Author Topic: Anemometer calibration comparator  (Read 1525 times)

0 Members and 1 Guest are viewing this topic.

Offline SLOweather

  • Administrator
  • Forecaster
  • *****
  • Posts: 3452
    • http://www.sloweather.com
Anemometer calibration comparator
« on: January 25, 2016, 05:27:23 PM »
I spoked off the code for my Rugid RUG9 rain gauge comparator to add some code for comparing cup anemometers. My ulterior motive is to put a Davis anemometer side by side with my copper "add-a-mometer" that goes on top of the gazebo/belvedere so i cab get some sort of a calibration factor for it.

In the Rugid, I programmed a 2.25 second delay timer (Davis's factor) and count pulses from the Davis anny while the timer is running. That's then latched for display between timer triggers.

And duplicated the code for the copper anny, only putting in a variable for the timer so I can try to match MPH on the fly.

The copper one has a much bigger radius, and therefore turns more slowly. I think I might steal Old Tele Man's idea from another thread, and add 2 or 3 magnets to the add-a-mometer to decrease the time between clicks for a given wind speed.
« Last Edit: January 31, 2016, 12:45:20 PM by SLOweather »

Offline SLOweather

  • Administrator
  • Forecaster
  • *****
  • Posts: 3452
    • http://www.sloweather.com
Re: Anemometer calibration comparator
« Reply #1 on: January 26, 2016, 02:26:54 PM »
Yesterday I temporarily installed a reed switch on the Wind column on the gazebo and tested it with a DMM..



At that distance it seems to have about a 60 dwell angle.

I dragged the RUG9 out to the driveway this morning, fished a Davis anemometer out of the box and clamped it to the center column, and hooked it all up.



The Davis counter timer is set for 2.5 seconds. The gazebo add-a-mometer counter timer is set for 5 seconds.



I used a leaf blower to run some tests. It works! After a few runs on both low and high speed, it seems like the calibration factor is about 2.

The highest real speed I saw from the Davis was around 83. I saw one run of 118, but that may have been due to switch bounce.

So, I can either set the gazebo sample time to 10 seconds (too long for me), or add a second magnet 180 away from the first one.   

I'll probably do both, and use the sample time to fine tune it over time.

Now then, where did my theoretical calcs go for this....

Offline SLOweather

  • Administrator
  • Forecaster
  • *****
  • Posts: 3452
    • http://www.sloweather.com
Re: Anemometer calibration comparator
« Reply #2 on: January 27, 2016, 04:04:51 PM »
Deciphering my calculations from a couple of weeks ago...

I'm using Davis's anemometer as a reference. It's the only type of manufactured one that I have, and their docs are pretty good on it.

The radius to the outside of the cups is 3". Circumference of the swept circle is then 2r(pi), or 18.85inches or 1.57 feet.

5280 feet in a mile/1.57 ft per rev =  3361 revs per mile in theory.

Davis says the anemometer rotates 1600 times per mile, or a correction factor of  0.476 from theoretical.

My copper anemometer has a radius of 9 inches, or circumference of 56.55 inches or 4.71 feet or 1120 revs theoretical per mile.

If I apply the Davis correction factor as a starting point for my copper anny, I should get 533 revs per 1 mph in an hour, or 533/60 8.89 revs per minute for 1 mph, or 1 rev every 6.75 seconds per 1 MPH of speed.

The hard part is getting decent comparison figures in the real world, due to the different sample times and the variability of the wind. Hmmm, I could try setting the sample times equal and to something big, like an hour, and then compare bulk revs for the hour from each device. That might be interesting.

Offline SLOweather

  • Administrator
  • Forecaster
  • *****
  • Posts: 3452
    • http://www.sloweather.com
Re: Anemometer calibration comparator
« Reply #3 on: January 28, 2016, 05:37:10 PM »
A little "real world" data finally...

About 11:45 AM PST, a weak front blew in. In 5 or 10 minutes the temperature dropped 10F and the wind went from 0 to 20 with gusts to 25 on the backyard anemometer.

Out front, the Davis and copper anemometers tracked each other within 1 MPH. The peak I observed out there was 18/19. Being lower than the backyard station isn't surprising considering the semi-sheltered location and being 4-5' off the ground. Mostly they matched or the copper one lagged the Davis by 1 MPH. A couple of times it was ahead by 1.
 I attribute this mainly to the copper one having more rotational mass and inertia. It's slower to respond to speed changes both + and -. It does seem to be more sensitive at really low winds.

