Wednesday, January 16, 2013

V3–Power Meter–The Complete How To

For the current visitors to the blog, it's finally here! For new visitors welcome to the complete how to. What I am going to explain start to finish how to make and advanced left and right measuring power meter. The quickest and easiest way to figure out how to put your own together or find out how other power meters work such as SRM, Quarq, Cycleops or Power2max is watch the videos in sequence below.DSC_9066
Left to Right V1, V2, V3 of my power meter. Questions?

V3 Short Demo Video

The first thing I’m going to do is a cost breakdown. It’s likely what is on peoples minds most. Most power meters are minimum of $1500 dollars, ranging to almost $4000.
V3 Parts Cost quantity price total
Crank 1 140 140
Strain Gage 2 12.5 25
Instrumentation Amp 2 4.1 8.2
Low power ADC 2 18 36
Micocontroller (Pro Micro) 1 25 25
ANT+  AP2 Module 1 24 24
Resistors 4 4.6 18.4
PCB (Rush Proto Service) 1 37.5 37.5
Misc (Caps, wire, pins, etc) 1 20 20
Build a PM for: 334.1
Garmin Edge 500 1 250 250
Strain Gauge Supplies 1 1000 1000
Gage Installation tester (used) 1 100 100
V2 boards + parts + spares 1 600 600
V3 Spares 1 400 400
Labour 1 500+ hours Free
Development Total: Total 2684.1
Development Total: Total w/tax 3033.033
The cost to build is pretty cheap at less than 350 dollars, however gearing up, and V2 developmental costs means this is much much more expensive to cut your teeth on.
However, lets look at projected cost for V4 if I can develop it. This is does not include the huge amount of labour so don’t get excited and expect power meters on the market at 250 dollars. It’s not happening.
V4 Projected Parts Cost quantity price total
Crank 1 140 140
Strain Gage 2 12.5 25
Low power ADC 1 13 13
Micocontroller (nRF51422) 1 2.3 2.3
PCB (Longer turnaround service) 1 20 20
Misc (Caps, wire, pins, etc) 1 40 40

  1. Introduction
  2. What is power, torque, and angular velocity?
  3. How is cadence and torque measured? What is a strain gauge? How do we measure such tiny resistance changes
  4. How do I install that strain gauge? What’s important to know.
  5. Let’s put it all together and see it working & double check the calibration and on Bike Demo.
  6. Short Demo (Previously Posted)
How to Links
  1. How to install strain gauges
    1. The Power Meter Origin – The abbreviated version of this and this page. Those pages are now deprecated and will shortly be edited to say as much.
    2. V3 Power Meter–Initial Instrumentation – How V3 was instrumented. Lots of pictures and explanation.
  2. Power Meter PCB File (And Eagle) – NOTE: See posting below “Hacking your own PCB”. I replaced the LM4140 with a voltage divider as it’s more stable but the PCB was not updated. The Board also has a coin cell holder, temporarily replaced with a lithium pack as it draws to much current for the coin cells.
  3. Code – Located here. Will be updated soon with better commenting.
Additional and Useful Links
  1. Vishay Data Books -
    1. Surface Preparation – PDF Link
    2. Technical notes – Lots of tech notes on installation, errors, sensitivity, etc.
  2. Intertechnology Ltd. -  Canadian distributor for Vishay products
  3. Digikey – My main supplier for components, they also have some Vishay strain gauges.
  4. Omega – Low cost instrumentation, sensors, and strain gauges + installation supplies. I’ve never used their supplies but the double shear gauge is used in V3
  5. Binsfield Strain Gauge installation videos – Part 1 – Useful to understand installation basics.
Important Posts (chronological)
  1. ANT+ For Embedded – Some history
  2. AP2 Module and design changes - evolution of V2
  3. Omega Strain Gauges – Receiving the new gauges for V3
  4. V3 Power Meter Initial Instrumentation – Important about how I installed the strain gauges
  5. V3 Reading sensors and transmitting power – The basics of getting the power meter working
  6. V3 – Power Consumption – Measures and outlining current consumption. NOTE: This is still not solved. Right now V3 is a power hog and the coin cells DO NOT WORK. I’m using a 110mah Lithium Polymer Cell until I sort it
  7. Hacking your own PCB – Critical, replace the LM4140 with a voltage divider, it will work better. Still works with LM4140 though.
  8. V3 – Why it works now and Power Meter Code – Found a short, solved all my problems by fixing it. 1.5 months of messing around.
Some more pictures
Installed on my bike on my cycle trainer.

You can see the lithium Cell that replaced the coin cell holder.
The furthest most chips are AD623, the ones closest to the Pro Micro is AD8321. The top where the LM4140 would go is replaced with a hack for a voltage divider.
34 Gauge Wire Coated in polyurethane heading from the “torque tube” to the PCB.
Strain Gauge inside to measure left leg torque.
Double Shear gauge to measure drive side torque.


