Author Topic: About the UV Sensor (Read 8948 times)

wxtech · « **on:** April 23, 2014, 06:32:02 PM »

I was hoping to draw a schematic of the UV sensor board, but I'll share pictures and someone else can analyze the trace pattern. I removed the circuit board from the sensor enclosure. The board was shielded with aluminum covers on the top and bottom sides, soldered to the board. The board was covered with silicone or RTV. I removed that to see the board traces. These pictures do not show the 4 conductor cable. Yellow wire is a +3 volt power pulse from the Sensor Interface Module (SIM). Red and black wires are common and are connected together in the trace pattern on the board. Green is the signal output to the SIM that is available when the power pulse is applied.
The TO5 windowed can labeled D1 on the circuit board is the light sensor. The sensor has 3 leads. The center lead is connected to the can. This center lead is connected to the common/ground plane of the circuit board. The can has numbers 3446 5C which probably are part numbers. The circuit board has U1 an 8 pin IC, and U2 a 5 pin SOIC with no readable numbers on either.

wxtech · « **Reply #1 on:** April 23, 2014, 06:33:39 PM »

UV sensor circuit board pictures.

wxtech · « **Reply #2 on:** April 23, 2014, 06:36:15 PM »

Does anyone wish more information or test before I reassemble the UV sensor?

linuxfreak · « **Reply #3 on:** April 23, 2014, 07:20:31 PM »

Can you read the markings on the red cap next to the 8 pin chip? The backside looks like a charge pump voltage source, u2 might be either a MOSFET or power supply regulator, SOT23-5 package.

George

linuxfreak · « **Reply #4 on:** April 23, 2014, 07:46:57 PM »

I did a quick compare of the pin out of the dip8 package against the tlv2475 datasheet, doesn't match the pins, who knows, might even be a PIC!!

But it does look like that is a voltage charge pump booster circuit on the back with the SOT23-5 U2.

George

wxtech · « **Reply #5 on:** April 23, 2014, 08:03:13 PM »

Age; probably purchased in 2006 +/- 1 year. The circuit board does not have a silk screened part number.
Using 2 magnifiers, I read V2472A1 on the IC. Above that is 51CF with 1 character before the 5 and maybe more after. Not sure about the F.
On the red component, W1MA (not sure); next line star 700, or maybe 4700. On the reverse is 10. The pictures attached show more silicon debris than looking by eye.

wxtech · « **Reply #6 on:** April 23, 2014, 08:21:34 PM »

Does anyone know of a Wiki location with solar and UV sensor information? I'll put what I know on my weather web site technical library page, there should be an easy to find source on solar sensing.
http://www.lexingtonwx.com/techdata/

linuxfreak · « **Reply #7 on:** April 23, 2014, 08:30:43 PM »

So it is a TLV2472A1!

The cap looks like a polyester metal film cap, 200 volt(?) 4700pF or 0.0047μF. The through hole resistor is a 1%, bands work out to 1-0-0-x100k-1% = 10megΩ 1%, guess that's why it was used rather than an SMD.

Now just have to figure out what the darn SOT23-5 is.

George

linuxfreak · « **Reply #8 on:** April 23, 2014, 08:37:29 PM »

Al,

Get this file off of my site, http://www.hamiltonweatheronline.net/downloads/Solar_sensor.zip

I put it together a few years back when I was trying to duplicate the solar sensor.

George

linuxfreak · « **Reply #9 on:** April 24, 2014, 01:48:45 AM »

Might have found one possible SOT23-5 chip, LTC1517-5, that fits the bill for a charge pump in a chip: http://cds.linear.com/docs/en/datasheet/15175fs.pdf

And seeing the op-amp is a TI part, this might also be a candidate: http://www.ti.com/lit/ds/slvs324a/slvs324a.pdf

And for a Motorola flavor: http://www.onsemi.com/pub_link/Collateral/MAX828-D.PDF

See attached pic for another idea using mosfets.

