Power meter and power oscillations

Hey, I’m curious how people use their power meters to hit a target output.

I have an issue with 3s power averaging causing me to bounce around my target power. My current plan is to put 1s power on the screen next to 3s power. What do you do?

Here’s what’s been happening:

Let’s say I’m just starting an interval. I’m currently pedaling at 60 rpm (one revolution per second) and for sake of easy calculations, 100W. Ideally, what I would do is jump straight to my target power, let’s say 1,000W. (No Pav doesn’t have me doing that! It makes the math easier.)

I can increase the power to my legs pretty fast, probably within one revolution. Let’s say I’ve timed this perfectly to a hill so that my rpm stays the same. After one revolution (one second), my 1s power is 1,000W. But my 3s power is 400W.

I get worried. My interval is only 30 seconds and it’s saying I’m not even close to my target power! I decide to pour on fuel and do 1s at 1,300W. I’m 2s into my interval and now my 3s power is 800W.

Gosh that must be right! I keep on at 1,300W for 1s. Now my 3s power is 1,200W.

Oops! Overshot by kind of a lot. Now I need to ride under my target to bring the average down. Maybe say 700W for 1s. After 1s at 700W, my 3s power is now 1,100W.

Shoot it’s still coming down. Okay, another 1s at 700W. Now my 3s power is 900W.

Well that’s closer! Okay let’s see if I can guess what 1,000W feels like. What was I doing before? Well, I first put down what I thought was 1,000W (and it was!) but it said 400W. I can’t do that math in my head, who knows what it should have been. Then I did something greater than 1,000W (actually 1,300W) and it said it was 1,200 W. Okay… maybe split the difference?

While I’ve been thinking 2s has passed and my 3s power is now 700W. AGHHHH! I try to split the difference and end up at 1,200W.

And the cycle repeats- high, low, high, low, and my interval looks like crap.

I’m seriously contemplating programming my own data fields to fix this issue. I think it would be nice to see a 3s power that doesn’t include data from the previous interval at the beginning, as well as the “instantaneous” power (averaged over the last revolution). Of course, I don’t know if the Garmin API would let me do that, but maybe it’s worth a shot.

What do you think?


Of course the real problem is I’m not experienced enough to just know what 1,000W (and everything in between) feels like.

On the other hand, if I knew that, I wouldn’t need a power meter… not to mention my fitness changes and so efforts feel different at different points in my training… hmmm…

1 Like

I’m sure Pav will have a say, but I think it’s just something that will come with experience. I think most people find that having the instantaneous 1 second power really isn’t very useful because it fluctuates way too much. I usually look at 3 sec power at my current power and also have average and/or NP on my data field, as well as lap avg/NP for intervals. It’s just going to be a challenge to keep the average power right on target for short intervals, but with some experience you should be able to gauge your effort more accurately. I think many of us tend to start out too hard on those short intervals, just make sure you pull it back slowly towards the target instead of letting it drop to way below the target.


Ah, somebody technical with the same issues.

So there are a bunch of things. ERG with ZWIFT. PITA. Oscillates around the target like crazy. I found that using ERG smoothing in my KICKR helps there. It’s kind of like applying 2 3 second filters cascading. You still get the same real power, just fewer oscillations.

Got to try that with just the bike computer for stationary workouts. Maybe the WAHOO allows more control there. Getting bored watching ZWIFT anyway.

Outdoors. For climbing I typically use 5 seconds as smoothing interval. Got a left sided only power meter and if I am grinding away at 49 rpm then 3 seconds is fairly pointless. Typically I use a secondary 30 second interval to check whether I am slacking or not. Then I do use laps. So at the bottom of a climb I start a new lap, and hence have the stats along the way.

For sprints, I gave up. Indoors the latency for the trainer is too log to get any real numbers. And outdoors, it’s tricky to do a 5 second sprint or a 30 second sprint. For me at least.

Having said all of that, I do intervals indoors. Seems to be more practical than going outdoors and wasting time trying to get to a place where you can do a decent workout.

Ah, when it says 5 Hours Base Z1-Z2 I tend to track the normalized power rather than the average. The average seems to be vastly off if your outdoor ride has climbs and descents.


Firstly - what @letsplaybikes said.
Secondly - when I started out with Pav I had a lot of the same questions. I am very data focused. Understand the intent of the intervals that you are doing. Anything shorter than 1 minute is a directional target power and not absolute.
For example - on 30 second sprints TP sets my target power at 750W. The intent of the intervals is to empty the tank on each and every interval. When executing the first 4 are all way over 750W and it is not until #5 that it starts to drop to that level. If I followed the TP target I wouldn’t get the desired training effect.
Net, on any short interval I judge my pace myself.


Some great feedback here! @Joel happy to discuss this more on our next call :slight_smile:

1 Like

@letsplaybikes Ah yeah, I’m trying to compensate for my lack of experience by leaning on math :nerd_face:

I think that’s a good point @Kicikacsa, the intent is the most important part. I probably don’t fully understand that!

@miffedoldpizza, I’m lucky enough to have a pretty steady and long canyon near me! I go outdoors for this stuff. I don’t have a trainer, and I need to learn technique and get practice, not just to improve my strength.

@pav Definitely! I’ll text you, other questions about things too, especially mental training.

1 Like