Looking for Templates for an Analog Circuit to Control LEDs from a Photo Diode

Discussion in 'Analog & Mixed-Signal Design' started by gebseng, Nov 24, 2017.

  1. gebseng

    Thread Starter New Member

    Nov 24, 2017
    27
    1
    Hello everyone on the forum,

    I am trying to build an electronic circuit that can drive a bright array (size appr. 40 by 60 mm) of white LEDs from a photo diode. By that I mean that the LED array should be able to change its brightness according to the brightness induced to the photo diode. This should work with a maximum frequency of 2 MHz. The circuit should also include a potentiometer to gamma correct the signal.

    I need this for a mechanical television system I am building, as described here: http://gebseng.com/11_big_paul/big_paul_brochure.pdf

    Since this system (the spinning Nipkow Disk) blocks out 99.99 percent (sic!) of the available light, the photodiode needs to be very light sensitive, and the LED array needs to be very bright.

    As you can probably tell from this writeup, I am not an electronic engineer myself, but I have decent soldering skills and a basic understanding of analog electronic parts.

    If you could suggest a tutorial, or even better existing examples for similar circuits that I could modify for my purpose, this would be very helpful. I am not sure what the right terms for this kind of circuit are, but I guess it can't be too exotic.

    best,

    geb



     
  2. RichardO

    Late Member

    May 4, 2013
    2,273
    894
    Thanks for the link to the historical information. :D:D

    You have a very challenging project. The best kind. :)

    Some quick questions:

    Which form of scanning are you using for the camera?

    Are you flooding the scene with very bright and placing the photodiode behind the disc? Or, are you doing a flying spot scanner with the light going through the disc and viewing the entire scene with the photodiode?

    I assume that the LED array is used in the viewer instead of illuminating the scene. Is this right?

    Can you tell us more abut the LED array. Mainly what voltage and current for operation.

    What optics do you have in mind for the camera and viewer?

    How do you plan on synchronizing the camera and viewer?
     
  3. gebseng

    Thread Starter New Member

    Nov 24, 2017
    27
    1
    Thanks for your reply!

    As for the camera part, the idea is to place a bright and continuous spotlight behind the disk, which, through the rotating disk, projects a line pattern on the subject to be scanned. On the subject side, a photo diode registers changes in the light reflected off the subject. So, flying spot I guess.

    Yes, the LED array is on the viewer side.

    I don't have voltage/current specifications for the LEDs yet, only that they will have to cover an area of 40 x 60 mm, and that they will have to be very bright, as the combination of the very small holes (0.25 mm or 0.5 mm diameter) and the fast rotating disk (900 rpm) will only let through around 0.01 percent of the original light.

    I am aware of the need to synchronize, but in a first step will only build one apparatus, which will have the camera on the left, and the viewer on the right side.

    best,

    Gebhard
     
  4. gebseng

    Thread Starter New Member

    Nov 24, 2017
    27
    1
    About the LED type: I am not sure if I can pick LEDs designed for lighting purposes. Are they dimmable like smaller LEDs, and can they be modulated at up to 2 MHz?

    If yes, how about this one:
    https://uk.rs-online.com/web/p/visible-leds/8792889/

    I would arrange 54 of those in a 6 x 9 grid.

    Specifications:
    LED Colours White
    Mounting Type Surface Mount
    Number of LEDs 1
    Forward Voltage 38 V
    Luminous Intensity 461.5 → 547.5 cd
    Number of Pins 2
    Viewing Angle 120 °
    Luminous Flux 1400 → 1800 lm
    Lens Shape Square
    Dimensions 7 x 7 x 0.7mm
    Series DURIS S10
    Colour Temperature 5700K
    Lens Colour Yellow
    Forward Current 20 → 400mA
    Lens Finish Diffused
     
  5. gebseng

    Thread Starter New Member

    Nov 24, 2017
    27
    1
    OK sorry, I guess this would be a total of 810 W, which is probably too much ;)
     
  6. RichardO

    Late Member

    May 4, 2013
    2,273
    894
    The LED you linked to is not mounted on a circuit board. That will be a problem. A star mounted LED solves 2 issues. First, it has the hard soldering done for you and, second, the PCB is designed to mount to a heat sink. Heat sinking is essential. The LED will burn up very quickly without a proper heat sink. (Imagine how hot a 20 watt light bulb runs).

    Maybe something like one of these:
    https://www.digikey.com/products/en...830&page=1&quantity=0&ptm=0&fid=0&pageSize=25

    I think just one of these is probably bright enough. You would need a diffuser to get illumination that is even over the entire viewing area. Buying one of the matching reflectors may help get a flatter illumination, as well.
     