George

wxtech · « **Reply #10 on:** April 24, 2014, 03:01:59 AM »

The solar sensor is back on the mast and working normally. I need to put this UV sensor back together and install it. I hope I didn't scratch the traces somewhere or otherwise destroy it. Oh well; for the team.
I can do a day of power on tests if you want to know what's going on actively. I'll use red & black as common and pulse the yellow with + 3 volts with the board in the sun. I'll probe around with a scope for waveshapes and voltage levels.
So, when there's no more interest, I'll re-apply RTV and reassemble.

wxtech · « **Reply #11 on:** April 26, 2014, 02:38:09 PM »

The UV sensor is back on the mast and connected. Not all the parts are back on yet. I didn't reapply the RTV yet. It still works OK after the digging that I did to it. When verified that it works OK, I'll finish putting it back together.
With red & black as common, +3 volts applied to the yellow wire, U2 pin 1 has -3 volts. Thanks for your analysis George.
http://www.lexingtonwx.com/vws/vws752.jpg is the latest VWS graph of UV. The graph will look more normal with time.
Al

kobuki · « **Reply #12 on:** December 29, 2014, 12:51:17 PM »

Sorry for resurrecting an old thread but I thought this info belongs to here than in a new thread. What I'm wondering is the filter cap above the sensor diode. It looks like some white plastic but it's probably some material able to let the required wavelenght through from the UV spectrum.

Does anyone have an idea what it might be made of?

linuxfreak · « **Reply #13 on:** December 29, 2014, 01:11:33 PM »

A WAG, Teflon or fused quartz glass.

George

kobuki · « **Reply #14 on:** December 29, 2014, 01:17:27 PM »

Sorry, I'm not familiar with the acronym 'WAG' (if it's indeed one) and a quick search didn't give me meaningful answers. Could you elaboarte on it a bit?

WA4TM · « **Reply #15 on:** December 29, 2014, 01:19:32 PM »

Quote from: kobuki on December 29, 2014, 01:17:27 PM

Sorry, I'm not familiar with the acronym 'WAG' (if it's indeed one) and a quick search didn't give me meaningful answers. Could you elaboarte on it a bit?

WAG = Wild a** guess.....

linuxfreak · « **Reply #16 on:** December 29, 2014, 01:20:09 PM »

Wild Ass'd Guess!!!

kobuki · « **Reply #17 on:** December 29, 2014, 01:20:46 PM »

LOL, OK

I have one too, but that's not the point...

SLOweather · « **Reply #18 on:** December 29, 2014, 01:40:15 PM »

I believe it's quartz, to pass the UV. Probably one of the more expensive parts of the sensor.

From https://www.weathershack.com/products/davis/manuals/uv_sensor.pdf:

Quote

Filter
With multiple hard-oxide coatings, the filter provides the Erythema
Action spectral response. It is stable in the presence of heat and humidity.

kobuki · « **Reply #19 on:** December 29, 2014, 01:40:22 PM »

OK, so one doesn't stop learning until death. Teflon/PTFE is indeed a good candidate to serve as diffusor for the UV sensor. It lets through the UV spectrum and is apparently in general use for the purpose.

Then another question: do we know the exact part number of the light sensor itself? All we know it has "3446 5C" written on it and has 3 leads and is a UV photosensitive element with a pretty large visible sensory surface under its glass.

kobuki · « **Reply #20 on:** December 29, 2014, 01:45:31 PM »

Quote from: SLOweather on December 29, 2014, 01:40:15 PM

I believe it's quartz, to pass the UV. Probably one of the more expensive parts of the sensor.

From https://www.weathershack.com/products/davis/manuals/uv_sensor.pdf:

Quote
Filter
With multiple hard-oxide coatings, the filter provides the Erythema
Action spectral response. It is stable in the presence of heat and humidity.

A-ha! So it probably has a Teflon diffusor as the outer layer for improving cosine response and a light filter to tune the spectral response. I'm starting to understand why this is one of the most expensive additions to the repertoire.

