I held it right in front of my eye, so it was pretty much straight-on. It was 2AM and I wasn't thinking all that well, I guess.
I would have thought that, what with the diode log curve, at some current the voltage drop would be too low to bandgap any photons. These LEDs run around 100 mV/decade current, as opposed to 60 for silicon. It would be really interesting to do a lock-in sort of experiment to see what happens at really low currents.
We used to run the old Cree SiC blue LEDs at 50 mA, as trigger indicators on our VME modules. Two 74F38 sections in parallel as the driver. A couple years later, the LEDs were much better and customers called to complain that they couldn't stand to be in the same room with them. We now use the beautiful Osrams at about 3 mA, backlighting a window in a frosty polycarb sticker.
I wanted to report that I ordered the two Vishay high-efficiency 7-segment displays, and did tests while driving them from a TI MSP430G2452 microprocessor. I had the main clock at 8 MHz, and the watchdog timer used that clock with a 32,768 divisor, yielding an interrupt rate of 244.1 times per second. I fired the segments in sequence, one segment per interrupt, then repeated that sequence over and over. So each segment is turned on for 4 msec 35 times per second, and of course is turned on 1/7 of the time.
I used a single 1K resistor in the common cathode, or anode as appropriate, with VCC at 3.6V, and the digit "8" appears to be quite bright, and appears to be constantly on so long as the display and the eye aren't moving. If there is any movement, the display sparkles a bit, but not enough to cause any problems at all. I also tried all this with a 2.2K resistor, and still thought it was bright enough to be fully useable.
So as far as power is concerned, this is equivalent to driving one segment through 1K continuously. I haven't done the test of comparing that to driving all seven segments continuously with, say, 7.5K or 10K resistors on each segment. It appears to be the consensus here that the brightness would be the same. I'm not sure about that.
Anyway, it does work, and it's fast enough not to flicker.
Maybe there's a difference for discerning illumination sources. But if you put a standard eye chart on the wall and start dimming, you can't read it at all, in light levels that still permit decent navigation, i.e., well above the threshold of vision.
But, unless your eyes are better than mine, you can't read at all with peripheral vision (e.g., displays). That requires the central vision (fovea), which has poorer sensitivity (but much better resolution and enhanced neural post-processing, e.g., object recognition).
Yep. The savings in daylight would be very considerable. For purely night use I think a straight LED display saves more power, since it avoids the LCD transmission loss.
I'd done those tests too when I sent you those parts, for a project that was, last time I looked years ago, hanging on someone's wall.
I was taking notes in the dark at lower and lower illumination levels for the tests. That's how I personally discovered that I couldn't read -whatsoever-, at light levels that really weren't all that low. Navigating was easy, but ability to resolve objects' features fell with light level, from full resolution to zip.
Having done some star-gazing, I tested looking off-axis. :) Optimum wavelength is aqua, oddly. Or maybe not odd, since it's very moon-like, thus a sensible adaptation.
A white LED does amazingly well--best I tested by leaps and bounds (I tested many, many LEDs & wavelengths). White LEDs' luminous efficacy is scads better than aquas'.
IME that means leaky, usually from static damage. Detectable as a high threshold of visibility under forward bias, or possibly by reverse leakage.
Could also be from ambient light too (if you're not ultra-careful to keep the LED dark), and/or older LED process technologies.
I measured white (blue InGaN + phosphor) 5mm 20mA LEDs, and found them quite linear from medium to low currents. I don't remember how low I went--milliamps to microamps? Low enough that I was having trouble seeing the darn things in the dark to measure them.
Not as big as your range, though.
I did see some clinkers, and all had been 'hurt'. (I was intrigued by one Chinese lot that was much more efficient than Nichias of that day at high current, but worthless at low. They'd been damaged, culls.)
Aqua is pretty close to 510nm peak response of the rods and on the turquoise side of green at least to my eyes. The OIII nebula lines at
501nm & 496nm look distinctly oily green to cyan to me in the handful of objects with surface brightness sufficient to see colours.
The peak scotopic response is quite sharp around 510nm so 550nm green is perceptably less bright almost by a magnitude like for like.
Blue is incredibly disappointing. I have green 532nm and violet 400nm laser pointers for doing stargazing talks and the beam from the green one looks pretty much like Darth vaders death ray allowing me to apparently point at individual stars on any night. The violet one is all but invisible except on the darkest of moonlit nights and even then a bit of mist helps. I had expected the extra scattering of shorter wavelengths to make up for the lower sensitivity but no such luck.
I was about to post something along the lines of the energy of the photon emitted must surely put a lower bound on the voltage across the bandgap for electroluminescence to occur. Then I noticed that in your low current limit below some of the device voltage drops are already well below that of the photons being emitted. I'm puzzled by this.
