[wxsig] snow depth sensor prototype

[John Bennett]

Back in mid July Jay Moyer posted a link to an ultrasonic sensor and 
asked if it could be used to measure water levels. This caught my 
attention since I have long wanted to experiment with measuring snow 
depth. The sensor was relatively in-expensive (about $30 in single lot 
quantities). However, it only had a one inch resolution. I started 
searching for other sensors, but all of them that were affordable had 
the same resolution. The specs did not look as good as the MaxBotix unit.

After much deliberation I decided to go ahead and work up a design 
around the MaxBotix EZ1 despite the one inch resolution. While the 
sensor will work in any geographic area, it is probably better suited 
for northern latitudes, the high plains, mountainous regions of the west 
or east. In those locations they measure snowfall very often in feet 
rather than inches - or in my case here tenths of an inch.

I now have a prototype installed and awaiting the first measurable snow 
event. So far, we have only had several dustings. We do get measurable 
snowfall here - sometimes quite a bit - not not all that often. Here are 
pictures of the prototype:

http://n4xi.smugmug.com/gallery/7010539_UjLZd#448817306_wK9vc

What's with the white board below the sensor? It is recommended for snow 
measurement. I would refer anyone interested to a publication from CSU 
entitled: 'The Snow Booklet' by Nolan Doesken and Arther Judson. Here's 
the link if you are interested:

http://climate.atmos.colostate.edu/snowbooklet.shtml

What I quickly discovered is that the output of the circuit has 
considerable jitter. I do not believe this is a problem with the sensor 
itself, but rather has to do with the hardware design. In order to use 
1-Wire bus and keep the cost down, I elected to use the analog output of 
the sensor and feed that to an ADC - the same one used for the 
barometric and humidity sensors in the TAPR kits. Since the sensor and 
the ADC are not synchronized, the result is jitter as each ramps 
independently. The solution for this was to add an integrator on the 
output of the sensor with a rather long time constant. This seems to 
work, although it makes the calibration a bit more tedious (long time 
between reads).

The sensor was not designed for outdoor use, so care had to be taken in 
the design to insure the PCB assembly was adequately protected. I 
believe I have achieved this. However, after installing the sensor and 
looking at it, I believe the installation of a hood over the sensor 
assembly should add further insurance that rain will not be blown up 
into the transducer cavity. Even at that, the transducers are about 
$5.00 each. If they will last a season, it seems reasonable enough as a 
replaceable maintenance item.

At this point I have not integrated the sensor into my WXN server code. 
When I do, there will be support for at least four units. I have enough 
materials to build two prototypes. When I get all the bugs worked out, a 
second unit will be installed at this location.

I have also given thought to designing a unit with four or five sensors 
on a single PCB spaced in such a way that 0.25" or 0.2" resolution could 
be achieved. Of course this would run up the cost on the order of $30 
per additional sensor.

If the prototype works and there is enough interest, there is the 
possibility of making a run of kits or publishing a construction article 
for the PSR and/or QST.

Comments are welcome.

73, John
n4xi