Temperature And Wind Offset

[82zman]

I have a new Ibike Aero with FW 3.08. I have been letting my Ibike stabilize at outdoor temps and getting very little wind offset drift. That being said all my rides to this point (this gizmo is new to me) have been in the part of the day where temperatures have stabilized.

Here is the issue. Today it was 44 deg. Shortly after bringing the ibike indoors at 70 deg the offset drifted to 4.2. which results in a 7 mph wind in my house as read by the ibike. I can see this being an issue as on my Saturday morning rides I could easily see a 20 deg increase from start to finish. This is very typical, especially this time of year. So here are my questions

1. Was the new FW supposed to fix the wind offset from drifting over temp?

2. Is this normal behavior and if so what practically can be done to compensate for it?

3. Has it ever been considered that these units should be calibrated over temperature at the factory? The result of such a calibration can be loaded into thermal compensation tables. I work with military electronics and we do this sort of thing on many of our products.

Thanks in advance for your help and suggestions.

[82zman]

I found this post from Coach Boyd on Oct 18th 2008

"I don't know how much I am allowed to divulge, but there has been a wind offset calibration done along all temperatures in the factory. This means if you start off at 40 degrees and it warms up to 80 degrees, your wind offset should change by less than 1.0. Changing wind offsets should no longer be a problem."

Clearly this is not what I am seeing. Here are some data points

I moved the Ibike outdoors

67 deg, Offset=3.4,Reported Wind=8mph
56 deg, Offset=1.0, Reported Wind=4mph
49 deg, offset=0, Reported Wind=0mph (BTW this was my initial cal temp so this is not surprising)

[82zman]

Ok here is this mornings ride. Ibike starting temp 39 deg, offset performed in garage. Ibike finishing temp 56 deg. Offset is 3.5. in the garage and the ibike is reporting a 7mph wind. The ride was about 3 hours.

I did 4 out/back wind scaling tests before and after the ride at the same location. The 4 tests before are right on the money almost 0 mph average. The 4 after the ride average over 1mph. As expected when I applied the ending offset number of 3.5 to the wind offset test performed after the ride using the analyze wind tool the ground winds went to zero, actually that's pretty cool.

Today's ride afforded me no locations along the way to do an offset which I would gladly do given the opportunity. This brings me to another question....... Is there a way I could cover the ibike with a baggie or something if there is no place to duck out of the wind to do an offset?

[Russ]

Hi 82zman,
"another question....... Is there a way I could cover the ibike with a baggie or something if there is no place to duck out of the wind to do an offset?"

A method I use, since I raised two kids, is to take a small baby washcloth with me which is fairly porous but not real thin. I double or triple it up and hold it gently against both the end port and the bottom port cupped in my fingers to block additional wind. I usually try to point into the wind when I do this so that my hand blocks the wind some more. I have had great success with this if the wind isn't real strong, not sure how good it is when the wind is real strong, might still be worth doing. The reason I use the baby rag is because it seems to me that it will not pressurize the ports like plastic might. Perhaps something else that has one side non-pourus and one side pourus might work better in strong wind.

Russ

[82zman]

Hey Russ I will try that. For now what I did is a little experiment and reprocessed the data.

1. Sent the iBike down to 39 deg and zeroed the offset replicating my ride from a day ago. The drift was very repeatable, by the time it got to 56 deg it was 3.7 which is within a few tenths of my ride. I did this twice and the numbers were almost identical. I plotted the drift for every 2 degree increase in order to collect data for drift at specific temperature points.

2. Using Weather Underground I collected the temperature data over time for the day I did the ride. This way at any point in time I knew what the offset was by applying the data from the chart I made in step one.


3. I segmented my full ride data into one hour segments and applied offsets for each segment based on the data I collected.

4. Finally I stitched the segments back together with the corrected wind offset into a full ride.

Here are some observations:

Uncorrected the average watts for 3 hours was 166. Corrected it came to 159. Why so little difference? Its because you must remember that the offset drift is averaged over time. That means we should not freak out (like I did) if we see drift after a ride.

