Topic: Hacking Davis VP2 Wireless sensors?

[SLOweather]

Has anyone hacked any of the VP2 wireless transmitters/sensors to use different sensors, or measure different parameters?

If so, I'd like details.

If not, is there any interest in it beyond whatever I decide to hack? I have a solar battery bank I'd like to monitor the voltage of.

Any thoughts on the best one to mess with? I'm thinking the leaf wetness/soil moisture box has the most potential.

Does anyone have the schematic for it?

[Bushman]

I'm SERIOUSLY consider building a series of third party sensors.  I have several RF EE firends that can help.  The stuff being pumped out now is so cheap it is not even funny.  My first sensor would be an indoor sensor about 1/10th the size of the current temp sensor.

[drew1021]

Yeah, there are thermistors out there with higher accuracy that would be great for us serious weather enthusiests. Say +/-0.2 

[wxtech]

Yeah, there are thermistors out there with higher accuracy that would be great for us serious weather enthusiests. Say +/-0.2 

Why?  Why do you need such great accuracy for climatology? 
I've worked in research labs with triple insulated temperature controlled vessels.  If you move the temperature sensor just inches up or down inside the vessel, you'll get a different temperature.
If you don't have a lot of air mixing in the vicinity of the sensor, you can slide the sensor up or down on the mast and get a degree or more difference in temperature.  So why try to measure to such accuracy?

For climatology, measure to less than a degree and round to the nearest degree.

[johnd]

Why?  Why do you need such great accuracy for climatology? 
...
For climatology, measure to less than a degree and round to the nearest degree.

Completely agree. Unless you're going to adopt standardisation to the highest degree (eg exactly what height are you going to keep the grass that's exactly 4 feet below the temperature sensor - it really does make a difference at 0.2deg temperature resolution) then estimating temperature to the nearest integer degree (F) is going to be as good as makes any sense.

[drew1021]

You have valid points wxtech, however there are times when the temperature is relitively consistent due to a well mixed atmosphere, esp. during the winter months. It is at times like these when I appreciate the added accuracy. I currently have 24hr FARS so I know how much more stable the temperatures are. Also if someone is comparing their site to an official NWS site then you want the highest accuracy possible. It would be nice if everyones home wx station setup were identical for comparison purposes but we know this will never happen
Personally I like the feeling of knowing my station is broadcasting the most accurate data possible under the circumstances.

[dalecoy]

Yeah, there are thermistors out there with higher accuracy that would be great for us serious weather enthusiests. Say +/-0.2 

OK, so why not  +/-0.01  ?

[drew1021]

Sure if there are thermistors out there with that high degree of accuracy. I've not seen any yet. Even if they were I'm sure the price would be way out of my reach.

[Bushman]

Digikey and Mouser sell them pretty cheap.

[dalecoy]

Sure if there are thermistors out there with that high degree of accuracy. I've not seen any yet. Even if they were I'm sure the price would be way out of my reach.

About the best you can do with a thermistor, as a practical matter, is roughly 0.1 °C.

However, you can do much better than that with an RTD.

But, assuming you were using a thermistor with a resolution of 0.2 °C, how would you manage to do calibration?  It's not just the resolution - you're wanting accuracy (and precision).

Also, how would you compensate for the non-linearity?  [That's not a big concern if you're only aiming for accuracy of a degree or two, but it's a big concern of you're trying to get accuracy of 0.2 °C]

[johnd]

Also if someone is comparing their site to an official NWS site then you want the highest accuracy possible.

I do think that there's a confusion of accuracy with resolution here. The only way that you're going to get maximum comparability with an NWS site is if you mimic the design and location of the NWS instrumentation in every possible respect. In reality this is impossible - there are so many factors, eg design of the shield, microenvironment of the sensor site, nature of the ground/vegetation, sensor measurement height, even as I said previously the length of grass, that can have significant effects on the temperature that you actually measure, nudging it up or down (and not just by 0.1deg but potentially by up to a whole degree or more).

The upshot of all this is that comparisons to the nearest integer degree are about all that you can meaningfully manage, precisely because you can't realistically control for all the other factors that have an impact on the temperature readings. The Davis sensors already manage a typical accuracy of 0.3-0.5degF (I know that the accuracy spec is 1degF but most people that have done accuracy checks find that the Davis sensors actually perform significantly better than their nominal spec) so, personally, I'd argue that any attempts to get better comparative data with other sites by using higher resolution temp sensors are going to give largely illusory results - they might make you feel good but there's no added scientific value.

[dalecoy]

I do think that there's a confusion of accuracy with resolution here.

It's very easy for people to confuse accuracy, precision, and resolution if they don't understand that these are three different terms.

Quoting Wikipedia:
The accuracy of a measurement  system is the degree of closeness of measurements of a quantity  to its actual (true) value.

The precision of a measurement system, also called reproducibility or repeatability, is the degree to which repeated measurements under unchanged conditions show the same results.

A measurement system can be accurate but not precise, precise but not accurate, neither, or both
[End of Wikipedia quotes]

Resolution is the ability to distinguish between two things.  For instance, on your TV screen, resolution is how many pixels per line (and how many lines on the screen).  The easy analogy in the Davis world, is that the resolution of the US rain gauge is 0.01 inch.  

One additional common problem with temperature-measuring systems is that they may show different results if a temperature is approached from below (warming up), and approached from above (cooling down).  That's a different slant on "precision" because the conditions have changed.  

[SLOweather]

My soil moisture/leaf wetness/temperature station arrived today. I've been doing some thinking....

Leaf/soil uses an AC excited sensor IIRC so the detachable sensor temp unit might be a simpler choice. Convert the voltage to a resistance following whatever algorithm you want by using processor with an A/D (PIC?) driving a digital pot connected in place of the sensor. No need to hack the Davis part.

I'd forgotten about digital pots. I think this may be the way to go. As you say, no hacking of the Davis part at all. And, in theory, it should even work with any AC excited sensors, even if the leaf/soil box uses AC to drive the temp sensor as well. That would be nice, as it allows for up to 8 sensors on one transmitter.

I don't know that I'd even mess with the leaf wetness inputs, with their limited range output (0-15). OTOH, the temperature inputs have a 190 unit range (-40 - 150F) and soil moisture has a 200 unit range. I don't even much care at this point how well the output as displayed on the console matches the numeric value of the measure variable, as I'd be feeding the raw Davis value into a database where I could manipulate it as needed for display. So, I could have an expanded range of say 10-15 volts for monitoring a 12 VDC nominal battery pack. 200 units into 5 volts would be 0.025 volts per unit.

Or, for a straight display, and wasting some resolution, 0-200 units into 0-20 volts would give 0.1 volt resolution.

Time to dig out the scope....

[wxtech]

If you're looking to convert a resistance temperature sensor input to measure voltage, use a JFET as a voltage controlled resistor.  You only need a JFET and two resistors.  Google "FET voltage controlled resistor" for the circuit.
Ref: http://www.datasheetcatalog.org/datasheet/vishay/70598.pdf