Right, so it's mounted and ready to go. We have total weight, it's leveled (so it can tell if it's climbing or descending) and it seems to be sensing wind speed OK. Now we need to calculate the aerodynamic drag and the friction between road and tyre. Now we can estimate this pretty well, but the "coast" test will actually time your deceleration run - ie measure the drag induced by you and your bike on the road. So out we went, ibike and I, on our Look KG76 for test number 1.
It's harder to find a flat, smooth quarter-mile of road than you'd think. Slightly uphill is good, downhill is bad, bad, bad as it distorts the results. So naturally I chose a road that looked flat-to-uphill but actually wasn't, so I got some fantastic results. Fantastic as in no way could it be real.
Look at this: 1459W, man! Beat that!
Oh well, back to the "coast" test. In fact I kept finding roads with dips, declines, potholes, corners and really smooth fast bits. Which raised a question or 2 in my mind. Like how accurate is it when road conditions vary? And how is it calculating wind speed, let alone direction? I guess it's a straight subtraction of total airflow "in" minus forward velocity, and angle isn't relevant, but the final figures look odd... anyway, wind aside, if I calibrate on a smooth fast road presumably I'll get errors unless I only ride on that exact same smooth fast road... so are the errors small enough that it won't matter? Or when I get to new territory should I re-calibrate?
So I chose to retest a few times (OK, about 5 times) and compare. Firstly the ibike captured the whole thing, despite my many, many retests - which is good - and secondly I never again got the sort of fantastic result I got with the first coast test. Instead of 1459W I was now in the region of 600-1000W tops (I was getting tired, too, after countless sprints!!). So which 'coastdown' is correct? Hmmm.
Now if you look at the screenshot on the left (of the ibike software) you will see a few strange things. Firstly it shows maximum Watts on this same ride as 1495, yet the LCD display showed a maximum of 1459! Oddly similar but dyslexically different. On the right of the pic you will see the figures for a precise moment in my ride. Using those figures (28kmh wind speed, 8.9% slope etc) you could indeed calculate that a 72 kg rider at 47.5kmh on that slope is indeed putting out about 2100W, not the 'fantastic' figure of 1459/95. But to me, fallible old me, I could have sworn the road was (a) almost flat and (b) that there was little if any wind.
If you take me at my word, that it was a flat road with nil wind then Kreuzotter calculates it as 715W. I'm happy with that. So - assuming a multiply-by-2 glitch occurred - there's an error of more than a percent or 2, isn't there? Hence my scepticism and need to rerun this "coastdown" test until it checks out against 'expectations'. Or am I too harsh? Did the mostly flat road dip and climb suddenly for an instant, or did I pull up on the bars, lifting the front wheel a tad (I was sprinting, after all)... and maybe the wind suddenly gusted? No, I reckon it was a glitch.
So, I think I've got the "coast" test figured out and I'll keep it "as is" for now until I see questionable figures. Certainly my max power figures have come back to earth. Some doubt remains over what happens if you ride very different terrain, but it's easy enough to re-do the coast setup if on super-smooth or super-rough road. Perhaps do the coast test just before a race on a new circuit? Certainly do it if you swap bikes, but that's a test I'm going to do later, just to see what the diffence may be... I suspect it'll be neglible, though, unless my race wheels really are that much better!