Training with Real Power

If you like more accurate data and the ability to use power while riding on the road, a power meter may be a better choice for training than virtual power. However, early into my journey with a power meter and the change has required a few adjustments on my part. In the long run, I suspect I’ll be happier, overall.

I’ve been doing indoor cycling training with ‘power’ for the past couple years. I’ve put power in single quotes because power training can take many forms. There are several different kinds of power meters, which all work a little differently, and there’s also ‘virtual’ power, which really isn’t your power output at all. Regardless of how it’s measured, the data is typically transmitted to a mobile/computer application or head unit like a Garmin cycling computer wirelessly by bluetooth or ANT+, so you can monitor it while working out and train to specific targets, usually based on FTP (the power output you could theoretically hold for an hour).

I’ve been using virtual power and been reasonably happy with it. The week before last I finally bought a power meter. Below is my experience thus far in moving from the former to the latter.

For the Uninitiated

Power meters, as the name suggests, measure your power output when you’re pedaling. Most often, these are custom single or dual crank arms, wheel hubs or pedals, which measure stress or torque and turn it into a wattage number. There are advantages and disadvantages to the various types of power meter, which can be found here.

With virtual power, platforms like TrainerRoad and Zwift can use the resistance curves of indoor trainers, along with data such as cadence and speed, to calculate a rider’s guesstimated power output. This measure still gives you data for training, but is less accurate than a power meter. It works pretty well and its main benefit is that you don’t need to shell out somewhere between $800 and $1500 for a real power meter, or considerably more for a turbo trainer. The main disadvantage of virtual power is that you can’t use it when you’re actually riding on the road.

Getting Rolling

So, with that out of the way, I thought I’d just post a brief overview of my experiences thus far. The thing which actually enticed me to buy a real power meter was seeing an ad for a used Stages Ultegra 6800 left crank arm on Facebook for $400. So, I thought I’d go check what a new one was selling for and found the same unit on sale for $525 at MEC. Since I wasn’t likely to find one cheaper, I thought I’d go for it. I have an older Ultegra 6700 drivetrain on my bike, but after a little research I realized the 6800 would work fine for me, provided I didn’t mind my new power meter crank arm to be a darker grey than the chain ring side.

My 6700 crank arms are about six years old now and the 6800 Stages is probably four or five, thus the closeout price reduction. I found out after the fact I could also have purchased directly from Stages online for $399 US plus delivery, which would have ended up being a little more expensive that my MEC purchase.

Anyway, installation was a snap – basically swapping a crank arm. I found initial pairing and calibration with TrainerRoad on my MacBook was finicky. I couldn’t really get it to keep a connection and calibration repeatedly failed at first. At that time, I never did get it to properly pair via bluetooth, but finally got a tenuous connection and calibration via ANT+ (the same way I’ve always connected for virtual power workouts). Similarly, I couldn’t get it to pair well with my Garmin 500.

First Ill-fated Ride

Nonetheless, I went ahead and did my Sunday longer ride. Two hours turned into 90 minutes because I found sweet spot intervals which should have been uncomfortably doable were just about killing me. I also noticed that the tenuous connection with my laptop was still pretty loose, with the power reading dropping to zero intermittently while riding, for the whole 90 minutes.

It was after this ride that I did some online reading and polling of others which I should have done first. First, given that I changed my power measurement tool, I should have absolutely re-done my FTP test (something that I normally do fairly regularly anyway, duh). However, since it was my longer ride day in my training plan, I did that instead. Even with the crappy connection, I noticed my power reading at a higher rate of perceived exertion (RPE) was lower. After my post-ride research I realized that my real FTP was probably at least 15 watts lower than my virtual FTP, which would explain why the ride felt so hard given that my targets were too high.

Second and equally as important, given how old my ‘new’ power meter was, I suspect the battery in the box was probably pretty weak and it also dawned on me that the firmware in the actual meter would have likely been several years out of date. I’m happy to report that since changing the battery and updating the firmware, my laptop pairing with TrainerRoad has been solid four rides in. I’ve been connecting it via both ANT+ and bluetooth and letting the software use the connections as it sees fit. I haven’t yet bothered with my Garmin, as I want to do that with an extra ride that isn’t part of my training plan. Plus, I’ve read that the old 500 often has problems with a power meter, since it’s kind of overloaded when connecting with a speed sensor and HR monitor, too. Luckily, the power meter lets you get rid of your cadence sensor, as that would be four sensors. However, I’m really not wanting to upgrade my Garmin, as the 500 works fine for my needs.

A Quick Comparison – Virtual and ‘Real’

I can’t really find good data on the variations between virtual power and power meter measurements, but most people I talk to seem to think virtual power overestimates your output by anywhere from 10-20 watts. The actual number isn’t nearly as important as it being a consistent measurement so that you can track changes over time, reassess and ensure your training targets are accurate enough to help you improve.

FTP … How Low Can I Go?