Offline SLOweather

  • Administrator
  • Forecaster
  • *****
  • Posts: 3452
    • http://www.sloweather.com
Re: Anemometer calibration comparator
« Reply #4 on: January 28, 2016, 05:51:58 PM »
Here's a 30 second video of the 2 annys running in the wind. You can see the rotational speed difference, and that I did the 2 magnet mod to the copper one.

BTW, if you think your anemometer spins fast, paint one cup a different color. :)

https://www.youtube.com/watch?v=bf9UlX3SUsU&feature=youtu.be

Offline SLOweather

  • Administrator
  • Forecaster
  • *****
  • Posts: 3452
    • http://www.sloweather.com
Re: Anemometer calibration comparator
« Reply #5 on: February 17, 2016, 03:10:37 PM »
In amongst other projects, I've been playing with the copper anemometer. Right now, I have 4 magnets on it, spaced at 90 degrees apart. This has allowed me to reduce the sample period from 5 seconds down to 2.5 seconds. At that period, the unit very closely matches the Davis, usually equal or within 1 mph in the 20 mph range. I'm chalking the difference up to the greater rotational mass of the copper one. And I'm not yet sure if the switch has time to open between magnet passes at higher speeds. In order to maybe mitigate that, I reverse the polarity of adjacent magnets.

I can't really put any more magnets on to reduce the sample period even more. Where it is, with 4 magnets and 2.5 seconds 1 click about equals 1 mph. So I've been thinking about ways to get the sample period down even farther.

Texas Electronics makes an anny with a 20 slot chopper wheel and an optical detector. I could try that, but it really messes with my current design.

However, there is one other way to do this. Right now I'm measuring switch closures per unit time, which is directly proportional to the wind speed. I could, instead, count some sort of clock pulses between switch closures. That would be inversely proportional to the wind speed, but give a way to measure very slow rotations of the cups. 

To make this easier to cipher out, I'll use the Davis 1 rev in 2.25 seconds = 1 mph as an example.

If I used a 0.01 second clock pulse, 225 pulses between clicks would equal 1 mph. 450 pulses would equal 0.5 mph, while 112 to 113 pulses between clicks would be 2 mph.

That mode starts to be less useful at higher wind speeds. 22 to 23 pulses would be 10 mph, 4 to 5 pulses wold be 50 mph, so resolution would drop off rapidly.

I could use a higher frequency clock. However, a good compromise might be to use both methods, counting clock pulses between clicks below, say 5 mph, and counting clicks per 2.25 seconds above 5 mph. In reality, I'd program it to run both ways all the time, and use logic to switch which reading is output.


Offline SLOweather

  • Administrator
  • Forecaster
  • *****
  • Posts: 3452
    • http://www.sloweather.com
Re: Anemometer calibration comparator
« Reply #6 on: May 06, 2016, 04:15:47 PM »
I've been a bit derailed on projects the last few weeks due to assuming some family cares duties.

However, one thing I've been able to do is gather some long term comparison data between the Davis anemometer and the copper one I built. To refresh, the copper one has 4 magnets taped to it to approximate the same sample period of the Davis. The controller is programmed to save and display the max value of each anny until I manually reset it. Over the last few weeks, I recorded the 2 speeds at random times depending on the outdoor wind conditions.

Davis   Gazebo   Correction
 32   41      0.780487805
 4   5      0.8
 25   30      0.833333333
 42   55      0.763636364
 13   15      0.866666667
 21   24      0.875
 23   27      0.851851852
 15   18      0.833333333
 16   19      0.842105263
         
Average      0.827379402



I put the data into Excel and graphed it and added a trendline.



First of all I was pleased to see that it's a linear relationship. While I had no reason to doubt that it was, a curve would have really messed with my head.

I did an average of the corrections. I looks like multiplying the copper anny by 0.83 will get a corrected wind speed. The better thing to do is probably to adjust the current sample period programming in the controller appropriately.

Offline SLOweather

  • Administrator
  • Forecaster
  • *****
  • Posts: 3452
    • http://www.sloweather.com
Re: Anemometer calibration comparator
« Reply #7 on: May 09, 2016, 03:08:35 PM »
Well, I thought it looked slightly concave as well. :) However, both devices report only in integer MPH, and the dynamics of each one are significantly different. The Davis has less mass and a smaller radius, so it responds faster, and the copper one has larger cup sizes, a larger radius and way more mass, thus lower threshold, more inertia/rotational energy and slower response.

I think I'll program up a data logger and trend the 2 on the same plot for a few days and see how track.

Excel says the Y intercept is -2.14
Slope is 1.31594161.
 That negative intercept may indeed make it a curve. The one obvious data point I haven't added is 0,0. Duh.
« Last Edit: May 09, 2016, 03:16:33 PM by SLOweather »

 

anything