Thread Rating:
  • 2 Vote(s) - 3 Average
  • 1
  • 2
  • 3
  • 4
  • 5

Wind sensor alternatives to FineOffset. Any ideas?
#1

FineOffset wind direction and speed sensors are cheap and easy to replace but not particularly accurate. Particularly the one that points the wind direction is known to be very "jumpy" (perhaps because it's too light) while simultaneously not picking the fine intermediate cardinal points as often as it would be statistically expected (meaning for instance than in the N-E quadrant it tends to accumulate readings in N - NE - E while NNE and ENE almost ever show up). There's a detailed study about this somewhere, I believe in SandaySoft forum, but it only takes a look at a daily windrose of readings to notice it.

My question then is: where to find a suitable replacement?

There is of course the Davis 6410, which I believe (correct me if I am wrong) can fit the WeatherDuino TX unit phone like jack without any modifications. It's nevertheless somewhat expensive (117 euros + 10 euros postage in the the linked offer) and according to a post somewhere in this forum not as durable as one would expect.

Does this leave us without options?
Reply
#2

(11-02-2015, 18:51)hvalentim Wrote:  FineOffset wind direction and speed sensors are cheap and easy to replace but not particularly accurate. Particularly the one that points the wind direction is known to be very "jumpy" (perhaps because it's too light) while simultaneously not picking the fine intermediate cardinal points as often as it would be statistically expected (meaning for instance than in the N-E quadrant it tends to accumulate readings in N - NE - E while NNE and ENE almost ever show up). There's a detailed study about this somewhere, I believe in SandaySoft forum, but it only takes a look at a daily windrose of readings to notice it.

This is a know issue with Fine Offset weather stations. The WeatherDuino Pro 2 does not solve this issue but you can greatly improve this behavior, with a very careful adjustment of R8 on the TX unit. Normally adjusting R8 to 10K will work, but you can do better... before install the wind vane in it's definitive place, connect it to your TX unit, manually rotate it and check that all cardinal points are detected. Then slightly adjust R8 in a way that the readings from the vane don't change from one positions to another with just a very small movement. This procedure also reduces the bouncing with low wind speeds.

(11-02-2015, 18:51)hvalentim Wrote:  My question then is: where to find a suitable replacement?

There is of course the Davis 6410, which I believe (correct me if I am wrong) can fit the WeatherDuino TX unit phone like jack without any modifications. It's nevertheless somewhat expensive (117 euros + 10 euros postage in the the linked offer) and according to a post somewhere in this forum not as durable as one would expect.

Does this leave us without options?

Currently the system does not support the Davis wind measuring instruments.
The mode of operation of the Davis wind direction sensor, though based on the same principle, is different from the Fine Offset sensors.
It is possible that at some point, I will add support for them. Is not technically very difficult, the major problem is that I don't have any of those instruments to test.

If support for other wind sensors were solely dependent on my will, it already existed, but ... good quality wind sensors, are few, and are expensive. I have already contacted with some manufacturers, for units to test, but even with their goodwill and some discounts, prices go beyond my means.

Anyway, I don't give up easily, in future I really want to add support, at least for some kind, of high quality wind instruments (speed and direction).

Reply
#3

So, I guess that by decreasing resistance (using an R8 component with 10 instead of a "15 to 20 KOhm, 3362P-1-203 Variable Resistor") we should get a "stronger" signal.

Seems it's a design issue as it tries to get 16 possible directions using only 8 switches (the combination part by activating two simultaneously does not work accurately).

[Image: W8681_wind_gauge_circuit_board_2.jpg]
from: W8681 wind direction indicator disassembly
Reply
#4

(12-02-2015, 11:39)hvalentim Wrote:  So, I guess that by decreasing resistance (using an R8 component with 10 instead of a "15 to 20 KOhm, 3362P-1-203 Variable Resistor") we should get a "stronger" signal.

Not exactly. Theoretically R8 can be a 10K fixed resistor, but even with an 1% precision resistor their value can be between 9.9K and 10.1K, and the most important: the Vref for the Arduino ADC is not always exactly 5V, so doing R8 variable, allow us to do some fine adjust. The procedure described on my previous topic, really helps to get a better precision from this instrument.
To better understand what I mean please refer to the document in this topic:

http://www.meteocercal.info/forum/Thread...nstruments

I think that the type of resistor that fits in the board does not exist with a value of 15K, so use a 20K, and to start adjust it as close as you can to 10K.

Reply
#5

(12-02-2015, 13:26)Werk_AG Wrote:  Not exactly...

Actually I misread (did not pay due attention to) the "Variable" (adjustable) part in the resistor description, thought we were talking about a fixed value.

I did not start collecting and paying attention to the smaller parts yet and it's a whole new world to me. Thanks a lot for your patience/guidance.

So,it seems the procedure is more like matching the Arduino and sensor voltage behaviors.
Reply
#6

(12-02-2015, 20:19)hvalentim Wrote:  
(12-02-2015, 13:26)Werk_AG Wrote:  Not exactly...

Actually I misread (did not pay due attention to) the "Variable" (adjustable) part in the resistor description, thought we were talking about a fixed value.

I did not start collecting and paying attention to the smaller parts yet and it's a whole new world to me. Thanks a lot for your patience/guidance.

So,it seems the procedure is more like matching the Arduino and sensor voltage behaviors.

This is the exact issue that brought me to WeatherDuino - I have tried all sorts of dampening, including an extension of the vanes on the wind sensor. What has shown most promise is arithmetic dampening of the wind vector. I have done this using an Arduino receiving the standard Fine Offset signal which is a sample transmitted every minute or so, which works well, but there is too much lag. What we need is an Arduino sampling the direction every second or so and averaging the wind vector to transmit a dampened signal to the receiver. The WeatherDuino TX seems ideal for this task.

Does this make sense?
Reply
#7

Yes, it does.

Currently the TX unit samples the wind direction every 5 seconds, and sends it, nothing more..
Due to the way that Fine Offset wind vane works, we can't just make an average between four or five readings and send the result. But certainly it can be done using more sofisticated mathematics. I just don't know how Smile

Reply
#8

(20-02-2015, 21:32)Werk_AG Wrote:  Yes, it does.

Currently the TX unit samples the wind direction every 5 seconds, and sends it, nothing more..
Due to the way that Fine Offset wind vane works, we can't just make an average between four or five readings and send the result. But certainly it can be done using more sofisticated mathematics. I just don't know how Smile

I think a new thread on the software forum would be the place to go into more detail (theory and code) on how I've been calculating the averages...
Reply
#9

Fell free to create the new thread .

Reply
#10

(22-02-2015, 20:27)Werk_AG Wrote:  Fell free to create the new thread .

Hi
Would it be possible to use a wind sensor such as the WSD1 from Environmental Measurements Ltd
This uses a continuously rotating potentiometer which could be connected as a voltage divider circuit.
http://www.emltd.net/products/wind/wind-...on-sensors
Thanks
Reply




Users browsing this thread: 1 Guest(s)