Are white LEDs "best" for LDRs of unknown properties?

[Mark Hammer]

LDRs all have different spectral sensitivities, with some more sensitive at THIS wavelength and others more sensitive at THAT one.  Sometimes, you get lucky and the LDR you buy has a defined spectral response that can be found in the corresponding datasheet.  Other times, though, you reach into, or order from, the "bargain bin" and what you end up with is an LDR that has a reasonably identifiable resistance range, but no identifiable spectral response....unless you have some very sophisticated test equipment...in which case what are you doing ordering from the bargain bin?

Looking at a flyer that came in the paper today, I was glancing at an LED flashlight on sale, and as always, I was pondering what ELSE I could do with such a thing.  Then it occurred to me: If you don't know what colour of light an LDR is most sensitive to, isn't the safest strategy to throw every conceivable wavelength at it?  In other words, when it doubt, shine white.

As always, there are bound to be some quirks I am unaware of, or that simply aren't visible to the naked eye. For instance, maybe "white" LEDs are a lot more "blue" than they appear, or maybe when you throw every wavelength at them, you have to severely cut back on overall illumination intensity to result in any appreciable LDR resistance change.

So, what say you?

[aloupos]

Grab some shrink tubing and some leds in different colors with similar intensities and chart the resistance across the LDR over a short time, say 10 seconds, using a pc based multi meter.  Not very scientific, because there are so many variables with LEDS, but for our purposes I think it would yield practical results. 

Could be an interesting experiment??

[Gus]

A white LED often is not white light.  Often a Blue/UV led with a yellow phosphor.
http://en.wikipedia.org/wiki/Light-emitting_diode 

[R.G.]

Good intuition, Mark. Yes, the spectral content of the light matters.

The sensitivity of cadmium sulphide (CdS) is a decent although not perfect match to the human eye's sensitivity range. This is one reason most of the available LDRs are CdS. LDRs don't have a couple of kilograms of neural matter second-guessing their responses, though, so they may not give you exactly what you think you see from an LED, especially if the LED is a couple of thin spectral lines and not really broad spectrum as Gus properly points out. The eye sees a few thin lines of the right relative intensities as even illumination (kind of) and the LDR is not fooled.

I've always preferred green LEDs for that reason. Almost all CdS peaks roughly in the green range, so you get close to maximum sensitivity, and I've always thought I could turn down the sensitivity either electrically, mechanically or optically if I had enough to start with.

aloupos has a good idea - test it. I've occasionally tested LDRs for sensitivity matching, more often for speed. LEDs turn on and off in microseconds if not faster, much faster than LDRs, so I would instrument up an LED/LDR to get a square pulse of current in the LED, then watch the LDR resistance expressed as voltage in a divider. This tells you the on-off time response directly and independently, and the flat spot in the middle tells you the bulk response at that illumination. The problem with this as a testing method is that "similar intensities". Unless you measure it, you have no idea about "similar intensities", and practically every detector you find that has a meter on it has a CdS cell inside. 

[frank_p]

Funny, I was looking for color identification circuits this week, so I just throw two links that might be interesing.

Even if it's not answering the question, I have the impression that comparative responses of the LDRs to primary colors and reaction trough different color filters could be a way to achieve this.

A couple of ideas here:
http://www.a2la.org/conclave2006/minutes/MTAC%20Meeting%20-%20Attachment%202.pdf
http://www.electronic-circuits-diagrams.com/sensorsimages/sensorsckt3.shtml

[Mike Burgundy]

As Gus states, white light LEDs are very limited indeed in the actual wavelengths they put out.
It's usually just two frequencies, not much else.

If anyone's interested:

Because of the way LEDs make light (exciting an electron into a higher orbit, which emits a foton when it falls back - the energy dissipated with that determines the wavelength exactly) the light they emit is in a very, very narrow-band. A measure of light "quality" (meaning how much of the spectrum is actually present in white light) is the RA-value:
sunlight is RA100, incandescent bulbs are 99-ish, high-quality fluorescent lighting is about 70 (which is more or less the line between good and BAD light), LED is horrendously low.
If you look at a colourchart:


You'll see all pure wavelengths along the "horseshoe". In between are mixes. This shows that white can actually be made with just two colours! If you have a 470nm blue (this is an actually available LED colour) and add a 570nm yellow (which is what the phosphorous layer mentioned by Gus is doing) you can get white.
Each (very narrow) colour needs it's own diode - making a better white LED is possible, by adding more and more LEDs on one die, but it adds to cost a lot, so manufacturers won't do that, at least not for signal-type LEDs. Also the number of available colours is limited. The difficulty in finding materials which send out different frequencies is evident if you think about how long you could only get red LEDs, then green, and orange (which then mostly was green+red...)
When incandescent is banned (which it will be soon) from production, LED substitutes will probably (hopefully) improve dramatically. I understand there's already a 5W LED claimed to have RA80 or so (Lumileds or Osram, I think)

[frank_p]

I've also heard that ordinary leds can be used a captor.
So if you could do a color sweep over some LEDs of different nature (and used as captor) in parallel with a LDR being tested, could we find out to what color(s) the LDR would be more sensible to by comparing where the maximum of the curves correspond  ?

http://en.wikipedia.org/wiki/LEDs_as_Photodiode_Light_Sensors

[George Giblet]

>  I've also heard that ordinary leds can be used a captor.

Try this trick.  Connect the terminals of an LED to your multimeter, set the multimeter to volts.  Then shine a light into the LED.

In effect the LED is working as a photocell.   Kind of reversibility: Current in -> Light out, Light in -> current out.

IIRC LDRs peak outs in the green/yellow region, which is confirmed here,

http://www.4-direct.com/resistor/photo-cds.htm

In the old days LDR used to make good cheap light meters because they weighted the spectra similar to the human eye (which peaks in the green region).  Of course there were more accurate and costly units with colour filters.

I know one guy spent a lot of time looking for the best opto LEDs and ended up using yellow.

[frank_p]

Try this trick.  Connect the terminals of an LED to your multimeter, set the multimeter to volts.  Then shine a light into the LED.

IIRC LDRs peak outs in the green/yellow region, which is confirmed here,

http://www.4-direct.com/resistor/photo-cds.htm

In the old days LDR used to make good cheap light meters because they weighted the spectra similar to the human eye (which peaks in the green region).  Of course there were more accurate and costly units with colour filters.

I know one guy spent a lot of time looking for the best opto LEDs and ended up using yellow.

Yes I played a bit with that a couple of months ago.  Great idea to start a couple of projects. But as RG stated, luminosity measuring and calibration seems to be a difficult topic in itself if you want to do some -scientific- work.  You can't do scientific experiments without a convenient measuring apparatus.

The problem with this as a testing method is that "similar intensities". Unless you measure it, you have no idea about "similar intensities", and practically every detector you find that has a meter on it has a CdS cell inside. 

[Jered]

  I'm building an Easy Vibe right now and have a question. Is it better to use clear or diffused LED's, or does it matter?
  Thanks, Jered

[George Giblet]

But as RG stated, luminosity measuring and calibration seems to be a difficult topic in itself if you want to do some -scientific- work.  You can't do scientific experiments without a convenient measuring apparatus.

What RG said is basically correct but if you have access to another piece of equipment like a light meter you can calibrate it.

An old cheap way to calibrate light meters was to use a 75W incandescent bulb of a known brand with a known light output.

If you want to do measurements where the color balance accurately reflects the human eye that an be very costly and difficult.  Doing it roughly is fairly easy, like the LDR approximation.

It's a real pain to make  calibrated sources for acoustics and light measurements at home, it's always easier to find/borrow a device which you calibrate off.  In both cases it's very expensive to buy sensors which have accurate absolute outputs.

[frank_p]

Well, thanks George (and all others) for all the good advices.
All my LDRs are from unknown source...

   

[Radamus]

A cheap idea from the guy who just finished CHM 101:

You can take a blank DVD and put it in a box at an angle. Allow a little bit of light in the top and a hole in the side to view the DVD. This will break the light and you'll know exactly what light your LED is emitting. Do it in the sun, and you get the full spectrum, like was already said. Do it in fluorescent lighting and you get several concentrated beams. Do it with an LED and see maybe just two? Just a thought. It's probably a bit easier than using more expensive equipment. I'm sure some of you engineers did this experiment at some point. Maybe if you know what light is being emitted, that will help the project?

Just a thought, and I know you guys are way ahead of me in all of this, but maybe simpler is better.

Curious for the results,
Conrad