EEVblog® Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: SharpEars on December 27, 2015, 11:28:16 pm
-
I just put some voltage into a blue LED and got it to glow. "So what?" you say.
Here is the amazing fact: The glow is visible at a current of a mere 500 nA, that's right as in hundreds of nanoamperes (at approx. 2.26 V of forward voltage). In fact, in a dark environment, even less current will make the thing visibly glow. I find the fact that it can produce a visible glow at potentially less than a microwatt of power to be unbelievable. I was at the border of my measurement equipment's capability just trying to measure the current requirements of the darned thing.
Don't try this with a red LED (you'll be lucky to see glow at hundreds of microamps - a full three orders of magnitude higher).
-
Cree RGB LEDs for outdoor displays (http://www.digikey.com/product-detail/en/CLV1A-FKB-CJ1M1F1BB7R4S3/CLV1A-FKB-CJ1M1F1BB7R4S3CT-ND/1987488 (http://www.digikey.com/product-detail/en/CLV1A-FKB-CJ1M1F1BB7R4S3/CLV1A-FKB-CJ1M1F1BB7R4S3CT-ND/1987488)) are also lighting up at about the same current. They are extremely bright when driven properly, but as bright as normal indication LEDs at very low currents.
-
I was wiring up a little test fixture a while ago with some random red and green LEDs I had on hand. From 5V, I think I finally settled on something like 1k for the reds and 47k for the greens to get them to a reasonably even non-blinding brightness. That's around 40uA for the greens, I'm not sure how low they would go, but it's really ridiculous how much light you can get with such little current.
-
I have a nice 1 li-ion torch with an XML in it. I have a tritium vial on the outside to help locate it at night. A few years ago I installed an 0805 sized blue LED into the head so it can be 'seen' next to the XML. The blue LED is direct driven from the li-ion cell through a 100k resistor. Works great. Total drive current is <10uA so effectively no issue in terms of long term runtime issues :)
cheers,
george.
-
dont forget that our eyes are very sensitive to green, so that may be why green leds seem to be brighter than other leds.
-
dont forget that our eyes are very sensitive to green, so that may be why green leds seem to be brighter than other leds.
... and are very insensitive to blue, making the <500 nA of drive current even more incredible.
-
Mike says he could see a red LED down to 500nA and a white one to 15nA(!) in this video:
https://www.youtube.com/embed/CIwlyy_KpXo (https://www.youtube.com/embed/CIwlyy_KpXo)
...and then shows a photomultiplier detecting the light at tens of pA.
-
dont forget that our eyes are very sensitive to green, so that may be why green leds seem to be brighter than other leds.
... and are very insensitive to blue, making the <500 nA of drive current even more incredible.
A white LED may be more visible due to the better wavelength for the eye, though the more diffuse source due to the phosphor may counteract this to some extent.
-
Off-topic (sort of) factoid:
Do you know why the telescope operators use red LED lights to read their maps/etc when they are out with their telescopes at night?
True, our eyes are more sensitive to green (thus you would think green is preferred).
However: They use red because the human eye's iris reacts the least to red....thus when the light is needed to read a star map or find an eyepiece, the light will not cause their "night vision" sensitivity to detail to be compromised.
It can take 30 minutes or more for a telescope operator's eyes to totally acclimate to the dark.
You would totally be schlonged (Dont you love the election season?) in that would would have to wait for 30 minutes to catch a good view of a DSO such as the Messier 101 pinwheel galaxy (or other DSO) if you used a green light to find that 8mm eyepiece. :D
v/r,
P
-
Mike says he could see a red LED down to 500nA and a white one to 15nA(!) in this video:
https://www.youtube.com/embed/CIwlyy_KpXo (https://www.youtube.com/embed/CIwlyy_KpXo)
...and then shows a photomultiplier detecting the light at tens of pA.
I should have probably mentioned that I used a regular non-superbright wide viewing angle diffuse blue led (i.e., not clear) which even at a full 20 mA is not all that bright compared to modern bright LEDs.
