Accuity V4.1–95% Complete!

It’s coded. It’s been running for about 70+ hours of testing and programming and forgetting to disconnect it on a CR2450 and still has life. It’s been extensively calibrated on the bike (this is necessary for removing bending from the left leg sensor). All parts (crank included) is less than $200 dollars. However skilled labour required is very high in terms of both skill and time. No, I won’t be selling these for $200 – my time is very valuable – but I still think a dent can be made in the $1500 mark. Powertap is forging the way.

To do:

1. Move the Reed Switch closer to the Quarq Magnet location
• My Madone has a different magnet taped on. The Nice Quarq ring magnet is on the old GT GTR Series 4.
2. Put on Chain rings
• RIGHT! I knew I forgot something.
3. Test on Trainer
• But I’m so out of shape.
4. Test on Road
• See above…? I’m really out of shape lately.
5. Finish Enclosure design
• Don’t ride in the rain until then!
6. Torque effectiveness and Pedal Smoothness
• Easy to calculate, but need to spend more time modifying the nRF51422 ANT+ library to support it.

It’s taken a while, but it’s doing well. It has 16 – 18 bits noise free. I’m using only 16 bits of the 24 the ADC is as it’s just about 100% noise free. It’s about 15 and 24 counts per N-M. Essentially this relates to < <1% error BUT keep in mind there is some sensitivity to pedal offset. It’ll likely come in at the competitions 1.5% – 2% quoted marks.

Okay, so maybe not 95%. I am very close. I am off schedule though, but honestly not by much:

I really didn’t include a prototype stage! Major oversight. However, if I build another 4 in a few days I’m 100% on schedule. No, I don’t expect this to happen. I need another month. So mid September for beta group I suspect.

What can it do?

It’s a powermeter so:

• Measure Left and Right legs torque individually
• Measure Cadence
• Calculate Power
• Transmit power data on the most common wireless powermeter standard - ANT+
• Calculates L/R power balance
• Costs a lot less than the competitors, but a lot of build time (Mixed feelings!)

Here are some pictures

Calibrating the non-drive side. The 5k on the right is to keep the rope stretched and the cranks level. 15kg “calibration weights” measured on calibrated scales at work which are good for 0.1g.

Debugging: Programmer talks to Accuity. Arduino receives the serial output. The .NETMF Fez Panda talks to another ADS1248 that’s also measuring one of the sensors to ensure Accuity is reading right. The Gripper is connected to a test strain gauge beam to ensure the Fez Panda ADS1248 is working.

Accuity up close. Personal. And poorly soldered!.

Serial Terminal and ANT+ Display Simulator.

So long for now

Testing with a Powertap G3

As I mentioned in a previous blog post I’ve been testing my V3 Powermeter. There is going to be a major overhaul with the next iteration but I can still learn a huge amount from V3. Just like how I learned from V2. I’m mainly concerned with testing the consistency of the readings. I’ll be making a post Sunday or Monday with the actual results but suffice to say the results are very similar though my calibration is off, it’s reporting lower power than expected. What you are looking at in the above picture is what I was looking at during and interval session with my prototype along with a Powertap G3 rented from Cycles London. Above you are seeing a movie on one monitor with two custom pieces of software I wrote. One I mentioned previously as my trainer software that I use for indoor training, and one is a program that I pulled together in 3 days of spare time. It connects to heart rate monitor, speed cadence sensor, and two power meters and records them synchronized at 0.25 second intervals. That’s right, 4 times a second.
Here is the catch on why 4hz recording. Powermeters transmitting on ANT+ can do so in a few ways. It can transmit the basic power profile which is mainly power and balance, crank profile which transmits more accurate cadence and torque but no balance, and wheel which does the same as crank. It transmits at 4.06hz. More after the break.

nRF51422 EK / DK Arrival

After some mix up and delays by Fedex my EK and DK kit arrived. Some might be wondering why I ordered both the EK and the DK. I don’t actually have an nRF6700 (nRFgo starter kit).

The DK, or developer kit, “requires” a starter kit for Nordic development – the nRF6700, and if I truly need it I’ll get it. However, the nRF6700 is used for essentially power and nothing else. The DK has pin headers on the back of the two boards supplied so I can just run some wires out there. It comes with five sample chips and a standalone programmer which I’ll need for developing my own board. That’s the bigger motivating factor here.

The EK, or evaluation kit, has the reference design for the nRF51422 plus a trace antenna and an on-board Segger J-link programmer and debugger. What you really need for development. Nuhorizons has these for 88 dollars, a drop from 99 dollars but only has the DK in stock. I think they have 2 EK on order though. Anyway, Honestly I think the EK price is a little high if we are comparing it to the the TI series or even an Arduino with on caveat – This is meant for ANT+ and as such comes with some Nordic backing in terms of an SDK.

The SDK comes with reference code for power meter, speed/cadence sensor, heart rate monitor, stride-distance monitor and a generic ANT communications design. In 10 minutes I had all the software installed, the ANT+ network key installed, and was transmitting the basic power profile to ANT+ display simulator. It’s programmed in a much more efficient way than my previous prototype, but it’s a good deal more complicated.

Also just rented a G3 Powertap. Time for some comparisons of my meter and a production unit. Might have to code up a multiple power meter recorder software.