  1. Amazing work! Do you have details of the synchronization of torque and angular velocity, as well as the separation of propulsive versus non-propulsive torque components? Seems the issue is strain is a tensor, and there's axial (diagonal) components and off-diagonal (shear) components. From that you need to torque vector, then only the component perpendicular to angular velocity, then align torque time-dependence with angular velocity time-dependence.

    1. Thanks!

      The strain gauges are designed in such a way that forces in the "incorrect" direction cancels each other out. For instance if I was to pull at the pedal mount co-linear to crank arm it would increase resistance of all four gauges. If I was to bend the arm inward or outward from the frame it would have the same effect... increasing or decreasing all gauges. If each leg of the wheatstone bright is made up of 4 x 350 ohm resistors and all are measuring an additional 1 ohm, so 351 ohms, the wheatstone bridge is still balanced and doesn't detect. I've done tested this with a NI Compact DAQ and with huge forces I've been able to get up to about 2 microstrain which is well below 1%.

      Mismatches in the strain gauges dimensions, resulting in slightly mismatched resistance, is the only way that the incorrect force can manifest itself. Since the gauges are 1% accurate to one another, this has a minimal impact and can only have a minor contribution. All power meters are susceptible to this. Keep in mind that SRM and Quarq have chain rings that are offset from the centerline of the spider causing torques at the measuring point in the wrong direction like you described. People just don't realize this. It's the exact same problem, it's just nobody realizes to ask... and the answer is the same as mine - that it cancels itself out.

  2. Hi Keith, you blog is amazing

    I am doing a project on Strain gauge sensor, I want to measure the power from the applied strain and I am using OMEGA SGT-2DD/350-SY13 full bridge strain gauge

    I have the following information

    model : SGT-2DD/350-SY13,
    Resistances: 350 ohms
    Gauge Factor: 2.14
    Maximum strain we can apply is 30,000

    I am mounting Strain gauges on either side of the shaft. I am struggling to adjust the gain for the output differential voltage which is 6.996 micro volts for 1 microstrain and 35.3 milli volts for 5000 microstrain.

    Can you please suggest me the range and suitable instrumentation amplifier for the strain gauge

    Thank you

  3. This is awesome. Just out of curiosity, what is your background? I'm currently studying mechanical engineering at Queen's and I'm very interested in developing similar products in the future. Where did you learn integrated circuit design? (I'm currently working with arduinos but my work pales in comparison)

    1. Don't underestimate what you can do with an Arduino and the really easy to use libraries. It's having the guts to just keep plugging away into the unknown. Don't forget the community. It loves to see and share your work as much as you can get from it.

      I have two degrees, both technically mechanical, but the Masters is more mechatronics. In my 8 years of university, I didn't learn much about this stuff that I didn't already know or couldn't have found online via sparkfun, hackaday, or make.

      If you're into sports specific stuff I'd suggest signing up to become an ANT+ adopter and start taking a look at their communications protocols. My code is for an Arduino Pro Micro, but will work with just about anything really. My new V4 code will use an ARM based nRF51422 chip so it is very different. A lot of improvements though in terms of power, size, and accuracy.

  4. Thanks for sharing excellent information. Your site is so cool. I’m impressed by the details that you have on this blog. It reveals how nicely you perceive this subject. Bookmarked this web page, will come back for more articles. Sgt. Hack

  5. Fantastic work, really impressive. One question though, why use the ADS8321? is the one in the micro not good enough?

    1. 10bit noisy on chip might manager 8 clean bits, and lose another bit for having to raise the level of the differential signal. So 7 bits, or 128 levels for full scale, and then losess due to offset so maybe 6 bits. I worked it out once to something like 4 N-M as the minimum resolution. That's quite useless. 16bit SAR sucks too I've found due to the noise and how it operates. Needs better analog electronics to keep things clean. There are other options that are better.... Sigma-Delta

  6. Hi Keith, amazing blog and I have only just scratched the surface! I have a question, apologies if the answer is obvious, but are you averaging the strain (torque) per rev? Or are you taking the instantaneous strain (torque) as the cadence sensor passes the pickup? Thanks

  7. Hi Keith,

    It appears that the link to the Eagle files is broken. I am struggling to get the AP2 module to send data. Can you upload the Eagle files?


  8. Keith stumbled on your site - looking for advice inspiration on how to acquire/development a module to read a power meter output so as to develop a mechanism to control resistance on bike rollers (using electro magnets and varying the voltage and thus the magnet field round the aluminum rollers).
    Just like to say fantastic information and dual sided power meters are heading south of $500 and interestingly the configuration looks kind've similar to yours in a very experimental fashion.