Topic: About the UV Sensor

[DW7240]

If anybody is interested (and I am) you may like to visit the link below, this company does a uv sensor with a max 3v analogue output, sounds much like the Davis model, but at a fraction of the price, you just have to provide a suitable housing etc.  Also amongst the items they sell is a analogue light sensor with the same max 3v output, sounds like an ideal solar sensor.

http://www.adafruit.com/products/1918

http://www.adafruit.com/products/1384

Any thoughts............

Nick. DW7240.com

[mcrossley]

The UV could be interesting, but the light sensor has a logarithmic output.

[K8POS]

Ordered one 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
Pictures are not that great, but looks very similar to what Davis is using.

[K8POS]

My sensor board arrived.
I used a 3 volt battery pack to power it and applied the Spectrum light we have at the lab to it.
The specs state that it is responsive to 200-370nm, actual is 205-372nm, it appears pretty blind to anything else.
I have not figured out just what the 2 trimmer/multiturn pots are for but working on that now.  Hooking it up to the ISS was pretty straight forward, but it still thinks the UV index is 2.0 in the dark.
I have it back on my battery pack to decipher what the trim pots are for.  I am sure one is output gain, but not sure about the other.  The board is painted so it is very hard to see where the traces go, and I have no schematic for it.
Still to work out:
1) Getting to read zero in total darkness.
2) Finding a opaque cover that will not block all the UV.  Guys at the college physics lab said to try a ping pong ball?
3) Making a suitable housing.
4) Calibration.
5) Checking the angular response.

But hey, only cost me $16.00 and gives me something to do.

[mcrossley]

Guys at the college physics lab said to try a ping pong ball?

It  may work, be aware they degrade quite quickly in outdoor UV light - you may have to replace it quite often.

[mcrossley]

Is the 3 V output tied to the input ground rail as reference?

[K8POS]

I do not think so.
Looks like a simple sensor into an amp type circuit.
No matter how or where I trim the pots to I can never get it to go below 27mv in the dark.
When hooked to the ISS this is giving me a UV index of 2.0 when hooked to the ISS.  I know at 27mv this should be near zero.
Going to try putting a pull down resistor between the signal and ground to see what affect that has.  My peek hold meter is broken so I can not see what the actual MV reading is when hooked to the ISS.
Have to go get my scope back.

[mcrossley]

Hmm, the Davis input has a zero offset on the input of 70mV, then 14.3mV per 0.1 UVI, up to 2.395V which gives 16 UVI, anything above 2.395V gives a reading of 16. So your 27mV should be OK

When it reads the sensor the SIM switches in a 100k pull-up to 3.3V on the signal line. This may be causing your problems.

[linuxfreak]

Wish people would read!!

Check the other thread from Wxtech, I mentioned in there that Davis uses a switching regulator to get a negative supply voltage for a ground reference!

You are using a unipolar supply, it has to be bipolar.

George

[mcrossley]

Got a link George?

[linuxfreak]

Here's the two threads from April 2014.

http://www.wxforum.net/index.php?topic=22414.msg215778

and

http://www.wxforum.net/index.php?topic=22427.msg215865

George

[K8POS]

Thank you George.
That would explain the pull UP resistor not a pull down.  Would also explain why I get different readings between the ISS powered and battery powered.
May have to strip the paint off this thing so I can follow the traces.
Uses a 4130A amp, found a data sheet for that so it will help.

[K8POS]

Oh well, see from the data sheet this thing really wants 5 volts input not the 3 volts.
Back to the drawing board.

[linuxfreak]

Davis uses 6v, +3v & - 3v on the amp.

George

[K8POS]

I know there has to be a simple (I hope) way to make this work.
I think I am just brain dead from reading all the post here and on other forums, including the ones Mark put on the sandaysoft forums.
It will end up being one  of those in the middle of the night eureka moments.

