AeroPod, PowerPod and Newton

I have not yet been able to obtain any credible data with the Aeropod. It's not for lack of trying, I just don't seem to be able to get any degree of reproducibility between trials.
I'm willing to accept that others believe they have been able to generate meaningful data.
One possibility is that I'm doing it wrong, but if a user with a graduate degree in engineering can't make it work then something is amiss.
That makes me wonder whether my Aeropod is functioning properly.
Is there any way to test it's functioning?
I did try checking the wind speed reported by the Connect iQ app using my Kestrel device as a reference, trying to note the wind speed reported by the Kestrel while holding it in the correct position and walking the bike around in circles. The iQ app showed a wind speed of zero the whole time, the Kestrel showed 1-3 mph depending on the direction the bike was facing. The iQ app DID show slope that varied as I moved the bike around but the wind speed never budged. The slope values showed a good bit of lag but they were believable.
I'm not sure that was a very reliable test.
Is there any reasonably reliable way to test the Aeropod and see whether it is operating "in spec"?

Please see my comments to your previous post.

I think what is happening is that Isaac CdA Analysis is being pushed beyond its current capabilities (we have more current algorithms in AeroPod than in Isaac).

Your "walk around in circle"s test won't work; AP intentionally shows 0 wind speed if bike speed is low (note that low absolute wind speeds have air pressures of just a few pascals).

velocomp writes:
"AP intentionally shows 0 wind speed if bike speed is low "
I have noticed this and really don't like it. If I'm stopped for some reason and the wind is blowing in my face AP should be showing the wind speed in the garmin IQ data field.
Is there any chance of this being changed?

Velocomp wrote: ↑Tue May 21, 2019 3:37 pm I think what is happening is that Isaac CdA Analysis is being pushed beyond its current capabilities (we have more current algorithms in AeroPod than in Isaac).

I hope that addressing the MANY deficiencies in Isaac is on the agenda for the future? I would have thought that since Isaac has access to a better processor and does not have constraints imposed by real-time processing it should be able to do a MUCH more sophisticated analysis.

Velocomp wrote: ↑Tue May 21, 2019 3:37 pm Your "walk around in circle"s test won't work; AP intentionally shows 0 wind speed if bike speed is low (note that low absolute wind speeds have air pressures of just a few pascals).

So it appears that there is no straightforward way to verify correct functioning, Aeropod is basically a "black box" from a user's perspective?
Personally I would prefer a way to operationalize a "trust but verify" approach.

ted.kelly wrote: ↑Tue May 21, 2019 8:32 pm velocomp writes:
"AP intentionally shows 0 wind speed if bike speed is low "
I have noticed this and really don't like it. If I'm stopped for some reason and the wind is blowing in my face AP should be showing the wind speed in the garmin IQ data field.
Is there any chance of this being changed?

AeroPod measures air pressure directly, not wind speed. Air pressure of near-zero wind speeds is only a few pascals (look it up). The measurement of near-zero wind speed is in the range of the native noise of the air pressure sensor, so we don't display it.

NOTE: when moving on the bike, the ground wind speed is added to the bike-caused wind speed. This is a pressure much different from zero, so the measurement is quite accurate because the signal to noise ratio is very high.

BillOsler wrote: ↑Tue May 21, 2019 9:30 pm

Velocomp wrote: ↑Tue May 21, 2019 3:37 pm I think what is happening is that Isaac CdA Analysis is being pushed beyond its current capabilities (we have more current algorithms in AeroPod than in Isaac).

I hope that addressing the MANY deficiencies in Isaac is on the agenda for the future? I would have thought that since Isaac has access to a better processor and does not have constraints imposed by real-time processing it should be able to do a MUCH more sophisticated analysis.

Velocomp wrote: ↑Tue May 21, 2019 3:37 pm Your "walk around in circle"s test won't work; AP intentionally shows 0 wind speed if bike speed is low (note that low absolute wind speeds have air pressures of just a few pascals).

So it appears that there is no straightforward way to verify correct functioning, Aeropod is basically a "black box" from a user's perspective?
Personally I would prefer a way to operationalize a "trust but verify" approach.

Your assumption about processor power is wrong. AP on-board processor has access to the raw data and works well; Isaac must re-process exported data.

Yes, it is a bit of a black box, though we have been very open about the principles of the box. Contrast this open-book approach to any other company talking about aero sensors; they do not provide ANY information about their methods.


See the last two posts of this thread for how-to-operationalize:

https://velocompforum.com/viewtopic.php?f=98&t=5107

Velocomp wrote: ↑Wed May 22, 2019 6:59 pm
Your assumption about processor power is wrong. AP on-board processor has access to the raw data and works well; Isaac must re-process exported data.

There are any number of posts in the forum that appear to suggest that the IBR file contains the raw data collected during the ride, though some of those posts relate to the Newton or Powerpod.
Does the IBR file for the Aeropod not preserve all of that data?

BillOsler wrote: ↑Wed May 22, 2019 10:35 pm

Velocomp wrote: ↑Wed May 22, 2019 6:59 pm
Your assumption about processor power is wrong. AP on-board processor has access to the raw data and works well; Isaac must re-process exported data.

There are any number of posts in the forum that appear to suggest that the IBR file contains the raw data collected during the ride, though some of those posts relate to the Newton or Powerpod.
Does the IBR file for the Aeropod not preserve all of that data?

Some of the direct-feed sensor data is sampled at 800x per second, and processed at that rate within AP. .ibr files cannot and do not retain all that raw data