I could have saved time and just "zeroed" the wind out since it was an out/back ride, but I wanted to come up with another method for correcting drift for a ride in the future that may not be an out/back. BTW when I zeroed out the wind in the uncorrected ride it came out to 159 watts! Proof that my method worked.

Look at the data I collected below and notice that over a broad temperature range the iBike wind offset is very stable. Its only the colder temps that give it a problem. On cold days I am going to make it a point to do an out/back ride so I can simply take care off the offset by setting the avg ground wind to zero.

This PM is not perfect and I hope more work can be done to correct the offset drift but in the mean time iBike did not leave us hanging and left us with some wonderful tools to post process the ride data.

Temp Offset
39 0
41 0.5
43 1
45 1.3
47 2
49 2.4
51 2.8
53 3.2
55 3.7
57 4.2
59 4.2
61 4.5
63 4.7
65 4.7
67 5
69 5
71 5.3
73 5.3

[82zman]

ok its me again...just found something that worked. I rode this evening and the offset dropped to -1.6 over a -10 deg drop in temperature which is no big deal. Immediately after my ride I used a gallon freezer bag and covered the ibike BUT MADE SURE to leave an open volume of air between the two ports by making sure that the bag was not near the ports. Its also important to point out that I did not attempt to seal the bag. The Ibike was placed into the wind so no air could enter the open bag. I did the offset into a 8mph wind and then moved it into the garage to recheck it and presto the offset remained at 0. This is what I will be doing on future rides that will vary quite a bit over temperature.

[Velocomp]

For various technical reasons the wind sensor is particularly sensitive to temperature.

The temperature performance of the Gen III has been improved amazingly; if 82zman did the same experiments with older iBikes he'd possibly see offset numbers between -30 and +30...

Folks, please remember that any electronic circuit is sensitive to temperature. And remember that most riders don't start at 39F and finish at 73F.

An advantage the other guys have is that you can't stuff a wheel or crank into a refrigerator and repeat these experiments. Not only won't their products fit, but it would be next to impossible to record the temperature variation data!

That said, we have DFPM data that shows that competitive circuits are even more sensitive to temperature than ours. Said differently: when we're riding in widely-varying temperature situations, we believe our data over theirs.

I like the fact that you are curious and I love your desire for us to achieve perfection. But perfection is generally not attainable, no matter how expensive the product is.

[82zman]

To be clear I am very happy with the iBike. Every part of it from the product itself, the SW, and especially the customer service is first class. My electronic engineering background is getting the better of me and that is why you see all this stuff in these post. I only wish to understand it fully and then after doing so optimize it for best performance.

[mtnsteve]

Have also noted changes in wind offset with changes in ambient air temp, which seems to be more marked at lower temps as previously mentioned. Have also found that if it changes from cloudy to sunny the temperature of the unit changes enough to require resetting the wind offset. Found that calibrating in the shade before riding on a sunny day quickly dropped the wind offset.

Does velocomp have any guidelines on how often the wind sensor should be reset re changes in temp (every 2, 5, or 10 degrees of change) or at what range of temps? Hate to stop too often to try to find a windless location if the results don't make a significant difference.

Have yet to find an ideal way to shield the sensor from any wind.

Thanks in advance for any guidelines.

[82zman]

I am no expert but you may just want to take some time to observe how your units wind offset reacts to temperature. On a day you are not riding set the unit outside (preferably in the morning or evening when the temps swing) for at least 30 minutes to acclimate to temperature then do a wind offset in a protected area such as a garage. After doing this simply monitor the offset versus the reported temperature as displayed by the ibike. This way you will know specifically how your unit reacts to temperature. Always check the offset in wind protected area. As an example I know my unit is very stable over certain temperature ranges (you can see my data in this post) so I will not bother with an offset.

