Hi and thanks for your advice, appreciate it...
1) i dont think it sounds like any device would solve as it seems less a device artefact and more a protocol consideration. Inherently, an out and back calibration assumes constant wind. That's virtually never the case but may produce ok results under that assumption when very little wind or where wind is constant... Whence why I was previously asking in this thread what level of cal fac reproducibility the gizmo cal fac gave in windy conditions and up to what speed of wind this was acceptable (if it had been established).. An initial reply I got was that it didn't require a windless day (a big range..), but then nothing after that, so it may be every device, it may not be, I don't know... Haven't heard how well the notio or other devices perform during cal fac in windy days, and what speed of wind it gets to before data becomes unreliable...
2) I would question whether you need to recalibrate as much as you are. For example it would seem to me that a material calibration factor changing as a result of minor setup change is less likely than a "change" observed but not real) in calibration factor because wind was not constant. That then opens up a whole other level of subjectivity, which using a cal fac (if accurate and precise) would avoid.. Does changing extension angle from 30 to 20 degrees give a genuine different cal fac? Does 30 to 10, to level? Changing front wheel/tyre? Surely the whole point is not to guess which set-up change may or may not affect the established baseline cal fac? Indeed Aerosensor appear to have now moved from "only apply a new cal fac if doing a major change" which was of course subjective, to now "one should now always correct (CdA) for current cal fac.."
2. If you are going to calibrate, doing multiple calibrations would help to average out variability, so if you are doing a simple out and back perhaps you can do more runs and take average of the calibration factors to smooth out the variations. That is what I am doing already, whence why I know the %error for the inter-runs can be 2,3,4% in non-windless days (Aerosensor guys say use the new cal fac only if inter-run %error is 1% or less...)
3. In my view you don't necessarily have to do separate calibration runs to testing runs. E.g suppose you do 4 out and back runs on setup A, and 4 out and backs on setup B. You could test using the same calibration factor and then post testing analysis check whether the implied average wind suggests that there seems to be a true difference in calibration factor - e.g. a clear wind bias towards positive or negative. in setup B might imply a difference in calibration factor. Aye, again that's what I am doing, and thought I mentioned above, the data can be post-corrected....
You could also check you Cda out Cda back - if calibration factor is set correctly you should see the value being equal, if you see a bias towards one direction it may suggest something is off (noting that this could also happen for other reasons e.g. CRR setting is not correct one direction has a lot more elevation than the other). Aye, looked at that too, and routinely within ~1%.... As mentioned the CdA estimation is very good in the wind conditions I've tested in, it's the cal fac.....
4. I would explore sensor setup/position/location to try make cal factor more "independent" of the setups you want to test.
5. Are there ways to make the calibration more independent of wind? Some ideas - the faster the ground speed, the smaller the wind component will be of total airspeed so would doing your calibration runs at a faster speed for example potentially help?[/quote] True to an extent maybe, but the legs have the power that the legs have (which ain't much..) and I doubt I could go so much faster as to bring 3-4% reproducibility error down to <1%... But worth a go and see how much, if any, difference it makes..
1) i dont think it sounds like any device would solve as it seems less a device artefact and more a protocol consideration. Inherently, an out and back calibration assumes constant wind. That's virtually never the case but may produce ok results under that assumption when very little wind or where wind is constant... Whence why I was previously asking in this thread what level of cal fac reproducibility the gizmo cal fac gave in windy conditions and up to what speed of wind this was acceptable (if it had been established).. An initial reply I got was that it didn't require a windless day (a big range..), but then nothing after that, so it may be every device, it may not be, I don't know... Haven't heard how well the notio or other devices perform during cal fac in windy days, and what speed of wind it gets to before data becomes unreliable...
2) I would question whether you need to recalibrate as much as you are. For example it would seem to me that a material calibration factor changing as a result of minor setup change is less likely than a "change" observed but not real) in calibration factor because wind was not constant. That then opens up a whole other level of subjectivity, which using a cal fac (if accurate and precise) would avoid.. Does changing extension angle from 30 to 20 degrees give a genuine different cal fac? Does 30 to 10, to level? Changing front wheel/tyre? Surely the whole point is not to guess which set-up change may or may not affect the established baseline cal fac? Indeed Aerosensor appear to have now moved from "only apply a new cal fac if doing a major change" which was of course subjective, to now "one should now always correct (CdA) for current cal fac.."
2. If you are going to calibrate, doing multiple calibrations would help to average out variability, so if you are doing a simple out and back perhaps you can do more runs and take average of the calibration factors to smooth out the variations. That is what I am doing already, whence why I know the %error for the inter-runs can be 2,3,4% in non-windless days (Aerosensor guys say use the new cal fac only if inter-run %error is 1% or less...)
3. In my view you don't necessarily have to do separate calibration runs to testing runs. E.g suppose you do 4 out and back runs on setup A, and 4 out and backs on setup B. You could test using the same calibration factor and then post testing analysis check whether the implied average wind suggests that there seems to be a true difference in calibration factor - e.g. a clear wind bias towards positive or negative. in setup B might imply a difference in calibration factor. Aye, again that's what I am doing, and thought I mentioned above, the data can be post-corrected....
You could also check you Cda out Cda back - if calibration factor is set correctly you should see the value being equal, if you see a bias towards one direction it may suggest something is off (noting that this could also happen for other reasons e.g. CRR setting is not correct one direction has a lot more elevation than the other). Aye, looked at that too, and routinely within ~1%.... As mentioned the CdA estimation is very good in the wind conditions I've tested in, it's the cal fac.....
4. I would explore sensor setup/position/location to try make cal factor more "independent" of the setups you want to test.
5. Are there ways to make the calibration more independent of wind? Some ideas - the faster the ground speed, the smaller the wind component will be of total airspeed so would doing your calibration runs at a faster speed for example potentially help?[/quote] True to an extent maybe, but the legs have the power that the legs have (which ain't much..) and I doubt I could go so much faster as to bring 3-4% reproducibility error down to <1%... But worth a go and see how much, if any, difference it makes..
Last edited by:
Fastasasloth: Dec 22, 23 5:31