K8POS · « **Reply #21 on:** December 30, 2014, 12:11:58 AM »

I can not find "3446 5C" Not 100% sure what they used, but that type of Photo diode/Photo transistor is not cheap.
I still think being the design is several years old, there is a cheaper way to arrive at the same result.
I was looking at this:
http://www.thanksbuyer.com/a24b-uvm-30a-uv-sensor-module-analog-default-0-3v-voltage-output-w-signal-amplification-linear-output-29218
I have asked them for more technical info. I know it says 5 volt supply, but think 3 would be ok too. Not sure how low the Batteries is the wireless unit can go before things get funky.
I had started a thread at: http://www.wxforum.net/index.php?topic=24761.new#new but think I will ask admin to close it and keep this one going.
In a nut shell, the sensor needs to be able to detect UV light in the 280-360 nm spectrum. It needs to out put at 150MV per UV index. The local NWS station is about an hour drive from me but does have a UV sensor on site I could use to compare to.
The dust cap on top can be just about anything as long as it does not BLOCK out UV as some glass will. Please let me know what I can do to help myself or anyone else come up with a working design.

kobuki · « **Reply #22 on:** December 30, 2014, 07:03:10 AM »

Quote from: K8POS on December 30, 2014, 12:11:58 AM

I can not find "3446 5C" Not 100% sure what they used, but that type of Photo diode/Photo transistor is not cheap.
I still think being the design is several years old, there is a cheaper way to arrive at the same result.

My fear with these cheap sensors are that they might be very imprecise bacause they have a very small surface. With low angles at sunrise/sunset this could be a larger problem. The larger the surface the more precise the measurement can be potentially.

Quote from: K8POS on December 30, 2014, 12:11:58 AM

I had started a thread at: http://www.wxforum.net/index.php?topic=24761.new#new but think I will ask admin to close it and keep this one going.

This thread is about the Davis sensor. Ask them to merge your second with your first on the DIY subject, not this one.

Quote from: K8POS on December 30, 2014, 12:11:58 AM

In a nut shell, the sensor needs to be able to detect UV light in the 280-360 nm spectrum. It needs to out put at 150MV per UV index. The local NWS station is about an hour drive from me but does have a UV sensor on site I could use to compare to.
The dust cap on top can be just about anything as long as it does not BLOCK out UV as some glass will. Please let me know what I can do to help myself or anyone else come up with a working design.

Well, there're mainly 3 important bands for UV: A, B and C. B is the most important for humans, AFAIK. You might want to focus on that and choose an appropriate sensor. OTOH, besides spectral measurements, the cosine response is also important, and that's where diffusors come into the picture. So of you take it a little seriously, then the cover is actually pretty important. If you just want something that gives an estimate of the UVB index, then you don't need to care about these, probably.

K8POS · « **Reply #23 on:** January 04, 2015, 10:11:01 AM »

I ordered on of these to play with:
http://www.thanksbuyer.com/a24b-uvm-30a-uv-sensor-module-analog-default-0-3v-voltage-output-w-signal-amplification-linear-output-29218
I work at a hospital lab and we have a multi spectrum light source that covers the UVA and UVB bands so I have a means to see just what bands it really does detect. Not knowing how big the unit actually is, or if it will operate from 3 volts is one challenge.
From what I can see from the places that produce professional sensors (the ones that cost big bucks) the covers are anything from plastic to glass, but NOT made of material that will block UV. I am a little concerned that the actual sensor may be square and not round, the round sensor would possibly lend to better 360 degree sensing.
The Davis sensor appears to use a photo diode/transistor that is not cheap. ($50-$100 for similar ones at mouser but I did not see ones with the correct wave lengths) So the top cap filer would have to block everything but UVA and UVB as kobuki suggest.

Has anyone tried a similar idea?

kobuki · « **Reply #24 on:** January 04, 2015, 10:21:18 AM »

Well, the best sensors don't use photodiodes at all. But the shape of the photodiode sensor area is not really important. What's important is the frequency band of the light you want to detect can hit it fully, unobstructed. If parts of it are shaded by the sensor body it becomes a problem. Using a sensor like this will not allow you to detect sunlight at very low angles anyway (or only in a very imprecise manner). IIRC the Davis sensor is specified to provide accurate results only above 10°. And yes, you probably want to use the proper filter material. I'd use a diffusor as well as I previously mentioned.

News:

Author Topic: About the UV Sensor (Read 8948 times)