That being said I do a wind offset at every given opportunity. For instance one of my routes has a bathroom large enough for my bike. I drag the bike in there and do an offset. The offset takes so little time this is hardly an inconvenience

Did you read how I have been using a gallon ziplock bag to do the offset? To perfect my technique I simply performed the offset outdoors and then checked it immediately in the garage and I have had great success with this. I am getting good enough at this to where doing it at a stop light is conceivable and the ziplock does not add many grams of weight

Do not take my methods as gospel I am just sharing what works for me. Lastly remember that the offset drift is averaged over time.

I have invested a great deal of time experimenting with this thing and most likely annoyed John by posting the results of some of these experiments. Soon I will be checking the ibike performance under water (Just kidding). But now that I know its "personality" I get very consistent results. If I owned a PT I would not have to do Tilt, Offset Calibrations, etc but I would have spent over 2x the dough. I also wonder what happens to other DFPMs over varying environmental conditions. Since this unit is innovative in its approach to power measurement and because of its small size it has been subjected to many experiments including the ones I ran myself. I wonder what would happen to a DFPM if the same scrutiny was applied?

[Velocomp]

It's getting a little geeky in the room!

As you'll see, this is much ado about only a little bit.

Let's start with an important fact: the iBike does not measure wind speed. It measures air pressure. The difference is important; air pressure is force / area and wind speed is...wind speed (how far do the wind molecules move in an hour).

Air pressure is directly related to opposing force, whereas wind speed depends on air density and temperature. A wind speed of 30 mph at sea level will result in a much higher opposing air pressure than a wind speed of 30 mph at 5000 feet. Why? Because the air is less dense at higher elevation (where the air is likely colder, too), so there are fewer and less energetic air molecules opposing your forward motion.

The number displayed in the wind offset screen is the air pressure in pascals. So, a wind offset of 2.0 (the 0 will be a small number) corresponds to 2.0 pascals.

Now, 2.0 pascals is NOT a lot of air pressure but if you look at your wind speed screen you'll see it corresponds to 4.1 mph. So, the easy conclusion is: "my reported wind speed is off by 4 mph and that's way too much".

Easy, but not true!

In fact, if your wind offset is 2.0, the iBike is reporting 2.0 pascals of air pressure when, in fact, the correct number is 0.0 pascals.

Now, a wind offset of 2.0 would be typical with the temperature compensation algorithms of the Gen III iBike, so let's see what it means in terms of real-life measurements.

At sea level, if you're pedaling at 20.1 mph into zero headwind, the actual opposing wind pressure is 48 pascals. However, with a wind offset of 2.0 your iBike sensor will report an opposing force of 50 pascals. Now, 50 pascals corresponds to an opposing wind of 20.5 mph. So, the wind speed reported by your iBike is "off" by only 0.4 mph--not very much at all!

Since your opposing wind force is measured in pascals, measured wind force is high by (1 - 50/48) * 100% = 4%. But remember, wind force is only one component of total opposing forces, so your watts error will likely be 2% when the other forces (hills, acceleration, and friction) are accounted for.

In the Gen III, wind offset is very stable at temperatures above 70F, generally varying by only 2 pascals between 70F and 120F. So, there's no reason to check offset if you're riding in warm weather.

Where I ride in Florida the wind offset varies by only about 1.0 during my rides. This corresponds to less than a 1% error--insignificant. In fact, I no longer do wind offsets prior to my outdoor rides.

If you're riding at temperatures below 70F, where the temperature varies by 20F or more during your rides, an occasional wind offset check is OK, but in general the Gen III units won't show offset deviations of more than about 3.0 over the normal range of riding conditions. So, even in cold weather you'll get good results if you choose not to perform regular wind offsets.

The test performed in this thread is extremely severe; most cyclists won't get on their bike at 39F, nor be lucky enough to get into 73F weather on the same ride. Note that if the wind offset had been performed at 52F, the deviation at 73F would be about +2.5 pascals, and -2.5 pascals at 39F.

Also, 82zman makes an excellent point about the competition. They don't publish the temperature dependence of their units, and for obvious reasons it is very tough for the rest of us to make independent evaluations! I do recall reading a lab test once that reported a 20% wattage deviation due to temperature changes for another well known (and extremely expensive) PM. Indeed, side-by-side test data from our own DFPMs causes us to conclude that the iBike's temperature dependence is less than the competitive units we've sampled.

