Sunday, 27 January 2019

LED backlights for birdbox illumination

All my birdboxes have Infrared (IR) and white light illumination, using 5mm LEDs.  The IR ones that I use give nice diffuse lighting, however the  white light ones tend to give a 'spotlight' effect.  I spotted these LED backlights recently from Pimoroni and once installed, they give a nice diffuse 'daylight' effect:

Underside of roof camera insert.  The circular black thing is an IR-cut module.

This year's new small bird boxes are equipped with two each.  They are wired in parallel to 5V with a 50 ohm resistor for each one, as well as another 50/100/200 ohm resistors between the ground and the uln2003 i/c transistor that is used to switch them for low/med/and blinding light levels.  Default is medium.

This picture shows the underside of the roof 'insert' for this small birdbox.  To the top and bottom are 5 IR LEDs in total.  These are wired in parallel, the only resistors are between the ground and the uln2003 (same arrangement as above there).  100 ohm seems the sweet spot.

The IR LED setting is still a bit spotty for my liking.. but I haven't found an IR led panel.... 
Last year's version of this used 4 IR LEDs, on in each corner, leaving a dim bit in the middle... which is exactly where all the action is... I fixed this this time by adding a fifth LED that points a the middle specifically.  The camera is rotated by 90 degrees to include the entrance hole in the image as well.  A 3mm green led indicator is added via one of the GPIOs (+10k ohm resistor) so that if I cant connect to it, it's obvious from the ground that there is/isn't power.

The diffuse white LED is an improvement on a previous version of this that uses 5mm white LEDs:

Nasty spotlighting in a previous version of the 2019 build, will be re-made with LED panel...

This is the business side of things:

Apart form the modifications described above, it's exactly the same setup as last season's box: Raspberry pi ZeroW (wifi), 8 MPixel v2 IR camera, IR-cut module. 5mm IR LEDs and White led panel (updated).  Illumination has high, medium and low (mood lighting) options.  Power is via 12v to 5v step-down. 

Video capture uses pikrellcam.  Video / image captures are stored locally, and can be archived off to a separate PC via wifi.  Local video storage overwrites if space drops under 20%.  This version uses the 'Lite' version of Raspian, so there is 13Gb available on the 16Gb micro SD card.  Based on last year's nesting activity that is good for approx 1 day of before a review is needed.  The prototyping board sitting on top of the Raspberry pi zero W is wired as described in this post (minus the rat-nest of wires for the LEDs...)

I wanted to add a microphone this year, but had real trouble getting a USB microphone adaptor working without noise interference.  The semicircular thing poking out the micro USB socket is a microphone, but its a fairly poor one.  I will probably not turn it on in practice... ironically a usb webcam I had spare gave flawless audio, but there was no way to fit it in and it seemed a bit silly to use an additional webcam as a microphone anyway...


  1. These backlight modules look great. I guess they could be mounted on the box side panel to provide more depth to the image (i.e. cross lighting/shadows).

    I think you have produced consistently better images than me from your camera/lighting setups now I stand no chance of catching you up!

    1. They are ace! and also cheap...
      I've got a 3 box production line for this year. I already completed one before finding these LED panels, so will do a quick re-work on the first. re the images quality - I just need the birds!. Sod's law says that they will nest in the boxes with the spotlight leds... ah well, at least I have incorporated IR cuts as standard now...

    2. I had a neighbour that was successful with his very first box. The following winter he took the box down from the tree and left it on a patio chair, complete with cables, and fitted a new super-duper system in its place.

      The blue tits obviously didn't read his memo, because they nested in the old box. Luckily he didn't have children or pets to bother them ...and didn't need to sit on the chair!

  2. Just thinking about your microphone noise problem, if the noise is broadband then you probably cant do much about it (except maybe extend the cable to move the mic away from the electronics). But if the noise is either predominantly high frequency or low frequency, you may be able to reduce it with digital filtering.
    e.g. Its easy to implement a low pass or high pass filter using a SoX command.

  3. I keep my SoX on my feet...
    At best its a background hum, at worst its a cyclic buzzing noise. I've tried smoothing the input voltage with various capacitors, also applying the same method at the USB audio card, with no effect. I did work out that the same mic + USb audio card gave different, reproducible interferences on different Pi Zero Ws. I tried multiple different mics as well as different USB sound cards with short/long USB cables - in the end decided it’s just too hit & miss (and expensive to figure out). I'll have a go applying SoX, as I think I can run it in situ with the box in a tree....

    1. My SoX already hum!

      Probably the easy way to experiment is to make a short audio recording with your current setup, then transfer the file to your laptop and try to clean up.

      Something like: sox {recorded file name} {output file name} sinc 200

      ...should filter out mains hum and other stuff below 200Hz.

      There may even be something useful in one of those clever Python libraries which you like using.