  7. RichardO

    Late Member

    May 4, 2013
    2,273
    894
    Got it.

    One way you can guess at the amount of light needed is to imagine how bright a light bulb of the equivalent lumens would be.

    You might try simulating the brightness. put a hole in a plate that is the same size as the one in your disc. Then view lamps of different lumens through the hole. I am not sure how this reflects the spot moving, however.

    That sound good.

    Just make sure that the LED does not have any resistors or other current limiting circuit in it.

    From what I know the LED's will be fast enough.
     
  8. gebseng

    Thread Starter New Member

    Nov 24, 2017
    27
    1
    Thank you so much, especially for narrowing down to LED options and suggesting the star mount!

    I will get started with this one:
    https://www.digikey.com/product-det...A-00-0000-0D0BN240E-SB01/1672-1050-ND/6012737
    which is star mount and has a good lm/W ratio. I think it would be better to start with a set of four (plus the diffusor), since the diffusor will take away more light, and 0.00114% of 6000 lumen is less than one lumen anyway.

    Now about the photo diode (assuming that a photo diode is the right pick at all), I picked this one:
    https://www.digikey.com/product-det...nics/SD003-151-001/SD003-151-001CT-ND/5222595
    because it had the lowest dark current.

    What do you think?

    best,

    geb
     
  9. Reloadron

    Distinguished Member

    Jan 15, 2015
    4,367
    2,129
    If I may interject something here and if my thinking is way off base please forgive me. Looking back at the first post:
    So my read here is you want a photo sensor to sense the ambient light and vary the LED intensity based on the sensed light. Should that be the case I am not sure a photo diode is the best choice for your light sensor. The reasoning is that a photo diode is generally the choice sensor for On/Off applications. My thinking is that a LDR (Light Dependent Resistor) may be a better choice as long as it has a fast enough response time. Here is what I am getting at: Difference between LDR and Photodiode.

    As to the actual control of your LED Intensity. You may wish to consider the use of PWM (Pulse Width Modulation). Using a small micro-controller like an Arduino for example the PWM can be controlled based on an input from a photo sensor. The lower the ambient light the higher the PWM duty cycle and the brighter the LED. This is a relatively simple circuit and the required code is very basic and readily available. Again, if I am off base with rhis thinking my apologies.

    Ron
     
  10. RichardO

    Late Member

    May 4, 2013
    2,273
    894
    [QUOTE="Reloadron, post: 1205155, member: 260269"So my read here is you want a photo sensor to sense the ambient light and vary the LED intensity based on the sensed light. Should that be the case I am not sure a photo diode is the best choice for your light sensor. The reasoning is that a photo diode is generally the choice sensor for On/Off applications. My thinking is that a LDR (Light Dependent Resistor) may be a better choice as long as it has a fast enough response time. Here is what I am getting at: Difference between LDR and Photodiode.

    As to the actual control of your LED Intensity. You may wish to consider the use of PWM (Pulse Width Modulation). Using a small micro-controller like an Arduino for example the PWM can be controlled based on an input from a photo sensor. The lower the ambient light the higher the PWM duty cycle and the brighter the LED. This is a relatively simple circuit and the required code is very basic and readily available. Again, if I am off base with rhis thinking my apologies.

    Ron
    [/QUOTE]
    I agree that he needsa photo-sensorwiththemostarea and therefore the most senesitivty as he can get. I do worry that any sensor other than a diode will not be fast enough. A bandwidth of 2 MHz was suggested. :eek:

    He can't do PWM because he needs such wide bandwidth. Well, maybe PWM would work if the PWM frequency was tens of MHz.
     
  11. RichardO

    Late Member

    May 4, 2013
    2,273
    894
    Here are some of my ramblings:

    I visualize a disc that is 1.6mm thick by 500mm diameter made from ABS plastic sheet. I would use 100 0.5mm diameter holes to give 100 scan lines.

    For testing, I would only have one disc. The top edge would be used for the viewer and the bottom edge for the camera. This might create an upside-down image, however.

    There must be a mask between the disc and the scene being scanned. This mask is next to the disc so that the scene is scanned by only one line at a time in a rectangular area.

    The camera light source must be a point source so that it can be focused to a spot on the scene being scanned. The pinhole in the disc will do some focusing but I think an additional lens will be needed. This lens must be bigger than the rectangular area of the disc doing the scanning.

    MR series halogen lamps might work for camera illumination .

    100 lines of 130 pixels at 900 RPM:
    900 RPM = 15 Hz ~= 60 ms. 60ms /100 lines 600us per line.
    600us /130 pixels ~= 5 us per pixel. 2 pixels per cycle = 10us = 100KHz.

