The goal is to take photos of some fairly quick process (well, at least too quick for the human eye) at exactly specified times with sub-millisecond accuracy. Of couse, with expensive special equipment at hand (such as a high speed camera), this is a trivial joke, but I wanted to do it in my room with the stuff and "toys" I have laying around as well as a normal digital camera (Minolta) and an ordinary photo flash.

The camera has a much too long delay from pressing the button until the exposure actually begins so one has no chance to catch the drop this way. Furthermore, the minimum expose time is 1ms but it would be better to further reduce it in order to get sharper images.

So, the basic idea is to make the room dark, open the camera's shutter, let the drop fall down, fire the photo flash at the correct time and close the shutter again.

Still needed, however, was the delay trigger to fire the flash light some specified time after the drop crosses the light barrier. For the height I was using, the base delay until the drop crashes was about 1/4s to 1/2s (250-500ms). To see the different stages of the crash, some precisely defined amount of time in the order of single milliseconds must be added to this time.

It took me half a day to get a working sub-millisecond precision delay trigger for the drops: First, I played around with various RC-based solutions but they all could not attain the precision (if they worked reliable at all) and had the principal inconvenience that I would need to calculate the time from the rotation angle of a potentiometer. A further problem was due to the fact that the light barrier could fire two peaks for each drop of transparent liquid: one upon entering and one upon leaving since the drop is about as transparent as air while being just in the middle of the light barrier.

I finally settled down with a really easy microcontroller-based circuit: A 1MHz clock source was driving an Atmel '2313 AVR RISC connected via SPI and the Latch16Bit board with my computer. The latter was only needed to hack together, compile and download the firmware of the controller which took about an hour. A switch was attached to the Microcontroller to increase/decrease the delay time from the first falling edge of the light barrier until firing the flash. Sub-millisecond precision was achieved easily by having the AVR RISC counting its own clock (i.e. executing a defined number of instructions) so that overall timer accuracy was a couple of microseconds.

I'm presenting the complete circuit schematic below here: