Detecting Ultra Violet

A quick search did not yield any posts, so I thought I'd offer this.
One of my pupils is making a cricket light monitor for which he will use an LDR, not least because its response reasonably matches that of the human eye. He then wondered about monitoring UV levels (sunburn etc) and spent a while searching for UV photo-detectors. Lots of ££sss!
Then I remembered a post somewhere (probably on this forum) about LEDs acting as PV devices that only respond to the wavelength of light they were designed to produce. A quick test with a 10Meg input impedance DMM, a UV LED and another powered UV LED (It was a cloudy day) showed a good 1.9 Volts when strongly illuminated. Next try was on a sunny day and equally useful readings resulted. Time to connect it to a PICAXE ADC input, bearing in mind that the 'input impedance' is nominally 47K. It reacted as I hoped (so the DC input resistance must be HUGE) but the string of LEDs did not decay when the sunlight was blocked. I wondered if the LED junction capacitance was storing the voltage, so I put a 10M resistor across the LED. Sorted, it responds very well to sunlight, yielding ever higher values the more intense the sun gets (nearer noon).

On this basis he can progress a very useful project

They (UV LEDs) would make good sensors for when the sun is actually out (as opposed to cloudy) for solar panel trackers, especially the amorphous types that only really yield in blue sky conditions.

I have used a yellow LED as a sun detector on model rockets to provide roll rate. The forward biased LED fed directly into the non-inverting input of a MAX4167 opamp with the gain set by feedback resistors on the inverting input. This high impedance buffer between the LED and the ADC input provided a quick response to changes in light level without any form of bleed resistor.

Interesting, response rate is fine for this application and not needing another component is a plus. What else did you log??

radiogareth said:

bearing in mind that the 'input impedance' is nominally 47K. ....

They (UV LEDs) would make good sensors for when the sun is actually out (as opposed to cloudy) for solar panel trackers,

especially the amorphous types that only really yield in blue sky conditions.

Click to expand...

Hi,

Can you provide any references for those statements?

The source impedance for the PICaxe A/D converters is "recommended" to be less than 10k ohms to produce a "negligible" error, but that would mean less than 1 part in 1,000 (10 bit A/D converter) so the "input impedance" would be >10M ohms. But quoting any input impedance for the A/D converter is largely meaningless, because it is in the form of a leakage current (dependent on temperature and largely irrespective of the actual voltage) and beacuse the converter operates by taking a sampling pulse, not a steady (dc) drain.

Certainly Amorphous silicon cells have a different spectral response to Crystalline silicon, but I thought that Amorphous correlated reasonably closely to the human eye, whilst Crystalline extends into the Infra Red (as does about 50% of the sun's radiation). IMHO if you're developing a "solar tracker", it makes more sense to use a sensor of the same response type as the solar panel, i.e. either amorphous silicon or a normal (Visible/IR) silicon photodiode, as appropriate.

Also, note that UV has various "colours" (UVA, B and C) just as does (white) light (Red, Green Blue, etc.). Basically UVA (closest to the visible spectrum) causes "sunburn" whilst UVC (fortunately, largely filtered by the Earth's atmosphere) is responsible for skin damage (Melanoma, etc.), which is what is normally quoted for the "UV Index".

Cheers, Alan.

radiogareth, In addition to the sun detector, the logger measured longitudinal acceleration and pressure. The logger was flown on a 2-stage rocket and data downloaded to a spreadsheet for post flight analysis. It was interesting to see the acceleration spikes when the first and second stages fired and to plot the relative altitude from the pressure data. An article about the project was published in the January 2009 issue of Nuts and Volts magazine and included code for the PICAXE 18X (back issues available from www.nutsvolts.com).

UV detectors are used in some fire / flame alarms - I don't know how they are priced but they work well...

premelec said:

UV detectors are used in some fire / flame alarms - I don't know how they are priced but they work well...

Click to expand...

I find that statement somewhat challenging to understand; with respect, I think you mean IR - (infra-red) detectors. Ultra-violet radiation is not emitted until an object reaches an extremely high temperature - we are generally looking at plasma arc welding equipment before ultra-violet radiation is generated. Infra-red detectors are used in most high-end fire alarm systems.

See here : http://en.wikipedia.org/wiki/Flame_detector#Ultraviolet

LEDS as detectors generally shift of the order of 25nm or so lower than their peak emitting wavelength and have a spectral sensitivity width of about 50nm. So if you want to measure in the UVA band, select a 400-405 UV LED. Forrest Mims has a lot of information in his Radio Shack "Engineer's Mini-Notebook" series (Vol. 1). His original circuit,

http://haze.concord.org/corrected.html

Calibration is always a major problem and there is a lot of good information here,

http://haze.concord.org/101.html#manual

Elektor has a nice article on measuring daylight (attached).

Baxter

@jedy... actually flames put out quite a bit of the shorter wavelengths - due to an eye injury and vitrectromy and no lens in one eye I was startled when looking toward a gas cooking fire how bright it appeared. I have also seen demonstrations of UV flame detectors where a guy in a glass booth pours flammable liquid on himself and lights it and the Halon fire extinguishers put his flame out fast and he's still talking - I've always assumed the guy in the booth took a deep in breath before starting the demo [PS wear your safety glasses!]

UV fire alarm detectors used where clear alcohol is stored, I.E. Vodka/Gin distillery storage tank area.

These detectors are not cheap.

Nor's the Gin or Vodka

Russbow: ha,ha - excellent!

In response to Allycats questions....it was posted on a thread I had on last year about minimizing current consumption on an 08M IIRC. I put a 470K pot across the power supply to reduce current wastage in standby, that produced lots of information/comments about the A/D input characteristics. All other comments noted and read with interest, its only for some 'indication' of UV presence, its not even a systems and control project, just a resistant materials one, with something useful in it that can light up LEDs, make a bleep or two and 'make cricket safer'

jedynakiewicz said:

I find that statement somewhat challenging to understand; with respect, I think you mean IR - (infra-red) detectors. Ultra-violet radiation is not emitted until an object reaches an extremely high temperature - we are generally looking at plasma arc welding equipment before ultra-violet radiation is generated. Infra-red detectors are used in most high-end fire alarm systems.

Click to expand...

I worked in an aluminium re-melt facility for a while and the safety circuit on the controllers for the gas burner do use UV sensors.

Just consider how effective using IR is likely to be for detecting whether the burner flame has gone out when the refractory bricks inside the furnace are glowing at somewhere above the melting point of aluminium, 660C.

Learned something. Good information. Thanks.

They (UV LEDs) would make good sensors for when the sun is actually out (as opposed to cloudy)

-Ive always heard you can get a burn on a cloudy day.
That aside, it sounds like you have come upon a very useful device. The only caveat is whether the led is responding too much to blue light >400 nm.
Several solutions:
One, get a true uv led from Nichia.
two, use 2 sensors, one with a UV blocking filter on. subtract that from the reading you get w/o the filter.
jb

JB, true indeed, you can get burnt, but in the absence of anything to calibrate it against its hard to see anything other than a definite difference based on the orientation of the sensor to the sun and the time of day/ strength of the sun. It's a wide angle uva 'straw hat' led, but easy to position it at 90 degrees to the sun within about 5 degrees. It will do as a simple warning but as usual this forum has educated as well as answered Gareth.

If you have a spare led it would be worth trying filing the lens flat as this will then make it not so directional to the angle of light, i often do it with leds if i want a wide beam of light compared to a narrow focused spot, all leds have different lens angles (most around 15 degrees) which can be rather narrow unless pointed directly at the source.

uv on cloudy days

radiogareth said:

JB, true indeed, you can get burnt, but in the absence of anything to calibrate it against its hard to see anything other than a definite difference based on the orientation of the sensor to the sun and the time of day/ strength of the sun. It's a wide angle uva 'straw hat' led, but easy to position it at 90 degrees to the sun within about 5 degrees. It will do as a simple warning but as usual this forum has educated as well as answered Gareth.

Click to expand...

And someone suggested filing it flat - good idea, and has been done before. (1) The idea I guess is to lessen the orientation issue...giving you an absolute measure of total uv, roughly.
If you look at sensitivity spectrum of "uva" led, you may get some sense of its responsivity to blue vs. uv. UV blocking filters are widely available for camera lenses, and this can be used for a
comparative measurement, against plain glass. As well, there is the issue of which wavelengths exactly cause sunburn. Does it extend to 420 or 450 nm. etc? (visible boundary is conventially set at 400 nm. You led may be
sensitive out to 450. And of course, the obverse is also true - perhaps sunburn only happens beyond 380 nm., etc. Nailing that down is a critical part of what you must do.)

just a few thoughts.

1. Mark Johnson, "Photodetection and Measurement." Used wet sanding.

A quick google produced this http://dermnetnz.org/reactions/sunburn.html

Yes I have used leds with the lenses sanded off and polished again (or not) depending on usage (useful when using ping-pong balls as diffusers). Not having to specifically orientate it will help it's usefulness a bit. Pupil has now decided he wants LCD dual bar-graph type displays rather than LEDs. More ploughing the forum to find if someone has done that already. Got some nice white on blue 2*16 backlit that work fine on other projects.

Gareth

Sending $FF or %11111111 to a lcd as a charactor should give you a single black (or white) square block, so all you need to do is adjust how many blocks you write to the display line to give you the bar graph result.

Be warned the white on blue LCD's are often next to useless out doors in sunlight, although they work well indoors.

http://www.coolcomponents.co.uk/colour-detector-sensor.html

Or search eBay for 'Arduino colour detector' you can get them for about £5 or much less, I2C interface I recall.

This circuit used an IR LED.
You could swap it for an UV LED and see how it goes.

http://www.picaxeforum.co.uk/showthread.php?7766-Handy-IR-Tester