I had a virtual FTP as high as about 270 watts a couple years ago. Last year I barely rode, trainer or on the road, so a couple reassessments had me at 254 and then 250. Finally, when I began my current sweet spot base program a few weeks ago, I reassessed my FTP and got a somewhat embarrassing 243, which was perfectly understandable given my lack of riding fitness.

With the first power meter ride a week ago so challenging, and knowing that virtual FTP is typically high, I figured I’d better reassess again as I suspected my target was too high. So, this past Monday I swapped the normal workout in my plan for another FTP test. 235.

Dear gawd, two thirty effing five?!?

Oh well, this number should finally be fairly accurate and there’s nothing like a low fitness assessment to motivate a middle-aged guy already suffering from diminishing returns.

So, my main takeaway with regard to the difference with FTP between the two? No matter what you think it is, it’s probably lower. Even if you recently reassessed your FTP with virtual power, you definitely need to reassess after you’ve installed a power meter.

Less Smooth, More Choppy

It’s definitely more challenging monitoring your power output as you ride with a power meter. However, I don’t think that’s a bad thing. When using virtual power, while it’s harder to maintain a smooth output at threshold and anaerobic output, I’ve never found the numbers vary that much. Once I hit the target for the interval, the fluctuations are a few watts at most and I get a pretty even line on the workout graph (below).

Virtual power graph – not TOO much variation

Since I’ve been using a power meter over the past eight days, I notice the number goes up and down with significant variability, sometimes 20-30 watts at a time (graph below). This is likely more pronounced on a single crank than dual crank meter as well, since it’s taking a single leg reading and doubling it through one revolution, as opposed to getting a reading from each leg per revolution.

Power meter graph – far more variation

As well, the effects of the power meter readings are more instantaneous than are the virtual power ones. If you reduce your output or stop pedalling, the reading change is almost immediate, whereas with virtual power there’s always a bit of a delay and it levels off smoothly before auto-pause kicks in. In general, training is just much more responsive with a meter than with virtual power.

I like this because I can always work on more consistent power output via ILT (independent leg training) intervals and my workouts will be a little more accurate with more responsiveness from the software. In the end, more accurate data will take a little getting used to, but is a good thing for training.


One thing I hadn’t considered, but which actually makes perfect sense to me, is that my speed would generally increase. The past four workouts since I switched have all been sweet spot (intervals that generally fall between 85% and 95% of FTP). It’s somewhat natural that, during the first workout prior to FTP reassessment, my speed was higher. After all, if my power output was actually lower, I’d have to pedal harder to hit a target that was too high for my fitness.

However, since I reassessed to the lower FTP, my workouts are still higher speed than they used to be. I’m using the same speed sensor, and my math-challenged brain figured that my speed would be more in line with my virtual power workouts, but that’s not the case. In general, the overall workout speed has been 1-1.5k/hr even after the reassessment. Not a massive difference, but just an interesting observation.


My final observation isn’t about the difference between training with virtual and real power numbers. Rather, if I’m going to realize one of the benefits of a power meter, it will be using it on the road as well.

After the battery replacement and firmware update, I was able to easily pair and calibrate the meter with my Garmin. However, I didn’t find it to be a satisfactory experience. Using Stages own recommendations, I disabled my HR and cadence sensors, while I did keep the speed sensor. Frankly, if I can only use power with the unit’s GPS while on the road (though Stages does replace a cadence sensor, so that helps), that doesn’t really provide much useful feedback data. Even with only the speed sensor enabled, Garmin had trouble maintaining a steady reading from the meter, dropping my signal several times. Given that this was stationary and inside, that problem would likely only be pronounced outside.

You can see the difference in the TrainerRoad and Garmin graphs below. TrainerRoad on the left has continuous connectivity (plus HR data, thus the read swimlane), while my Garmin 500 graph has a ton of drops and dips. I’ve since deleted the Garmin 500 activity, but I recall all the data being way off, as a result of so many drops. My distance and speed were both significantly lower, but happily the power was quite accurate when compared with TrainerRoad. This would be a bigger problem on an actual road ride, though distance and speed will be captured with GPS.

Stages also recommends per second data collection, which should improve accuracy. Given all that and that at least a couple of people have since told me that the old 500s have pairing conflicts with more than 2-3 sensors, I suspect I’ll need to upgrade my cycling computer to use the power meter on the road. Newer ones like the 520 or 820 apparently handle more simultaneous sensors fine, along with bluetooth to help with connectivity and several other features.

In the meantime, I’ll continue playing with the 500 and may possibly just need to go without power on the road and use power only while doing trainer workouts. The only way I’m going to really find out is to test it with only power and GPS on my next road ride, likely coming up as soon as next weekend. If that works better, I’ll see if I can add HR on a subsequent ride.

Overall, while it’s taking some getting used to I’m definitely a fan of more accurate data and I want a truer representation of my power output. As well, once a few kinks are ironed out, being able to use it on the road, particularly to compare things like hill repeats or Cypress climbs, can only help improve my cycling. Finally, seeing the larger variations in power output gives me not only a new target to focus on in my training but might finally make me want to do more of the dreaded ILT drills (which I hate).