-
I have observed this same thing. some of the LED bulbs I have in my house stay luminescent for several minutes after turning them off from what I assume is the residual charge on the filter capacitors staying just barley above the voltage of the forward voltage drop of the led.
-
At low light levels, blue colors are at the peak of human vision response, while reds are nearly invisible. Check the attached curve. The key is the Scotopic curve, which applies at low illumination levels. (Think moonlight or less). At normal illumination levels the Photopic curve applies, and blue doesn't do so well.
-
At low light levels, blue colors are at the peak of human vision response, while reds are nearly invisible. Check the attached curve. The key is the Scotopic curve, which applies at low illumination levels. (Think moonlight or less). At normal illumination levels the Photopic curve applies, and blue doesn't do so well.
Now that(!) is an amazing fact.... :-+
-
Let your eyes adjust to darkness and you'll be able to see 10x lower current, or thereabouts. Which I believe is down in the endothermic range for most LEDs (i.e., it gets ever so slightly cooler as it produces light, because Vin < h*c / (e*lambda)).
Tim
-
Years ago a friend of mine built some crystal sets that used an LED as a detector. When tuned to a strong signal he described the LED as "like the cigarette of a sentry glowing in the dark"
-
I have observed this same thing. some of the LED bulbs I have in my house stay luminescent for several minutes after turning them off from what I assume is the residual charge on the filter capacitors staying just barley above the voltage of the forward voltage drop of the led.
I noticed this, too, with some power LEDs. I think it is some kind of fluorescence of the phosphors and not related to residual currents, because I was able to observe it on the lab bench with the power LED inputs both grounded :-)
It also is dependent on how bright and how long the LED was on.
-
I have observed this same thing. some of the LED bulbs I have in my house stay luminescent for several minutes after turning them off from what I assume is the residual charge on the filter capacitors staying just barley above the voltage of the forward voltage drop of the led.
I noticed this, too, with some power LEDs. I think it is some kind of fluorescence of the phosphors and not related to residual currents, because I was able to observe it on the lab bench with the power LED inputs both grounded :-)
It also is dependent on how bright and how long the LED was on.
I've also noticed this.
It's definitely the phosphor because I tested it with an LED which hadn't been used for days. I excited the phosphor with a blacklight and it carried on glowing when the light was turned off.
-
One more interesting fact about blue LEDs - the light capabilities at low currents will suffer and the light output will be noisy if the LED sustains a static damage. I've designed in the past a very high dynamic range (over 100000:1) controlled light sources for PMT calibration with a blue LED and a closed-loop photodiode control for the light output. You have to handle these LEDs with all anti-static precautions, otherwise they may become unusable for such an application, very noisy and non-linear at low currents.
Cheers
Alex
-
At low light levels, blue colors are at the peak of human vision response, while reds are nearly invisible. Check the attached curve. The key is the Scotopic curve, which applies at low illumination levels. (Think moonlight or less). At normal illumination levels the Photopic curve applies, and blue doesn't do so well.
This is a bit misleading. The scotopic curve is actually centered around green, not blue. (https://en.wikipedia.org/wiki/Luminosity_function)
The green curve in the following diagram is the scotopic one:
(https://upload.wikimedia.org/wikipedia/commons/a/a0/Luminosity.png)
Look at the visual spectrum to see the actual colors represented by the wavelengths. The blues in the 450-465 range are pretty low on the above curve, compared to the greenish peak:
(https://upload.wikimedia.org/wikipedia/commons/0/06/Srgbspectrum.png)
-
Here is the amazing fact: The glow is visible at a current of a mere 500 nA
500nA is still over 1,000,000,000,000 electrons per second... each one is a possible photon.
-
At low light levels, blue colors are at the peak of human vision response, while reds are nearly invisible. Check the attached curve. The key is the Scotopic curve, which applies at low illumination levels. (Think moonlight or less). At normal illumination levels the Photopic curve applies, and blue doesn't do so well.
