Tempest's wind speed accuracy difference between units

[fgentili]

Hi, i own 3 tempest units and i logged the data of the 3 tempest units in a homemade wind tunnel with same hub to test accuracy. Was tested one tempest at time in the same position x.y and z.
And i find out they have huge different values between each other. Values below.

FAN AVERAGE WIND SPEED TESTED WITH WEATHER FLOW HAND ANEMOMETER = 10.7 km/h

Tempest 1 average value in 5 min interval= 10.4 km/h . DV.ST = 0.483
Tempest 2 average value in 5 min interval= 9.2 km/h . DV.ST = 0.234
Tempest 3 average value in 5 min interval= 6.9 km/h . DV.ST = 0.200

Anyone knows why is that?

[peter]

Can you share pictures of your homemade wind tunnel?

[fgentili]

Yes. uploaded.

[sunny]

were all tempest in the same orientation? perhaps make a better windtunnel. I made one a long time ago, also had problems getting good results.

[fgentili]

Same orientation, tested several tries, also in another orientation, they all show huge values differences .

[peter]

I wonder if you are having issues with turbulence. You could try a longer averaging period and see if that works out better? Turbulence should be entirely random, and if you sample for long enough, its effect should average to zero, but otherwise it could be causing the differences you are seeing.

A good check would be to compare successive 5 minute samples from the same Tempest. Assuming the device measurements are stable over that period (which they will be unless the Tempest is broken), then any variability between the 5 minute samples will be due to turbulence in the air flow.

[fgentili]

I’ll perform another test
What is the the calibration process of the units, they have all the same calibration values, or each unit has it’s own calibration values?.

[peter]

There is some info on that here: SKY / Tempest: wind sensor initial self balancing

[fgentili]

Thank you for the reply. I didn’t know that i had to use them in normal mode for a while for self calibrating. i only used one of them in this mode, and looks like is just the one that the meassure is ok.

The calibration happens inside the Tempest unit, the hub unit or online?. I wonder if the values depends on tempest unit and the hub its comes with. Or no problem to use tempest unit with another hub?

Thanks.

[wlbryce]

It is terribly difficult to do testing like you are trying to do. The fan does not produce a laminar flow so the “wind” is no consistent or stable for testing of the device. In order to get good data, you must have a diffuser. A series of long tubes or screen to channel the air and create a laminar flow for the testing. Their purpose is to make the velocities of the wind in the wind tunnel equal by eliminating fast and slow air velocities. As the air passes through the diffuser, the resulting drop in pressure across the diffuser creates constant air speed in the tunnel.

[fgentili]

Thanks for the reply. I undestand that.

In this test i still see repeatability and reliability with same unit ( enough for a first approach). The difference between the units maybe is because two of them has not passed the self calibrating step.

[wlbryce]

With a open fan, it will never be repeatable because the air around the fan, the fan speed, and many other things are always changing even though you don’t perceive it as changing.

[fgentili]

But between 2 and 4% Coefficient of variation for this simple test, don’t you think is good?

[sunny]

rotate it around and see if you still have 4% Standard deviation.

[fgentili]

This could be a good test. Same unit, same run, rotating every a few min. Ill try it tomorrow.

[gizmoev]

The self calibration should be relatively fast. I would just make sure to rotate the Tempest a full 360° in the wind stream before starting the test after it is powered on.

[fgentili]

Thank you for this information. Ill try this!

[sunny]

i think you are right, but the advise from weatherflow in my case was to put it outside for a few weeks instead of the wind tunnel. So I’m not 100% sure.

[marcolinger]

I have a Tempest and a Netatmo. Both are verry close on most of the informations, even on rain despite the fact of the completely different type of method. Wind also is almost the same.

[gizmoev]

Here is what I’m referring to from SKY / Tempest: wind sensor initial self balancing

@WFmarketing

The initial in-field process requires collecting a certain amount of data at all four cardinal points. This initial process runs through a check to ensure all four sonic sensors are balanced properly.