There recently has been a paper published in the International Journal of Sports Medicine claiming to have validated the G-Cog versus a SRM and PowerTap.
Here is the Link to find the paper link
*Due to copyright laws I am unable to post the paper
Below you may read my (George Costa, Owner/Engineer) Synopsis of the Paper. These researchers did not understand how to properly use a G-Cog device. I reached out to both the Journal and the Authors.
This is the response from the IJSM responsible editor.
Here you can read the response of the main author W. Bertucci
Synopsis of the Paper
"Validity and reliability of the G-Cog BMX Powermeter."
Authors: Bertucci W, Crequy S, Chiementin X. Université de Reims Champagne Ardenne, Laboratoire d'Analyse des Contraintes Mécaniques (LACM-DTI, EA 4302 LRC-CEA n° DSM0534), Reims, France. firstname.lastname@example.org
I have summarized what are errors in the authors methods and understanding in using a G-Cog into four key components of the paper.
1. In performing field tests the authors decided to use as verification hardware a SRM and PowerTap (2Hz and 0.8Hz sampling). Apparently these are accepted pieces of hardware because of their "Gold" standard. Yet they write in the paper and I quote.
"For the BMX discipline , only a specific SRM crank power meter has been developed. However, This device samples at a very low frequency (2Hz) for an explosive exercise. In this condition, The PO measurements are interesting only over the entire race (averaging the PO) but are not relevant for the first pedaling cycle after the start which is considered as a key moment of the race."
So in breaking this down they admit that an SRM shouldn't be used unless your averaging across the entire race and it will not provide the data required in the first few pedal cycles. Yet they discredit The G-Cog which can sample at 250Hz because the SRM is the "Gold" standard? But they report in their findings only max power not the average power across the duration? Sounds like a contradiction.
Further issue with how they decided to validate the G-Cog device is any creditable engineer/scientist would never attempt to resolve/validate a 250Hz dynamic signal (G-Cog) with a 2Hz one (SRM)? But yet the authors representing Université de Reims Champagne Ardenne, Laboratoire d'Analyse des Contraintes Mécaniques apparently think that this is acceptable? This is not an accepted practice in signal processing, not sure why they feel it is acceptable for cycling anaylsis.
Please see the attached figure I provided (sample data chart) showing the data samples collected on BMX power data across one second, I have marked the graph showing a typical SRM data collection (0.5sec intervals, Theorized). I would really like to know how anyone continues to believe this is an acceptable form of data collection for a BMX sprint event. All of the relevant information is missed.
2. All of the results of the laboratory experiments using the Fortius trainer should be tossed as they serve no relevance to the scientific community. One major critical flaw that they ignored even after I presented to this research group in France after the 2009 SX World cup final is that the sensors in G-Cog are piezoelectric and are not suitable for constant load application. By design these torque sensors measure dynamic loads only and when a constant load (Fortius) is applied they dissipate much like the charge placed on a capacitor. Their results show exactly what happens no matter the applied constant load the G-Cog power output remains low and unchanged this is expected but only if you know how the device works which clearly the three authors did not.
G-Cog was designed this way not to be compared with road power meters but due to our space constrictions of having a 0.1x2.0inch stress region piezo was the only way to go. The negative results obtained are expected when using the G-Cog outside its intended use platform. G-Cog is not a steady state road powermeter. It is a high data rate dynamic sensor collection platform. It would be unacceptable to use a yard stick to create precision machined components in a machine shop. The same applies for using a SRM or Powertap to collect highly cyclical dynamic BMX data at 2Hz and 0.8Hz respectively, it shouldn't be done no matter what the prior "research" states. I would urge other researchers to keep this in mind and use the correct tool for the job. We do not recommend using G-Cog for road applications.
3. On page 3 of the paper they state the following and I quote:
"For the Sprint tests , when we analyzed the difference between the maximal PO values of the SRM and the G-Cog with the same sample rate (2Hz)"
How did they achieve this? The lowest possible setting in the G-Cog user software is 25Hz??? The only way to achieve this is through a firmware/software change which we have never performed and can only be performed here at the factory (see screen shot of online G-Cog manual, coach.bmp). Yet they state they ran the two systems side by side at this frequency? This is impossible to do without data manipulation post collection. They claim the study follows the guidelines of the ethical standards set forth by the Int J Sports Med? But there is no explanation of how they performed the impossible and ran a G-Cog system at 2Hz? The simple answer is they did not and they manipulated the data.
4. The legend labels of figure 6 in the paper are incorrect. The authors
never make any reference to collecting SRM and G-Cog data at 0.5Hz (should
be listed 2Hz, but again see item 3). This is a small mistake but nonetheless
a mistake. Errors like this and the rest that were made throughout the paper
should have never have received approval.
We have reached out to both the journal and the authors. This page on the website was created to help shed light on what has been released to the community without our consent. Independent scientific research is wonderful and we welcome it but not when the methods are flawed and this gets published as fact. It is very disturbing also that the IJSM has taken the stance they have regarding this issue.
Any questions/comments can be directed to George@G-Cog.com