Apple iPod Nano 6g v1.2 (Part 2)

Back in early October, I gave a favorable report on using the internal accelerometer on the Apple iPod Nano 6g (when updated to v1.2), see here. I decided to test it again on the treadmill.

This morning, I was too fatigued to do the Al Corniche run I did two days ago (see here). (I was suffering perhaps too much from a skin allergy condition that has recently flared up again.)

As a result, I decided to simply run with the iPod Nano on the treadmill in cool air-conditioned comfort at the Mövenpick Towers hotel (see left) for about 30 minutes. The gym there does have a spectacular view of the Gulf from the 25th floor. Better than feeling sorry for myself and doing absolutely nothing I suppose.

But instead of setting a steady, fixed pace (like last time), I varied the treadmill speed up and down. In fact, I ran for the first 10 minutes at 12.5 km/hr (7.8 mph). Then bumped up the speed to 13.5 km/hr (8.4 mph) for 5 minutes, and dropped it to 10.0 km/hr (6.2 mph) for 5 minutes to recover. Finally, back to 12.5 km/hr for the reminder of the run.

So what would one expect the device to report? Well, one would expect something like the red line shown below (possibly shifted slightly vertically, modulo speed calibration of the treadmill).

I can report that the blue data points seem to track the red line fairly well. Though it's not clear there is much difference in the data recorded between the 12.5 km/hr and 13.5 km/hr sections...

(Note: I downloaded the raw data points and directly plotted it in Excel myself. For those interested, you can find them in the directory /Volumes/iPod Nano Name/iPod_Control/Device/Trainer/Workouts/Empeds/nikeinternal/synched when mounted on a Mac. )

This ridiculously heavy-handed smoothing is potentially a huge problem that renders the device useless under non steady state conditions.

(Ignore what I said above. In a previous edition of this post, I said the iPod Nano smoothed away the speed changes. Mea culpa. I screwed up. There was a mistake in my Excel formula.)

From my treadmill speed settings, I should get an average pace of 7'54" min/mile, assuming the treadmill was calibrated perfectly. The internal accelerometer gave me a 7'25" min/mile average or about 6.5% off the treadmill settings.

Apple iPod Nano 6g v1.2

The other day Apple introduced an update to the iPod Nano 6g that I use for running with the Nike+ receiver and shoe transmitter.

The update included the opportunity to dispense with the white Nike+ dongle that hangs off the 30-pin port (shown here on the left). In fact, it permits for the first time the use of its internal accelerometer to record run pace and mileage instead of relying on the separate shoe sensor.

This is quite exciting because...

• There is only one 30-pin port, and the iPod Nano lacks integrated bluetooth. So if that port was freed up, one could use a small bluetooth dongle and pair the Nano with bluetooth earphones. Yes, in other words, ditch that pesky cord that is always bouncing around and interfering with arm motion while running.
• Also, I wondered whether the Nano accelerometer would be less accurate than the accelerometer built into the shoe sensor: reasoning being that footfalls should be more accurate than torso movement: the Nano being clipped to one's clothing.

So I eagerly downloaded v1.2 of the system software for the Nano and installed it via iTunes. Things went smoothly and one good piece of news is that stored data was preserved: in other words, it didn't wipe out my pre-existing records of workouts on the Nano. So far, so good. Let's see if the internal accelerometer does as good a job as the Nike+ foot sensor.

To do the experiment, I ran on a treadmill for exactly 12 minutes at 7.5 mph (8:00 min/mile pace), once with the internal sensor and once with the dongle inserted.

The new version of the iPod Nano user guide recommends "Clip iPod nano to your waistband for better accuracy." So I obeyed and did the experiment so configured.

(Note: I began the timer when the treadmill was at 3.0 mph (walking pace), and held the increase-speed button until it hit 7.5 mph. Total time spent including the ramp up in speed was 12 minutes. Assuming the treadmill is accurate, the total distance should be just under 1.5 miles.)

Here are the results:

Neither sensor was calibrated beforehand. I was pleasantly surprised how close they were to each other. (The one on the left is the run recorded by the internal sensor.)

(Seems within the margin for error I've experienced with repeats using the Nike+ foot sensor. Also both sensors were slightly optimistic. Even without calibration, assuming the treadmill is correct, it is within single digit percentage points of reality.)

I was disappointed though that when I synced the iPod Nano using iTunes, it only uploaded the Nike+ foot sensor session to the Nike+ website. So it could only display one of the graphs.

I went into the Nano while it was mounted as a drive on my Mac. And extracted the raw data.

(See how to do that in my 2007 blog post here.)

It is mildly interesting that using the internal sensor records only distance (in km) every 10 seconds. But using the Nike+ foot sensor it recorded both speed and distance data every 10 seconds.

I plotted and overlaid the distance data (converted to speed) on top of one another in Excel:

As you can see the data are pretty comparable.

Hmm, now the dongle has been deemed unnecessary, I guess I could go wireless. I see the preferred set-up is to pair that tiny i10s bluetooth dongle for the iPod port with a pair of those Sony Ericsson HBH-IS800 earphones. My birthday is coming up. So I think I will reward myself...

Update: December 1st 2011: I have new and serious doubts about the usefulness of the built-in accelerometer based on a treadmill run. See Apple iPod Nano 6g v1.2 (Part 2).