I’ve joined two groups: 670G Users, which has a couple of hoops they want you to jump through before joining (they really only want people who have the sensors and are using–or at least have used–Auto Mode), and 670G Support, which isn’t quite as closed. I think the level of expertise and experience on the Users group is a bit higher, so it’s worth going through the rigmarole, but they’ve both been helpful to me.
Biking is my primary form of exercise–about 80 mi. a week as long as there’s no snow/ice on the ground. I’m still struggling with a.m. BGs, though the pump is getting better at keeping them at least in the >150 range. Low BGs after biking are more of a problem, though the temp target thing is pretty good at keeping them from being scary-low, as long as I remember to set it 1.5-2hrs ahead. Which means you’re really committed to doing it, so it does make the spontaneous ride harder to deal with.
I’m curious about this though:
The basal rates you set while still in Manual Mode don’t have any effect on what the microboluses do in Auto. The only parameters you can adjust in Auto are Carb Ratio and Active Insulin Time. Maybe you meant CR? I’ve had to tweak AIT to its lowest setting and I’ve got five different CR’s going now, all of them very aggressive compared to what the pump seems to like (“Your setting is outside the normal range” blah blah blah). One thing I’ve learned is that these parameters are only notionally related to what your actual CR and AIT would be. At least, I would say that the farther out your particular metabolism is from the middle of the bell curve the engineers are designing to, the more notional they are. Once everything goes into the Algorithm it affects everything else, and they end up being more just a way of forcing it to give you the results you want than a rigorous setting derived from a rigorously accurate measurement. For instance, on paper CR would seem to only be applicable to bolus estimates, but it definitely seems to have an impact on microbolus rates in some way shape or form. Or at least that’s the consensus on the user sites. One of the frustrations is the fact that the MdT engineers don’t seem to want to share too much detail about what the algorithm actually does, so we’re all forced to do a lot of collective guesswork.