This is a bit misleading. The scotopic curve is actually centered around green, not blue. (https://en.wikipedia.org/wiki/Luminosity_function)
The green curve in the following diagram is the scotopic one:
(https://upload.wikimedia.org/wikipedia/commons/a/a0/Luminosity.png)
Look at the visual spectrum to see the actual colors represented by the wavelengths. The blues in the 450-465 range are pretty low on the above curve, compared to the greenish peak:
(https://upload.wikimedia.org/wikipedia/commons/0/06/Srgbspectrum.png)
What is probably more misleading is that some manufacturers call an LED at 505 nanometers "greenish blue" which is where I got the long wavelength end of the blue range. Most people would probably call that a green with a touch of blue. It is still true that under low light conditions many "blue" LEDs will get a stronger eye response than many "red" LEDs. at 460 nanometers, what I would call a true blue the low level eye response is about 60%, while the red response is somewhere around 1 or 2 percent. What most people would call green is at about 540 nanometers and is not at the peak under either high or low lighting conditions, but is very high on the curve. In the intermediate, or mesopic range the eye response is some sort of blend of the two, and green is very near the peak of that response.
All of this is very dependent on the exact LEDs used, the viewing conditions, and how long the eye has accommodated to the light level. It helps to put color spectrum on same scale as the eye response plot when visualizing all of this.
-
Still, if that curve is in linear scale, then that's only -3dB, a very modest visual difference.
Speaking of cheating colors, there's "high efficiency red", which most people would still call red, but they're noticeably orange, and thus that little bit brighter visually.
Tim
-
Here is the amazing fact: The glow is visible at a current of a mere 500 nA
500nA is still over 1,000,000,000,000 electrons per second... each one is a possible photon.
At about a 20% probability (hence the typical efficiency of LEDs!); it would be 100%, except that:
- Some electrons are shunted through impurities or defects (especially at low currents and with shitty dice)
- Some recombine (spontaneously, or more likely at impurity sites) without emission
- Many photons get stuck inside the crystal (which, remember, has quite a high index of refraction, and the photons are created in all directions)
- And a few get stuck in the package/lens, or reflect back to the die
The first one is very apparent in circuit. Very cheap dice/emitters from China contain bargain-bin or rejected parts, often due to high leakage. These defects typically behave ohmic (~ohms to Mohms), so prevent emission at low currents (the voltage isn't high enough), but are not so great that operation at rated current is severely impaired (i.e., emission might be half or worse, but you probably already guessed that from the sticker price, riiiight?).
This is why, in a series-parallel array type emitter, you can get different intensities within a given string, at low current. If the dies were good quality, all the parts in a string would light evenly, as you would expect (you'd still expect variation between series strings though).
Tim
-
Ignoring the sensitivity of the human eye, the efficiency of LEDs is greater at either extremes of the visible spectrum. Deep red and blue/indigo LEDs are more efficient than orange, yellow and green LEDs.
Higher efficiency yellow and green LEDs which use blue LEDs and phosphors are being developed.
http://phys.org/news/2009-07-yellow-gap-full-conversion-blue.html (http://phys.org/news/2009-07-yellow-gap-full-conversion-blue.html)
http://www.compoundsemiconductor.net/article/91299-cranking-up-the-efficacy-of-green-leds.html (http://www.compoundsemiconductor.net/article/91299-cranking-up-the-efficacy-of-green-leds.html)
https://www.osapublishing.org/view_article.cfm?gotourl=https%3A%2F%2Fwww.osapublishing.org%2FDirectPDFAccess%2F27C75349-BD7A-D82E-896CD0A86BEB6A6F_199457%2Foe-18-11-11063.pdf%3Fda%3D1%26id%3D199457%26seq%3D0%26mobile%3Dno&org= (https://www.osapublishing.org/view_article.cfm?gotourl=https%3A%2F%2Fwww.osapublishing.org%2FDirectPDFAccess%2F27C75349-BD7A-D82E-896CD0A86BEB6A6F_199457%2Foe-18-11-11063.pdf%3Fda%3D1%26id%3D199457%26seq%3D0%26mobile%3Dno&org=)
-
It's definitely the phosphor because I tested it with an LED which hadn't been used for days. I excited the phosphor with a blacklight and it carried on glowing when the light was turned off.
