Wednesday, March 14, 2007

Can O' Worms (Part IV)


So, I was unable to stop myself yesterday and I continued on with some of my investigations. Yesterday's ride was done with all PMs in place and I just rode around on the CT however I felt like I wanted to at the time. While letting my mind wander, I realized that the reason the SRM creeps higher on me is due to heat. That is, the BB itself heats up during a trainer ride. So, how much? And, can I correct for it?




Well, the how much is certainly negligable. We're taking about a 3.5w drift over the course of twenty minutes. That's not enough to effect anyone's training.

As for correcting it? Well, yes. You can see by the second graph that you can create a simple linear regression based upon the torque values calculated by WKO+. If you then recalculate the watts based upon torque (WKO+ calculates torque for you whenever both Watts and Cadence are available), the difference goes away.


Miraculously, when you compare the predicted SRM watts to the Ergomo (used only as a comparison here), you see the two line up very well in terms of consistency. Yes. The SRM is reading higher. But, that's not my point as I can correct for that by adjusting the Ergomo's K-Factor upwards.


This also means that SRM users should follow the advice provided by the manufacturer. That is, perform the Zero Offset (ZO) anytime the ambient temperature changes as well as at the beginning of your ride. And, I would add that you also consider performing the ZO about 20 or 30 minutes into your workout as well.

Monday, March 12, 2007

Can O' Worms (Part III)

Last night, I found what I was looking for.... The Computrainer has a Potentiometer screw that allows for adjustment of the resistance (thus allowing someone to raise or lower the reported watts). Now, be warned, if you move the screw and return the load generator to RacerMate, you will be charged for recalibration against their dyno. Fair enough.

The truth is, there probably isn't any reason to turn the screw anyway unless you're bothered by looking at results compared against another Power Meter and even then as you're about to read, it still might not matter.

So, this morning, I rode using the SRM, Ergomo and the CT. I didn't follow any protocol. I just rode the bike. In my haste, I forgot to perform the zero offset on the SRM. Well, I figured that since the bike hadn't been removed from the CT since yesterday, and the temperature is stable ... well, what would have changed to give me unexpected results? I did ride for 10 minutes in Ergo mode and then performed the RRC.

The results? Well, for some reason, the SRM was 12w higher than the Ergomo - the result from not performing the zero offset I presume. But, the slope still showed a positive bias meaning that I probably still need to lower it some more.


What about the CT? Well, the good news is that the slope of the differences between the CT and Ergomo was -0.004. In my mind, that's closer than anyone could hope. And, I feel vindicated in that the differences that I sensed between my CT and Ergomo were in fact real. More importantly, it shows that the differences are steady. The bad news is that my CT now reads (even after adjusting the potentiometer screw all the way left (raise watts) a clear 36-38w too low.

But, wait. Ergomo's and SRM's are supposed to read higher than a CT. In theory, drivetrain losses (chain, derailuer, hub, spokes and tires) will lose some power. So, the next question is exactly how much should I have expected to lose and are there things that I can/should do to the bike to salvage some of that loss?

I don't know the answer to those questions yet. If/When I find out, I'll write about it.

Sunday, March 11, 2007

Can O' Worms (Part II)

Yesterday, I set out to adjust both my SRM and Ergomo Powermeters so that they both at least showed similar readings. Without having the benefit of first calibrating my SRM externally (using the Mfgr protocol), I set out to just "eyeball" the differences.

The SRM allows for the adjustment of Slope - that is the response to torque. When you do the zero test, you are setting the Offset (or Y-Intercept). All of this comes straight from your Intro to Statistics class.

The Ergomo allows for the adjustment of K-Factor. Unfortunately, I don't know what K-Factor is. I think it's actually closer to being a Y-Intercept.

Yesterday's testing showed that my SRM consistently tracked higher across a range of wattages. However, the differences grew as watts increased. Thus, the slopes of the two devices are/were different. So, I began to adjust the SRM's slope in the hopes of finding something where the differences tracked evenly across various wattage ranges. This morning I continued this approach until I reached a trend in differences that was very close to zero.


However, this still left the ergomo consistently reading about 10w lower than the SRM. So, I began to adjust the K-Factor on the Ergomo by small amounts and repeat the protocol.



You can see by the graph above that I think I ended up being pretty successful. The difference between the two is 0.5 - 1.8w depending upon how you look at the data.


Another question, though, that I wanted to answer, was how much does a simple change in K-Factor effect watts? Well, I was able to gather a small set of data and do a simple analysis. The end result was that for me, increasing the K-Factor by 1 unit results in an increase of 3.6-4.2w for the cadence zones that I'm interested in (80-100 rpm). Obviously my sample set is very small, so your mileage may vary.

So, what have I really learned from all of this? Well, not much except that two of my powermeters now read very closely. Unfortunately, I don't have the means this moment to say that they are both reading "correctly". Nor am I all that worried about it either. Funny thing is, they're both on the same bike. So, why does it really matter? :P

I did notice that the SRM is quicker to record/capture spikes that occur when a rider quickly increases cadence. And, some of the differences are clearly due to the Ergomo's dependence upon the left leg's input. All in all, both of these seem inconsequential to anyone doing Time Trial riding. If you ride the track, maybe you'd feel differently.

