NP < AP on VO2 intervals?

OT....

What power meter do you use on your mtb? I would love to get a PM for my mtb but dunno which one I would get...

jetnjeff said:

Any body have a clue why?

Deviations from the average power (either positive, or negative) are exaggerated through the NP algorithm. Riders typically see the positive side of this over rides where the majority of the power is at one level, and large power increases during short efforts (hills, surges, sprints) produce an effect on NP which is greater than the effect on AP. However, the inverse can also happen during efforts such as you described, where the majority of an isolated period (ie, one interval, as opposed to an entire ride) is at a consistently high power, but there may be short periods of significantly lower power due to grade changes, soft-pedalling through a loose section, or whatever.

My guess is that your intervals have some "negative power spikes" in them where you didn't hold power through the grade changes, which is dragging down your NP. It's not anything to be concerned with, however.

frenchyge said:

Deviations from the average power (either positive, or negative) are exaggerated through the NP algorithm. Riders typically see the positive side of this over rides where the majority of the power is at one level, and large power increases during short efforts (hills, surges, sprints) produce an effect on NP which is greater than the effect on AP. However, the inverse can also happen during efforts such as you described, where the majority of an isolated period (ie, one interval, as opposed to an entire ride) is at a consistently high power, but there may be short periods of significantly lower power due to grade changes, soft-pedalling through a loose section, or whatever.

My guess is that your intervals have some "negative power spikes" in them where you didn't hold power through the grade changes, which is dragging down your NP. It's not anything to be concerned with, however.

Thanks,

That makes perfect sense. I was trying to hold power durring the grade changes. I noticed that was a problem last week and still have work to do to make sure I increase my cadence and don't get lazy. Things I wouldn't have known if I did not have a power meter!!!😄 😄

mattv2099 said:

OT....

What power meter do you use on your mtb? I would love to get a PM for my mtb but dunno which one I would get...

I have a Power Tap Pro. The SL does not come in a 135 axle yet.

I like the concept of the Ergomo better but they are still working on the MTB version and it cost more particularly the computer, which is the most likely thing to get wrecked in a crash. I usually break every thing, but I have only 8 rides on my PT. 4 were real MTBing and one was in pretty severe terain. So far so good.

It is great bio feed back.

jetnjeff said:

I just checked out my file for ~6 minute VO2 max intervals. They were done in a MTB up a gravel climb that varied between Steep and very steep. On all five inervals my NP was about 5 Watts less than my AP. The grade changed a lot. I would expect the NP to be about the same(due to the short controled duration) or a bit higher due to the varying pitch.

Any body have a clue why?

In theory (i.e., from a mathematical perspective), normalized power should always be at least equal to, if not greater than, average power. Due to the use of a 30 s rolling average, however, on short efforts it is possible for normalized power to be slightly less than average power. I wouldn't recommend putting too much stock in normalized power during such short efforts in the first place, though, so I don't think it is worth losing any sleep over.

jetnjeff said:

I have a Power Tap Pro. The SL does not come in a 135 axle yet.

I like the concept of the Ergomo better but they are still working on the MTB version and it cost more particularly the computer, which is the most likely thing to get wrecked in a crash. I usually break every thing, but I have only 8 rides on my PT. 4 were real MTBing and one was in pretty severe terain. So far so good.

It is great bio feed back.

I was unaware that the pro came in a 135mm axle. I've checked the website and found no such info. I've got a SL on my roadie and I would love a pro on my mtb.

acoggan said:

In theory (i.e., from a mathematical perspective), normalized power should always be at least equal to, if not greater than, average power. Due to the use of a 30 s rolling average, however, on short efforts it is possible for normalized power to be slightly less than average power. I wouldn't recommend putting too much stock in normalized power during such short efforts in the first place, though, so I don't think it is worth losing any sleep over.

i have noticed that sometimes on my ride files the NP for an hour will be say 320 and then i go to the athlete homepage and look at the mean maximal power and it'll say 350 for that same ride, which number is more reliable?

mattv2099 said:

I was unaware that the pro came in a 135mm axle. I've checked the website and found no such info. I've got a SL on my roadie and I would love a pro on my mtb.