    If my assumptions are correct then you only need about 100 KHz of bandwidth for your photo detector. This is a lot easier than 2 MHz. :)
     
  12. RichardO

    Late Member

    May 4, 2013
    2,273
    894
    I looked at your photo detector. You need a photo detector that better matches the human eye color response. Green is about 550 nm.

    upload_2017-11-26_13-47-59.png

    The LED looks like it will work. I am cheap -- I would try just one of them before springing for more. It could turn out that one is bright enough or you might find that it does not work at all for what you want.

    Don't forget to heat sink the LED. At 25 watts (for only _one_ LED) and a temperature rise of 50 deg. C you need a heat sink of less than 2 deg. C per watt. This is a big heat sink unless you add a fan. Four LED's would absolutely need a fan _and_ a large heat sink.
     
  13. Reloadron

    Distinguished Member

    Jan 15, 2015
    4,367
    2,129
    Ah, I think I better get it now. Well at least I can see where I was off base in my thinking as literally I was not seeing the whole picture. Thanks for helping me understand what was going on.

    Ron
     
  14. gebseng

    Thread Starter New Member

    Nov 24, 2017
    27
    1
    Thanks! As far as I know, LDRs have a way too slow response time (I read 30ms on one datasheet, is that 330 Hz?), but I need something that works at 2 MHz. So, a photo diode can only "switch" at a certain light threshold? Are there other light sensors that could work in he 2 MHz range and also be very light sensitive?

    best,

    geb
     
  15. gebseng

    Thread Starter New Member

    Nov 24, 2017
    27
    1
    Thanks! overlooked that this photo diode works in the infrared spectrum. But actually, this might prove useful, since I then could use an infrared light source for the camera, which would interfere less with ambient light. But I can also look for another diode.
     
  16. gebseng

    Thread Starter New Member

    Nov 24, 2017
    27
    1
    OK; I overlooked your response to Rons contribution first. Yes exactly, that's why I thought of a photo diode instead of an LDR.
     
  17. gebseng

    Thread Starter New Member

    Nov 24, 2017
    27
    1
    Thanks for your considerations! I am way past the stage of thinking about the disk, since that part is finished already.
    My disk has a diameter of 1490mm, is made from 6mm aluminum and at the moment has 120 holes, 5mm diameter each, that will be plugged with small inserts with a conical bore and a center hole of 0.5 mm (see attached image and pdf).

    My thinking with the frequency was: 120 vertical lines (= holes) by 100 lines (horizontal resolution) by 15 frames/second equals 180 kHz (?).
    But I want to have reserves for another version with 240 holes (0.25 mm hole) and a horizontal resolution of 200 lines, which in m thinking would equal 720 kHz. I "rounded that up" to 2 MHz to err on the safe side...

    Thanks also for your ideas about the camera construction. Yes, I might need shielding against ambient light, and might have to focus to scan beam. But maybe a spotlight source with parallel light beams will prove to be enough.

    W0RUDpDTSo2FAZSrq+9ykA.jpg
     
  18. gebseng

    Thread Starter New Member

    Nov 24, 2017
    27
    1
  19. gebseng

    Thread Starter New Member

    Nov 24, 2017
    27
    1
    How about this "Avalanche" type photodiode? It has an extremely low dark current and is fast enough:
    http://marktechopto.com/pdf/products/datasheet/MTAPD-06-001.pdf

    I read on wikipedia (https://en.wikipedia.org/wiki/Photodiode):
    - - - - - - - - - - - - - - - - - -
    Photoconductive mode
    In this mode the diode is often reverse biased (with the cathode driven positive with respect to the anode). This reduces the response time because the additional reverse bias increases the width of the depletion layer, which decreases the junction's capacitance. The reverse bias also increases the dark current without much change in the photocurrent. For a given spectral distribution, the photocurrent is linearly proportional to the illuminance (and to the irradiance).[3]

    Although this mode is faster, the photoconductive mode tends to exhibit more electronic noise.[4] The leakage current of a good PIN diode is so low (<1 nA) that the Johnson–Nyquist noise of the load resistance in a typical circuit often dominates.
    - - - - - - - - - - - - - - - - - -

    So, if the photocurrent is linearly proportional to the illuminance, couldn't that be used to build a circuit that measures this current and uses the result to drive the LEDs?
     
    Last edited: Nov 27, 2017
  20. RichardO

    Late Member

    May 4, 2013
    2,273
    894
    It is very hard to guess what things look like in the infrared spectrum. Using an IR photodiode might add confusion when you are trying to get yuor project working.
     
Loading...