBA and the resolution of your camera, if you position the PCBA so it fills the field of view of the camera you should get a resolution around 0.2 mm/pixel.  This seems quite adequate to measure your LEDs, unless they are very tiny LEDs.  I don't see any reason to use multiple cameras.

Bruce

[yangyang-rexxar]

Hi Bruce

Now I have positioned my camera at 50cm far from PCBA and I can capture half of the PCBA, so plan to use 2 cameras, the camera is 1037*779 pixel camera. It is almost 4 pixels/mm, it seems enough to capture color and intensity. But the LED color display as a round, and at the center the RGB are all 255, so how can I determine the color of this round pane, how to calculate the intensity of this round pane. The intensity at center is 255. If I turn the apeture too small, I can just capture a small point. So could you have some advice on this.

[BruceAmmons]

That's the down side of trying to measure the color of LEDs.  They are so bright that they wash out the color.  I was thinking you were just looking at presence/absence and size.

The only method I can think of is using a series of color filters that won't let through the other colors.  That is a cumbersome method, though.

Bruce

[DOC-doc]

Hi PaulBin, Hi Bruce

I am not a great scientist, but reading what you want to do, I would say that if you want to measure the intensity of the leds, you have to take into consideration the fact that (for most LEDs) the light intensity is not constant with the viewing angle : if you look at the LED from the Top (zero angle) there is more light than if you look at it from the side. Here it depends a lot of the LED kind, but you can fine these figures at the LEDs manufacturers website.

Concerning the saturation of the LED :
1 - make a nonlinear LUT. But this will decrease the brightness measurment sensitivity
2 - make a second picture with smaller aperture time
3 - If you can, reduce the light intensity in the LEDs or flash them with a very short pulse.

Cheers

[yangyang-rexxar]

Hi Doc

I will follow what you said tomorrow moning. But seems I can just implement item 2. 🙂

I understand the angle problem so I want the camera so far from object to reduce the impact, now I prepare to use 2 or more camera to do this.

This afternoon I take some photos about my PCBA with different aperture, I just use one camera to take a small part of the whole board, it seems the intensity and color are captured perfectly, the aperture just take effect the dot's size, but whether I adjust the aperture, long or short, at the center of the led, the RGB are all near to 255, I just can distinguish the color from the edge.

Could you tell me in this case how to calculte the color and intensity generally?

[DOC-doc]

Intensity : I = (R value + G value + B value) / 3.

Color : the balance between R/I, G/I and B/I, where I is the intensity and R, G and B the Red Green and Blue value

Cheers

Again : be carefull. The Intensity you will measure can vary a LOT with the angle from which it is seen. For some leds a 10 or 15 degree value is enought for a 30 to 40% intensity variation

[yangyang-rexxar]

Hi Doc

The LED is 1.6x1.5mm BI-COLOR SMD CHIP LED LAMP. It is used as a status indicator on a PCBA. I take some photos about it with different aperture for reference. Which one is the best for analysis?