An interesting article on night vision

Aviathor wrote:

The G1000 I fly with is too dim if I leave it on auto-brightness. I need to adjust the brightness manually to around 10% at night.

That’s funny, I find the one I fly too bright, and have brought it down to about 1% which I find more comfortable.

Auto:

1%

Neat…

To continue this topic: I was able to find the time and installed Peter’s suggestion of an “adaptive” behavior into my night-vision-preservation-lamp. It was a lot easier that I originally thought, only two components were required: a photo-resistor and a plain resistor, connected to an unused analog input pin of the microcontroller. Everything else is already there. Double-clicking the on-off-button will start a 4-step measuring cycle. First, all lamps are turned off and the light of the environment is measured. This measurement will later adjust the overall brightness, very much like in most avionic units which have automatic brightness adjustment. Next, the colors red, green and blue are flashed in a brief sequence and the reflected light is measured. This measurement will set the colour of the lamp.

The German VFR ICAO chart for example is overall mostly greenish with labeling mostly in black:

Illuminating this chart with red light is not really good, because red is complementary to green. The background of the chart becomes very dark when illuminated by red light and reading black text from a dark background is very difficult. My lamp (I tried it out…) will set itself to the darkest possible yellow/green when used with the chart, resulting in black letters on a significantly brighter background – much easier to read. Unfortunately, due to the bad weather we are experiencing these days, I could not try it out in real life (telescope and airplane) yet.

I know that this whole exercise is mostly academic because paper charts have almost disappeared from both aviation and astronomy, but I think I gained a lot from the experience.

This is the simple modification to the lamp, I mounted the little photo-resistor in an aluminium tube underneath to shield it from the lamp’s own light:

Is it “adaptive” by itself i.e. it senses the ambient light or the colour of the surface being illuminated?

I am afraid not (yet)… “Adaptive” was one of the buzzwords in the early 1990ies when I did my ph.d. work in engineering (everything was called adaptive then). So I continue to use the term to make things sound more interesting

Adding a sensor for the ambient light should be no problem: a photodiode and a resistor is all the hardware required. I still have one unused analog input pin of the controller. That value could be used for setting the initial brightness when turning on the lamp. Sensing the colour of the illuminated surface is a bit more complicated. I will try to incorporate into the next prototype…

Long time no see… but finally my prototype “adaptive night vision lamp” is finished. The housing I bought was a tad on the small side, so a lot of tinkering was required to fit everything inside.

This picture shows the finished (so to say) lamp attached to a table using a photo clamp and a “magic arm”.

All this stuff is inside:

The lamp itself consits of 6 RGB LEDs that can be tuned to every colour of the rainbow. Additionally, the brightness and saturation (from “pure colour” to white) can be set. It is controlled by two rotary encorders with “click” function. One for ON/OFF and brightness, the other one for colour/hue and saturation, selectable by clicking the knob. Pressing the ON/OFF knob for long will store the current setting as default for the next power-up.
Additionally I added a 10W white LED (sufficient for a car headlight using the proper reflectors…) for white lighting. Astronomers need that for setting up their equipment during dusk and for locating parts that get dropped into high grass…

In order to prevent the bright white light to be turned on accidentally, I use a key switch for that.

Power supply is by XLR plug. 6V to 15V are acceptable with the voltage regulators and resistors I use, but the white LED will only work between 10V and 14V. A voltage sensing circuit makes sure that nothing goes wrong (during tests I fried two LEDs and one microcontroller, but now it should be OK…).

The heart of it all is an ATMEGA 328P microcontroller. I use an “Arduino Nano” board, because from eBay China than can be bought for 2 Euros, less than the chip alone would cost here. It comes with a voltage regulator (not stong enough for all the LEDs, so an external one is used here) and an USB interface that is very handy for quickly changing the software. And the “Arduino” software is quite easy to develop (C/C++) and runs on my Macintosh.

I know, Peter can make better PCBs than I, but mine are more colorful

Over the next few weeks, I will try out on the telescope which colour and brightness works best with my night vision and let you know in due time.

kwlf wrote:

I’d be interested to hear how it goes.

I will keep you informed once I have tested it. Just tonight I talked with a fellow amateur astronomer who is a biologist in real life. He was also interested in a variable color light because he thinks it could help him with microscopy, enhancing different colour contrasts. So I will make two of these “rainbow lights” for completely different purposes.

… it might be a pair of glasses that blacked out one eye instantly …

Quite a few astronomers do it that way, not with these instant LCD welding goggles, but with a pirate-style eye patch with which they preserve night vision for one eye all the time. I have tried that, but it gives me headaches and nausea if my eyes are illuminated in such an asymmetric way.

My idea was to use a combination of a deep-red LED, a yellow/orange LED and a blue LED. The red LED would be the brightest – for good visual acuity. The yellow/orange LED would provide a degree of discrimination along the red-green axis, and the blue LED would provide a little discrimination along the blue-yellow axis, all the while whilst stimulating the rod system minimally. I had a multispectral illuminator to hand, that I’d made for other purposes. The idea wasn’t to make the colours look normal, but to provide a degree of colour information, whilst preserving night vision.

My recollection is that it worked – kind-of. It was very dependent on the maps, so would work for an ordinance survey map, but not for a colour photocopy. Another problem – again my recollection is dim – was that at least some of the blue LEDs caused the paper to fluoresce. One of the issues with looking at maps in red light is that you can’t see the red danger areas. With my light source you could… But they didn’t ‘pop out’ at you in the way that they usually do.

For the photopic regime, you can use a combination of red, green and blue lights in essentially any balance, and you still retain colour perception once you’ve adapted to the light. It doesn’t matter if you make the red light source 10 x brighter than normal – you might perceive that there’s a red cast to the light, but provided you have some green and blue light, you can still discriminate colour. I’m not aware of anybody testing this in the same way in the mesopic regime – not that I’ve looked – and would anticipate that if the green and blue lights are very dim, colour vision might fail. But in retrospect I can see that using a blended combination of RGB lights as you’re proposing might actually be the most practical proposition. I’d be interested to hear how it goes.

If I were to invent any night-vision gadget, it might be a pair of glasses that blacked out one eye instantly, if exposed to bright light, akin to some welding goggles that only blacken when you make an arc.

Funny coincidence! On my breadboard I have a circuit with RGB LEDs, a microcontroller and two rotary encoders with which I can adjust color/hue over the entire visible spectrum and brightness. The aim is to build an astronomy flashlight. I am waiting for some parts in the mail and then I am going to assemble and field test that thing. In astronomy, dark adaptation is much more important than in the cockpit. One really needs 30 minutes for full adaptation and one second of light can ruin that. I always found reading (star) maps under red light very difficult – for my eyes it is almost impossible to focus in dim red light – and will do some experiments with my “rainbow lamp”, clear nights permitting.
In the cockpit of a (civilian) aircraft it does not matter much. As others have said, the screens are so bright even at lowest intensity that no real dark adaptation will be achieved (*). Therefore the aircraft manufacturers have dumped the red lights decades ago.

* Just look outside the cockpit window at night with the lowest level of instrument lighting and count the visible stars. Hardly more than a few dozen, even in the stratosphere. After half an hour in your back garden, you can see 1000-2000 in a clear night.