This is a two-fold question (or topic). These discussions keep popping up, often in relation to non-round chainrings and in particular whether power numbers are accurate when using osymetric.I feel there's a lot of confusion around this, and I'm calling on anyone with some actual specific engineering knowledge to help clear this up.
Firstly, I'm curious about what assumptions and models are used when converting an electric reading from the strain gauges into a torque value. I know that this is reasonably linear, so that we normally assume that T = aX+b with b being strongly affected by temperature and needing frequent (automatic) recalibrations and a being generally quite stable. But if this was the full picture there could really be no dependency on the type of the shape of the chainring, so there must be something else going on. The torque value is obviously sampled discretely (anyone knows what frequency is used?) and averaged. So averaging over one revolution, for instance, there are clearly some assumptions that can affect the result; do we assume constant torque (clearly not true, neither with round nor non-round rings), do we assume constant cadence (again, may be true on flats but certainly not true grinding it out on a hill?
Secondly, in the real-world, there certainly seems to be large differences depending on the cranks and the chainrings used. Looking at quarq data (http://www.quarq.com/ring-difference), there are some shockingly large differences. SRAM Red 53/39 -> SRAM TT 55/39 yields a difference of nearly 5%. And, incidentally, osymetric rings do seem to give a higher reading than many other combinations so perhaps that could explain a little bit of the mysterious tail-wind effect they have ;-). The question is -- why? The torque is measured on the spider -- in the static case, i.e. ignoring any shearing or flex, how can it matter how the it gets there?
Lastly, this applies to power2max. They don't seem to give any calibration data at all, and generally just say that this is nothing to worry about. Does anyone know if they are doing anything substantially different or if they are just glossing over potential miscalibrations figuring that their customers are better off not worrying about a measly 5% here or there?
--
When I channel my hate to productive, I don't find it hard to impress
Firstly, I'm curious about what assumptions and models are used when converting an electric reading from the strain gauges into a torque value. I know that this is reasonably linear, so that we normally assume that T = aX+b with b being strongly affected by temperature and needing frequent (automatic) recalibrations and a being generally quite stable. But if this was the full picture there could really be no dependency on the type of the shape of the chainring, so there must be something else going on. The torque value is obviously sampled discretely (anyone knows what frequency is used?) and averaged. So averaging over one revolution, for instance, there are clearly some assumptions that can affect the result; do we assume constant torque (clearly not true, neither with round nor non-round rings), do we assume constant cadence (again, may be true on flats but certainly not true grinding it out on a hill?
Secondly, in the real-world, there certainly seems to be large differences depending on the cranks and the chainrings used. Looking at quarq data (http://www.quarq.com/ring-difference), there are some shockingly large differences. SRAM Red 53/39 -> SRAM TT 55/39 yields a difference of nearly 5%. And, incidentally, osymetric rings do seem to give a higher reading than many other combinations so perhaps that could explain a little bit of the mysterious tail-wind effect they have ;-). The question is -- why? The torque is measured on the spider -- in the static case, i.e. ignoring any shearing or flex, how can it matter how the it gets there?
Lastly, this applies to power2max. They don't seem to give any calibration data at all, and generally just say that this is nothing to worry about. Does anyone know if they are doing anything substantially different or if they are just glossing over potential miscalibrations figuring that their customers are better off not worrying about a measly 5% here or there?
--
When I channel my hate to productive, I don't find it hard to impress