Ballpark numbers off back of envelope:
red 630nm ~ 2eV (1.72!!) yellow 590nm ~ 2.1eV (1.77!!) green 550nm ~ 2.3eV (2.62) blue 480nm ~ 2.6eV (2.65)
Be interested to know the terminal voltage at which they stop emitting as judged by a fully dark adapted eye. There is clearly some multielectron process going on in the red/orange/yellow LEDs.
The other thing that bugs me is the huge variation in individual LEDs low current performance in domestic LED lighting (bare LEDs visible). The average is that they stay lit for a few seconds after switch off but now there are outliers that stay lit for minutes. They seem to be in a pretty basic series configuration and were presumably picked and placed off the same tape from the same batch so why so different?
That isn't normal - at least in my experience. A lot of people do have trouble reading by pure red light to preserve scotopic vision as was common in observatories and submarines but these days I think almost everyone has switched to faint white light for reading charts.
Seriously here - have you had an eye test recently? You really don't want to lose your sight and the brain is rather good at filling in gaps.
That is probably due to geometrical aberrations in your eye lenses that only really kick in when the iris is fully dilated. You could have it corrected although it is a tough call to get your optician to do a test for a pair of glasses dark adapted.
The standard computer generated test charts are way too bright for scotopic vision testing so you need old style charts and the time needed for dark adaption makes it an expensive luxury to have a set measured and made unless you have a tame optician friend.
If you try looking through binoculars can you still get good focus on stars and if not then what about the moon?
Equally how do the stars look to you at night? Looking for Alcor & Mizar in the Plough is a reasonable test of normal night vision.
I am aware these days that with my uncorrected vision stars are no longer tack sharp. In my youth I could just about pass the Centurions lookout test on a good night but never see Jupiters moons naked eye.
Let's see, I think the ~100 mV/ decade of current (vs ~60mV) is due to the non-ideality factor... which is close to 2 for pn diodes and it looks like led's too. (And that has something to do with where the carriers recombine, I think it's two if they all recombine in the depletion region, and one if move into the doped region before recombining. Though I may have that backwards.
As far as the "non" energy conservation. h*f > e*V (f is the photon frequency) I think the rest is made up for by thermal energy... only those electrons on the edge of Boltzmann's tail get excited.
There was a report (from England?.. or MIT?) about getting more energy out of a led (in light) than went in, I*V, at low forward current. "Harvesting the thermal energy".
I'm sure it's completely normal. It's why "shadowy figures" are "shadowy" rather than merely dimmer, fully-detailed figures.
It's also seen in the medical literature for scotopic visual acuity.
Try sitting across from a bookshelf with a variable light source, and watch, as you dim the source, how your resolution fades, which details and print fade first, etc.
I first did that testing optimum LED flashlight power-levels.
That may be because the eye has marked chromatic aberration--different focal lengths at different wavelengths, longer at red, shorter in blue wavelengths. Red light's great if you're nearsighted--it shaves about a diopter off your myopia. Green is neutral, blue is awful.
I was doing those tests fifteen years ago, and yes, my eyes were fine, and are still fine. I studied eyes in some depth once upon a time, and have subjected mine to all sorts of tests of acuity, focal length, optical error, resolution, etc. Better than most AFAICT.
I can measure focal length to The standard computer generated test charts are way too bright for
No problems whatever.
I was myopic, so never had that resolution without correction, and had discovered by 10 years of age that spectacles reduced my dark vision.
Half(?) your pixels and all your color, high-res vision is in the fovea. Once you drop below its threshold of vision, your resolution drops markedly, it's as simple as that. Cones don't work in low light.
Reading benefits from the high density of cones in the eye's fovea region (the center of your field if view, where you stare at something). So, it's normal, when the light is too dim to see with cone cells, for the rod cells to take over. Rod cells are far apart in the fovea, see with best resolution off-axis, and only render monochrome.
So, when your eyes dark-adapt, staring straight at text makes it hard to resolve; you need to see it out of the non-central part of your visual field, and it'll never be as distinct as with daylignt vision.
The classic trick is to use dim red light to read, and depends on the rods' insensitivity to red. The dark-adapted rods don't get dazzled, because red is outside their range of sensitivity. The cones in the fovea can render text by the red light, but of course a map or chart with color cues gets... to be a puzzle.
So is this an LED form of Maxwell's demon? If it works the way you describe it violates the second law of thermodynamics.
Unfortunately my brain seems to not be working up to snuff today as I can't remember how thermodynamics even related to my heat pump! I'll need to refresh my recollections a bit and see if I can find the loophole for your LED demon.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.