The fluoro over my desk does the same thing to the screen on my scope. Eeerie to see it glowing after I turn out the lights.
Yes, all the diffuse led bulbs in our house have the afterglow too.
-
The afterglow is of course due to white LEDs and CRTs using a phosphor for light conversion. It remains active for a little while even after the original stimulating source is removed.
cheers,
george.
-
Plotted it again on a log scale. At low light levels both green and blue will be good, within 3dB of each other. But red will be at least two orders of magnitude worse, and possibly three orders of magnitude, depending on the exact wavelengths of blue and red.
This terrible response of the eye to red at low light is why red lights are used to preserve night vision. It just doesn't trigger the chemical reactions in the cone cells used at those light levels. An interesting question, which I haven't probed, is why evolution chose fewer, but higher energy photons to see in low light conditions. The answer isn't obvious to me.
-
The afterglow is of course due to white LEDs and CRTs using a phosphor for light conversion. It remains active for a little while even after the original stimulating source is removed.
Yet the LiFi guys (at least the slightly crazy LiFi proponents does) claim that they can be used for 200+Gbit data transfers....
-
You know that not all LEDs use phosphors, right?
-
You know that not all LEDs use phosphors, right?
If this is a come back to LiFi mention, then you are wrong, since all lighting LEDs use some sort of compound with afterglow.
And even if it was not the case, all light fixtures in my house have an enclosure that has afterglow.
And light communication using non-lighting LEDs has been around for quite some time in form of lasers.
-
You know that not all LEDs use phosphors, right?
For sure I do. But the LiFi people usually claims that your regular white LED lighting in your home and office will be used.
-
You know that not all LEDs use phosphors, right?
Yes, obviously. I did write WHITE LEDs (the typical ones we see from folk like Cree etc) when I stated phosphors.
cheers,
george.
-
Plotted it again on a log scale. At low light levels both green and blue will be good, within 3dB of each other. But red will be at least two orders of magnitude worse, and possibly three orders of magnitude, depending on the exact wavelengths of blue and red.
Your attachment contains an error: the wavelength should be nanometres not micrometres which is in the far infrared spectrum.
This terrible response of the eye to red at low light is why red lights are used to preserve night vision. It just doesn't trigger the chemical reactions in the cone cells used at those light levels. An interesting question, which I haven't probed, is why evolution chose fewer, but higher energy photons to see in low light conditions. The answer isn't obvious to me.
I don't know the exact answer but could guess it's because blue light is refracted over the horizon so at dawn and dusk there's not much red light.
The afterglow is of course due to white LEDs and CRTs using a phosphor for light conversion. It remains active for a little while even after the original stimulating source is removed.
Yet the LiFi guys (at least the slightly crazy LiFi proponents does) claim that they can be used for 200+Gbit data transfers....
That shouldn't be a problem. The phosphors only emit the longer wavelengths so add a blue filter to the sensor and the phosphor afterglow is no longer a problem.
-
Hint: Lifi doesn't communicate in yellowish wavelengths.
Tim
-
Plotted it again on a log scale. At low light levels both green and blue will be good, within 3dB of each other. But red will be at least two orders of magnitude worse, and possibly three orders of magnitude, depending on the exact wavelengths of blue and red.
Your attachment contains an error: the wavelength should be nanometres not micrometres which is in the far infrared spectrum.
Thanks for catching this. I usually work in the infrared and my fingers entered the units without checking with my brain. The chart is now corrected.
-
Plotted it again on a log scale. At low light levels both green and blue will be good, within 3dB of each other. But red will be at least two orders of magnitude worse, and possibly three orders of magnitude, depending on the exact wavelengths of blue and red.
Your attachment contains an error: the wavelength should be nanometres not micrometres which is in the far infrared spectrum.
Thanks for catching this. I usually work in the infrared and my fingers entered the units without checking with my brain. The chart is now corrected.
Heh, you now have two wavelength axes one in microns and one in nanometers.
-
Off-topic (sort of) factoid:
Do you know why the telescope operators use red LED lights to read their maps/etc when they are out with their telescopes at night?
