[TowerTalk] Wired Anemometer for Tower?

David Gilbert xdavid at cis-broadband.com
Wed Apr 30 03:10:52 EDT 2014


That initial degree of uncertainty only exists for 12. 5 milliseconds at 
25 mph (80 pulses per second).  How is that even the least bit relevant 
here??  No way a tower or its antennas respond mechanically to anything 
even remotely that brief.  After the first few pulses, the uncertainty 
rapidly decreases as the number of measured intervals accumulate and the 
measurement accuracy becomes dependent upon the linearity of the 
anemometer (rated at +/-4%) plus whatever inertial effects create 
acceleration/deceleration delay for gusts.  I would find it very 
difficult to believe that response times less than several hundred 
milliseconds would represent anything other than pure scientific 
curiosity, and there isn't a shred of justification for your original 
contention that this kind of application requires expensive equipment.  
It simply does not.

Besides, the positional accuracy for eight pulses per revolution is not 
+/- 12.5 % ... it's +/- 6.25 %.

On top of that, in real life the anemometer would have been spinning all 
along, happily generating pulses that were being measured and averaged, 
and the only way that the full +/- 6.25% inaccuracy would be of concern 
for ANY 12.5 millisecond window is if you erroneously thought that wind 
gusts could spike with an infinite ramp.  They don't ... they have their 
own time constant and it isn't anywhere close to being in the single 
digit range of milliseconds.

Dave   AB7E

On 4/29/2014 9:08 PM, Al Kozakiewicz wrote:

This is the key thing that misses the entire point.  It's not the accuracy of the time measurement that's the issue.  It is the uncertainty of when position transitions are crossed.  With either electromagnetic or optical sensors, that uncertainty is 1/8th of a revolution with an 8 ppr transducer.  The fact that you can measure the time between "pulses" accurately is not disputed.  The issue is that the rotational distance (hence velocity) 8 pulses represents is known only to +/- 12.5%.

Again, no practical matter since no one would try to measure wind speed with such a small sample and expect high accuracy.  Nor can I think of a reason why you would need to in this application.


TowerTalk mailing list
TowerTalk at contesting.com

More information about the TowerTalk mailing list