I have to thank Ray from www.dcrainmaker.com for his post. I think it paints a pretty good picture of things. His posting did highlight something. My initial reluctance to BTLE. This stems from a personal experience choice, but mainly cost. However, if down the road I can get some form of backing and some help with coding I will be dropping in the nRF51922 microcontroller (ANT+/BTLE simultaneously) when released for dual support of both protocols.
One point he made was about getting a product to 95%. I think it’s being misunderstood that the accuracy of my meter is +/-5%. This is NOT the case. My torque testing is easily measuring in the 2% described by other’s products. I believe he means in terms of finishing. That’s a big hurdle. Redesign – Packaging – Waterproofing – Validation – FCC / CE approvals - ANT+/BTLE approvals. That’s the last 5% of developing a product and it’s a big 5%.
I’ve been having good progress with the nRF51422 (ANT+) microcontroller and I’m about 99.9% settled on the ADS1247/8. The ADS1247 is a 2 differential (4 single sided) ADC while the 1248 is 4 differential (8 single sided). Why I might chose the ADS1248 is so that I can have a secondary force sensing arrangement. That means I could do the fancy directional force output that Pioneer has shown off. It’s information overload during riding, but it might be useful. I think I’m going to try and design the board to accept both. Future proofing I’ll call it.
I’m glad a couple of people commented on the gyro situation. 4ma doesn’t sound like a lot, neither does 2ma, but when your coin cell’s life curve shows a maxim total draw for 2ma then that is a problem. I have a gyro I like being put on a breakout board by Protoadvantage now. I’m hoping to have it next week. I like coin cells, but the gyro could kill that… or sticking with coin cells could kill the gyro. If you have an opinion, post in the comments or drop me a line.
Now the other thing which this post is titled. Beta Testers. If you are in any of the above listed area’s, an avid cyclist, and interested in access to some beta prototypes drop me a line at kwakeham@gmail.com or kwakeham@accuity.ca. I’m looking for local first so I can drive over and debug / check / upload new firmware / get your feedback.
I’m looking at two months to get the new design in order and a prototype built. After that I suspect it’d be another month or two before I’ll put a meter in anyone's hands. However, if you are further away and interested drop me a line too. I’ve had several people ask. There is a list being compiled. When www.accuity.ca goes live there will be a sign up section. This will be a closed Beta, more like an Alpha test really. Generally that means I’ll own the device, you have a loan of it and I will eventually take it back.
Thanks for following! And Thanks to Ray again for supporting the industry.
Hi, enjoying these posts hugely.
ReplyDeleteI would vote for lower power consumption and sticking with the coin cell, but then I'm probably not sure what the implications are.
Am assuming you want to fuse accel/gyro data to get an accurate crank position & therefore cadence? My gut feel is that accelerometer with sufficient filtering would do the trick, but then again I've never put an one on a bike! I'd imagine there would be a serious amount of short term noise to filter out once you're rattling down the road...
Regards,
Jon.
Hi Jon.
DeleteThe short is that due to start up times of a gyro (in the range of 30ms to 120ms) I have to keep it powered up at all times adding 4ma power draw. On a CR2032 this is beyond the continuous operation. They are at best 225mah, making powering the gyro last at most 56 hours, but there is going to be cut off voltages which stops it from operating due to the higher draw. So now we are down to 30 hours before powering the microcontroller and the strain gauges. I would be forced to the larger CR2477, but that is stop gap at best and won't be competitive in terms of the others 300 claimed hours.
Resolving Cartesian acceleration to position and rotational velocity is a tricky proposition. A gyro is very direct -- for the angular velocity. It would have to be integrated for position -- this doesn't actually work accurately in reality. My masters degree had me trying that for velocity with a MEMs Accelerometer. After 5 seconds it was drifting, after 10 seconds it was way off. However, an accelerometer could theoretically detect orientation due to the ability to sense gravity. This is and "absolute" sensing of orientation, meaning if you go up a hill it'll be skewed.
I think Rotor is doing it by sensing force peaks of each leg. This is an option to recalibrate the gyro / accelerometers - but then we are into some annoying math. The simple is the magnet and reed switch (or hall sensor) method.
how about using a magnetometer? Stick a magnetic source on the bottom bracket shell and you could deduce rotational position. Or you could use one of those new inductance-to-digital convertors that TI just brought out (LDC1000).
Delete