True, our eyes are more sensitive to green (thus you would think green is preferred).
However: They use red because the human eye's iris reacts the least to red....thus when the light is needed to read a star map or find an eyepiece, the light will not cause their "night vision" sensitivity to detail to be compromised.
It can take 30 minutes or more for a telescope operator's eyes to totally acclimate to the dark.
You would totally be schlonged (Dont you love the election season?) in that would would have to wait for 30 minutes to catch a good view of a DSO such as the Messier 101 pinwheel galaxy (or other DSO) if you used a green light to find that 8mm eyepiece. :D
v/r,
P
You've got quite a bit of information lost in your retelling of this, the iris is very quick to respond (seconds) but only has an order of magnitude or so of range. The full range of human vision extends from around 1:100,000 instantaneous out to 1:1,000,000,000 or more through biochemical adaption (the slow process) with some individuals exceeding 10^14 dynamic range.
https://en.wikipedia.org/wiki/Adaptation_(eye)
https://en.wikipedia.org/wiki/Purkinje_effect
https://en.wikipedia.org/wiki/Accelerating_Dark_Adaptation_in_Humans
The sensitivity to bleaching is dependent on wavelength and you can maintain much of the dark adaptation even in the presence of yellow and green monochromatic light (LEDs can be broad enough they extend into the blues, or just have odd emission bumps), but red is the least bleaching. Its not as simple as the sensitivity curves presented here so far.
-
Another good read just about how sensitive the eye is: http://math.ucr.edu/home/baez/physics/Quantum/see_a_photon.html (http://math.ucr.edu/home/baez/physics/Quantum/see_a_photon.html)
... sounds a bit like a biochemical PMT, doesn't it?
-
I agree, the whole subject of eye response is very complex. FOVs, surrounding fields, illumination history, individual differences and a whole host of other factors. Measurements are difficult. Think of what the CIE engineers had to do in the 1920s and 1930s to try to generate calibrated amplitude monochromatic light sources to generate the originals of these curves. But engineering is powered by useful approximations, and if you are looking for qualitative explanation of why a blue LED might seem surprisingly bright at low power levels compared to a red LED these curves go a long way, even to the point of roughly explaining the power differences observed. If you are looking for time-nut or voltage-nut accuracy human visual response will be a very frustrating field to pursue.
-
I agree, the whole subject of eye response is very complex. FOVs, surrounding fields, illumination history, individual differences and a whole host of other factors. Measurements are difficult. Think of what the CIE engineers had to do in the 1920s and 1930s to try to generate calibrated amplitude monochromatic light sources to generate the originals of these curves. But engineering is powered by useful approximations, and if you are looking for qualitative explanation of why a blue LED might seem surprisingly bright at low power levels compared to a red LED these curves go a long way, even to the point of roughly explaining the power differences observed. If you are looking for time-nut or voltage-nut accuracy human visual response will be a very frustrating field to pursue.
Except that the steady state relative sensitivities you show ignore the huge photochemical effects which dominate the dark adaptation, and are an additional 3 orders of magnitude to consider/correct/understand.
-
Since we know little of the original posters experimental conditions we don't know if dark adaptation plays a role or not. Agree that if he is comparing a red LED viewed after a few minutes in a bright normally illuminated room to a blue LED in a dark room after his eyes have acclimated for half an hour there are several orders of magnitude more to try to account for. I assumed since he was commenting on the nanopower output of the blue LED that that observation was in a dark room and then further inferred that he was comparing a red LED in similar conditions. There are an infinite number of other possibilities. Which is why I try to do quantitative calculations of illumination, power, reflection etc in MKS units, and try to leave lighting evaluations to something on the order of: If the spousal unit likes it, it is good. Otherwise it is not good. if it is not good I ask questions like more? Redder? Successive approximation leads to an answer which is valid for a while.