Bob

[kobuki]

My sensor board arrived.
I used a 3 volt battery pack to power it and applied the Spectrum light we have at the lab to it.
The specs state that it is responsive to 200-370nm, actual is 205-372nm, it appears pretty blind to anything else.
I have not figured out just what the 2 trimmer/multiturn pots are for but working on that now.  Hooking it up to the ISS was pretty straight forward, but it still thinks the UV index is 2.0 in the dark.
I have it back on my battery pack to decipher what the trim pots are for.  I am sure one is output gain, but not sure about the other.  The board is painted so it is very hard to see where the traces go, and I have no schematic for it.
Still to work out:
1) Getting to read zero in total darkness.
2) Finding a opaque cover that will not block all the UV.  Guys at the college physics lab said to try a ping pong ball?
3) Making a suitable housing.
4) Calibration.
5) Checking the angular response.

But hey, only cost me $16.00 and gives me something to do.

Nice equipment you have access to! That makes things easy 

I'd suggest using a small disc made from a Teflon sheet as diffusor to improve the cosine response, as I've previously suggested in one of your threads on this subject. I think you also mentioned this DIY Pyranometer solution - it provides a good description of the diffusor and how to make one using a small teflon disc. A ping-pong ball won't hold for too long in direct sunlight, as it has been mentioned.

About the circuit output. Isn't this possible to just shift the values in  the console? It'd be nice to avoid designing a completely new circuit.

There's this thread about a similar subject - it's about injecting additional solar sensor data into existing transmitted packets by relaying them with a different station ID. Maybe you could even work together on a combined solution.

[K8POS]

I am on a wired station.  As far as I can tell, and I have tried, there is no way to change the calibration of the UV sensor using the console.  The Cumulus software does allow for this.
I want to get the electronics working before I worry about the rest of it.

[kobuki]

Yes, of course. But I think it's worth mentioning that the injection scheme is very much workable on a cabled station too, but on the RS-485 bus.

[K8POS]

My sensor arrived some time ago and I have been playing with the output of it.
The amplifier has a 1.5K multiturn trim pot on the output that is way to high.  Just looking through my side window (non uv coated) I can read 10 times what the UV actually is being reported by other near by stations.
It is kind of hard due to the station only checking the sensor every minute.  Wish I had a recording dc volt meter.
I ordered a 500K trim pot to replace it with and see If I can  dumb down the output some.  I have also considered putting a 1.5 or so trim pot on the actual output and see where that leads me.
The sensor is responding to the correct UV wave lengths, just puts out almost 10X what I should be getting.  Once I can get the output under control I will work on a housing and calibration.

[kobuki]

Returning back to the original topic a bit. The datasheet specifies 3V ±10% as supply voltage on the yellow wire. On paper it should be fine, but would a regulated 3.3V supply cause any difference in the output? The sensor has been dissected in this thread but I can't identify anything that resembles an onboard voltage reference or a regulator. So the question is, would output sensitivity change according to supply voltage or not? I'm hoping to hook up the sensor to an MCU that already has a 3.3V regulator onboard. I'd need to know if i have to scale the sampled output values or not.

[K8POS]

I do not think it will affect it.  The actual sensor produces current that is then amplified into voltage that the ISS/Console reads.
I had thought about using a picmicro and program it to send out the correct voltage, but then one would need a regulated constant voltage supply.  I have decided to take just the senor from the UV sensor I had bought, and I am making my own amplifier circuit that has known amplification and better adjustment than what I bought has.  The one I have incorporates a 5 volt amplifier that does not give reliable readings on 3 volts.  The OpAmp I will be using will operate from a min of 2 volts.

[kobuki]

Yes, after giving it some thought that's my own assumption as well. The opamp scales the voltage detected on a shunt and in essence the output should not be affected by its voltage rails (within specs, of course).

[K8POS]

Would be so much easier if I could alter the program in the console that figures out the UV index.

[ct]

Why does the UV sensor have 2 ground wires (black and red) ?

When connecting to a MCU do I connect both black and red wires to the MCU ground, or just one of them?

[kobuki]

You can connect both or only one of them, they're internally connected. Though if you run the cable on extended lenghts I think it's a good idea not to leave one of the pair floating.