So remember: when you have your DFPM and your iBike together and you're riding in varying temperature conditions, it's not at all clear where the measurement errors reside!

Finally, please note that there is NATURAL unit-to-unit deviation in the temperature performance of any electronic product. Your iBike isn't "wrong" if it operates outside the numbers reported in this posting.

[mds]

Thanks John for your very helpful information about wind offset. On my iSport with software version 3.03, I also sometime see wind offsets of 2 or 3pa and corresponding windspeeds of 4 or 6mph at the end of my ride that was cool at the beginning and then warmed up. But I have never seen a big variation of power that you would expect to see with a 4 or 6 mph difference at higher speeds. In other words, a 4 or 6 mph variation at 20mph on the flats would correspond to a big change in power due to the quadratic nature of drag. I have never seen that and have been wondering why. Your explaination gives the reason. Thanks.

On additonal comment: In sunny conditions I sometimes see temperature displayed at maybe 10 degrees higher than ambient. I have a black unit. I am wondering if it would be better to have a white unit so that more of the heat would be reflected away. Does unit color make any difference?

[Velocomp]

Color doesn't make much of a difference. On my Sunday ride here it was sunny and 78F. My white iBike read 91F.

We've tested black iBikes in the Red Sea desert in the middle of the summer and they work just fine, even when internal temperatures are 120F.

[82zman]

That's an excellent description John but it still leaves me with one question.

You say "Now, 2.0 pascals is NOT a lot of air pressure but if you look at your wind speed screen you'll see it corresponds to 4.1 mph. So, the easy conclusion is: "my reported wind speed is off by 4 mph and that's way too much". As you explained this actually computes to about a .4mph error at 20mph (50 vs 48 pascals).

Then the simple question is is why is the wind screen reporting 4.1mph when the offset is only off by 2 pascals? What exactly does this number represent?

[Velocomp]

As stated in my prior post, wind speed and air pressure are not the same thing.

In fact, wind speed increases only as the square root of air pressure. That is, if wind pressure doubles then the wind speed only goes up by a factor of 1.41.

Starting from zero wind, very low air pressures result in seemingly significant wind speeds. In fact, 2 pascals of air pressure will result in a 4.1 mph wind, just as your iBike says. It doesn't take much air pressure to cause a low speed ambient wind; the gentle breezes on a nice day are the result of extremely small pressure differences in the atmosphere.

The incorrect math is to think that, with opposing air moving against you at 20 mph (equivalent to an opposing air pressure of 48 pascals) that an additional 2 pascals of air pressure will cause the opposing wind to increase to 24 mph. That this does not happen is a result of the square root relationship between wind speed and air pressure.

In bicycling terms, we all know that it is much easier on the flats to hold our bike's speed at 10 mph than at 20 mph (double the speed). The reason is that the opposing air pressure builds enormously as bike speed increases. Since air pressure builds as the square of the wind speed, the opposing air pressure is 4x greater at 20 mph than at 10 mph. And here's the kicker: since power is force x speed, in fact it takes 8x as much power to overcome opposing air pressure at 20 mph as at 10 mph!

[82zman]

Thank You! Completely understood now. Duh on my part! 2 pascals computes to 4.1mph. The iBike is so much smarter than me

As usual your support is outstanding!

[rruff]

Finally, please note that there is NATURAL unit-to-unit deviation in the temperature performance of any electronic product. Your iBike isn't "wrong" if it operates outside the numbers reported in this posting.

If the deviation is consistent... and it seems to be for me and 82zman at least... might it be possible in the future to compensate for this via firmware or software? I'm thinking of having a customizable curve that the user could modify. Even though an error of ~.5 mph on wind speed at 20 mph doesn't seem like a lot, it is still ~4% error in power... and it isn't variable, but rather a constant offset that will make the reading for the whole ride be biased by that amount. Unlike the tilt it is difficult to compensate for this after the ride.