-
Since we know little of the original posters experimental conditions we don't know if dark adaptation plays a role or not. Agree that if he is comparing a red LED viewed after a few minutes in a bright normally illuminated room to a blue LED in a dark room after his eyes have acclimated for half an hour there are several orders of magnitude more to try to account for. I assumed since he was commenting on the nanopower output of the blue LED that that observation was in a dark room and then further inferred that he was comparing a red LED in similar conditions. There are an infinite number of other possibilities. Which is why I try to do quantitative calculations of illumination, power, reflection etc in MKS units, and try to leave lighting evaluations to something on the order of: If the spousal unit likes it, it is good. Otherwise it is not good. if it is not good I ask questions like more? Redder? Successive approximation leads to an answer which is valid for a while.
You've also used normalised curves which hide the relative wavelength dependencies of the eyes components, here is a modern reference with much better presented data:
http://webvision.med.utah.edu/book/part-viii-gabac-receptors/light-and-dark-adaptation/ (http://webvision.med.utah.edu/book/part-viii-gabac-receptors/light-and-dark-adaptation/)
and this is still an interesting area for research.
-
I have one of those cheap alarm clock/mini digital picture frame combos (separate blue LED backlit LCD for the clock) that runs off an internal Lipo battery and can be charged over USB. I put a 100k resistor across the backlight transistor but it turns out that made the clock too bright at night! I added a piece of pink post it note between the diffuser and LCD both to change the color (much easier than trying to change the tiny SMD LEDs) and attenuate the light. Still too bright, so I increased the resistor to 470k and it was just right.
-
Since we know little of the original posters experimental conditions we don't know if dark adaptation plays a role or not. Agree that if he is comparing a red LED viewed after a few minutes in a bright normally illuminated room to a blue LED in a dark room after his eyes have acclimated for half an hour there are several orders of magnitude more to try to account for. I assumed since he was commenting on the nanopower output of the blue LED that that observation was in a dark room and then further inferred that he was comparing a red LED in similar conditions. There are an infinite number of other possibilities. Which is why I try to do quantitative calculations of illumination, power, reflection etc in MKS units, and try to leave lighting evaluations to something on the order of: If the spousal unit likes it, it is good. Otherwise it is not good. if it is not good I ask questions like more? Redder? Successive approximation leads to an answer which is valid for a while.
You've also used normalised curves which hide the relative wavelength dependencies of the eyes components, here is a modern reference with much better presented data:
http://webvision.med.utah.edu/book/part-viii-gabac-receptors/light-and-dark-adaptation/ (http://webvision.med.utah.edu/book/part-viii-gabac-receptors/light-and-dark-adaptation/)
and this is still an interesting area for research.
I've not used normalized curves to hide the relative color sensitivity, the relative color sensitivity is fundamentally unknown. For a brief explanation see (http://hyperphysics.phy-astr.gsu.edu/hbase/vision/colcon.html (http://hyperphysics.phy-astr.gsu.edu/hbase/vision/colcon.html)). We can say with some certainty that a blue light under some well specified set of conditions causes a perceptual response that can be related to the perceptual response to a red light under the same set of conditions. We can't say whether that is because the blue cones responded differently than the red ones or because the downstream processing provides different emphasis. Much of human vision is unknown physics, or perhaps psychology or even training. Think of the phenomenon of color constancy. Agree that there is much room for research.
-
I've not used normalized curves to hide the relative color sensitivity, the relative color sensitivity is fundamentally unknown.
You have normalised the photopic and scotopic spectral sensitivities, when they should be relative to each other (both changing in absolute and relative by amount of dark adaption).
-
I've not used normalized curves to hide the relative color sensitivity, the relative color sensitivity is fundamentally unknown.
You have normalised the photopic and scotopic spectral sensitivities, when they should be relative to each other (both changing in absolute and relative by amount of dark adaption).
I now understand your comment and agree. I should have placed them on two separate graphs, each of which would still be normalized to one for the maximum response. The concept is so deeply engrained for me that I didn't imagine that they would be interpreted as directly comparable. Sort of inherent in the definition. I can think of no correct way to represent both responses with an appropriate relative scaling on a single graph since each curve applies over a multi-order of magnitude brightness range even without the added complication of adaptation.