One feature that I'm sure that everyone with multiple Powermeters would crave is having the ability to just raise or lower the displayed/recorded data if for no other reason than giving us a warm and fuzzy feeling when we switch bikes. Unfortunately, there isn't consistency among the manufacturers. So, this becomes such a difficult task (whether you own different brands of PMs or not).

Now, I've still got two more PM's to try and bring into sync. So, wish me luck!

Saturday, March 10, 2007

Can O' Worms

One of the benefits of my job is that the very best training equipment and devices often find their way into my world. In short, I get lots of toys. One of the worst things is that I have to use them all within my own little world of training too. Of course, I'm not actually complaining about getting to play with all these power meters and HRMs and GPSs. But, sometimes the data that I get from them is well .... how do I say it nicely? .... "Suspect"? No. That's not correct. Sometimes the data from one makes me question the data received from another.

Stop reading right now if you think this is going to end up being an article that claims one PM is better or more accurate than another.

We've all seen the posting where one guy claims his original power meter is, has, and will always be the most correct and then he laments how the new one that he just purchased reads way too high or too low. I've been there. It's frustrating. I'm pretty sure, though, that 90% of the time, the poster is either just plain wrong or, more likely, he's trying to ask the right question, but doing so the wrong way. What he really wants is to know how to get the new meter to display and record the same watts as the old workhorse so that he can count on some level of consistency. Amen to that.

Well, our lamenting forum poster often has one disadvantage. That is, he's got PM's on different bikes, usually with different positions (road vs. TT, road vs. MTB, etc.). Hmmm, well.... I'm not really sure what to say or do about that. But, I might have something in the near future.

I have sort a unique situation in that my TT bike sits on a computrainer and has both an SRM Pro and an Ergomo Pro power meter on it. Yep. 2 PM's simultaneously helping me crank out the horsepower (how come I can't go twice or thrice as fast?). And, like our lamenting forum poster, I too become distracted (for lack of a better word) at having not two, but three numbers in disagreement in front of my face when I'm training.

So, today I decided to take matters into my own hands. My goal was to get all three to essentially agree. But, how do I do that? Folks, I'm just winging it.... I'm lost this minute on what to do about the Computrainer. But, I know that both the SRM and the Ergomo allow a bit of tinkering. So, that's where I'm heading first....

Given the 3 devices, ideally I should start by externally calibrating the SRM using weights and a calculator. Or, some might suggest using the CT as the benchmark. Unfortunately for the CT, it's reading 40-60 watts lower than both the Ergomo and the SRM. So, it's "Wrong". :) Next, I'm not completely convinced that SRM's published calibration routine is completely correct (especially if you expect someone like me to follow along). And, more importantly, I have lots and lots of data based upon my ergomo. So, ideally, I want that one to more or less become the benchmark.

During my exploits today, I was happily changing slopes and offsets and k-factors on both units. At first, I tried to eyeball the numbers and then I eventually broke out the spreadsheets. In the end, I think I've got my Ergomo Pro and SRM to be really close. Close enough until I get more time to narrow it all down even further, that is. But, in the end, it all of a sudden struck me that the best results ended up coming from my PM's when my Ergomo was restored to it's original K-Factor setting (the one that I've been using for the past year or so). So, I feel a little vindicated even if I haven't yet "proved" to myself that the slope adjustment on the SRM *should* have been changed.

In the end, however, I realized that everyone with multiple PM's needs more data. So, this task of "calibrating" is really turning into a session of "Hey, what happens if I do this?" and eventually someone can create a handbook of "how to get your multiple PM's to agree".

So, what is the result of today's exercise? See for yourself. This graph is just a very small sample. But, it shows that both the ergomo and the srm read very closely. You can also see that slope adjustments on the SRM have some effect. Note how the dashed-green line moves slightly upwards from lower watts to higher watts (meaning that these settings give the SRM a bias for under-reporting lower efforts and over-reporting higher efforts). This indicates that the SRM is still giving a bit too much value to higher torque. (My SRM is three or four years old and hasn't ever been back for factory testing of the slope. It's slope is marked as 21.0. But, for this testing I had dropped it down to 14.0 and should still probably lower it by 1.0 to 13.0 instead).

One thing that becomes very clear from this exercise is that both the SRM and Ergomo Pro track together very, very nicely. But, I have no idea as to which one is more correct. +/-3 watts is close enough for my needs.

Also, this analysis is done on the downloaded information and not the subjective measure of the displayed watts. Both the SRM and the Ergomo display differently than what is recorded. To help equalize things, I had set the Ergomo to update it's display every pedal revolution. I believe that the SRM updates it's display every 1+ seconds. So, close enough. And, before analyzing this final set of data, I did have the feeling that the ergomo was reading a bit higher in the low watt ranges and a bit under on the higher watt ranges.

Next time..... I am hoping to figure out something for bringing the CT into the mix. Or, maybe a little experiment to show the effects of changing K-Factor on the Ergomo. Oh, you better believe that I'm looking for a PT too.