Actually the website says both the Pro and the SL are available in 135, but in reallity only the Pro is. Saris actually markets the Pro 135 axle as MTB and you can get pre built wheels. No Disc brake option though. So I got a hub and a ceramic mavic 717. I like V-brakes and ceramic rims over Disk anyway, lighter more standard parts etc.

acoggan said:

In theory (i.e., from a mathematical perspective), normalized power should always be at least equal to, if not greater than, average power. Due to the use of a 30 s rolling average, however, on short efforts it is possible for normalized power to be slightly less than average power. I wouldn't recommend putting too much stock in normalized power during such short efforts in the first place, though, so I don't think it is worth losing any sleep over.

I forget what is the minimum time that you would want to use NP for?

cassette player said:

i have noticed that sometimes on my ride files the NP for an hour will be say 320 and then i go to the athlete homepage and look at the mean maximal power and it'll say 350 for that same ride, which number is more reliable?

In the file, CPS finds the highest 60min AP and then gives you the NP for that particular period. In hompage, it actually seeks the highest 60min NP. Highest AP and NP often don't match.

beerco said:

In the file, CPS finds the highest 60min AP and then gives you the NP for that particular period. In hompage, it actually seeks the highest 60min NP. Highest AP and NP often don't match.

what does AP stand for

acoggan said:

In theory (i.e., from a mathematical perspective), normalized power should always be at least equal to, if not greater than, average power. Due to the use of a 30 s rolling average, however, on short efforts it is possible for normalized power to be slightly less than average power. I wouldn't recommend putting too much stock in normalized power during such short efforts in the first place, though, so I don't think it is worth losing any sleep over.

Nothing earth-shattering from me as usual but I think I see why CPS often reports lower NP than AP for short(ish) intervals.

If we consider a pure square-wave interval, say 450Wx5min ON and 200W.

Clearly AP over the ON period is 450W. But when we consider the effect of the 30-sec RA, it takes 30-sec (or 24 std. PT samples) for instantaneous NP to reach 450. It's then 450W for the rest of the interval .... no problem.

Averaging NP (taking P(t)^4, then ^1/4 of the average) over the same 5-min period, we catch the 30-sec ramp from 200W to 450W, then 4.5min at 450W, but none of the carry-on effect of that 450W effort on the rest period NP.

Or for my example, NP= 442.7W versus AP=450W.

I believe this is the problem. Not that it's a big one -- just for intervals where 30 seconds is an appreciable percentage of the duration.

On the road, I suspect normal power variation masks this effect such that NP is nearly always higher than AP but not as high as it 'should' be.

Am I off-base here?

rmur

rmur17 said:

Am I off-base here?

I think your logic is right on, but the way Cycling Peaks applies the averaging seems a little different than you describe. Looking at the graph of an L5 interval I had done on the trainer, and then incrementally increasing the averaging interval, CP appears to average the power at a particular point in time with the *next 30 seconds* as opposed to the previous 30-sec history. The result is that if a range is defined on a square wave interval, and then 30-sec averaging applied to the display, the ramp-up falls before the defined range and the ramp-down falls within the range.

Like I said, I think you were right-on with your thinking, and the shorter the range, the larger the difference between AP (ave power) and NP.

frenchyge said:

I think your logic is right on, but the way Cycling Peaks applies the averaging seems a little different than you describe. Looking at the graph of an L5 interval I had done on the trainer, and then incrementally increasing the averaging interval, CP appears to average the power at a particular point in time with the *next 30 seconds* as opposed to the previous 30-sec history. The result is that if a range is defined on a square wave interval, and then 30-sec averaging applied to the display, the ramp-up falls before the defined range and the ramp-down falls within the range.

Like I said, I think you were right-on with your thinking, and the shorter the range, the larger the difference between AP (ave power) and NP.

okay if CPS 'shifts the averaging window' by 30-seconds - delaying averaging until 30-sec into the interval but counting 30-seconds beyond the e/o the interval, my example yields the same power ~442W.

It seems the only way to get the correct result in this example would be to delay the averaging until 30-seconds in but stop right at the e/o the interval. That yields the correct 450W but means NP is for 4.5 rather than 5-minutes ... probably not a big deal at all.

In any case, it seems to be a coding issue -solvable by perhaps simply setting the rule NP always >= AP :-)

rmur

rmur17 said:

It seems the only way to get the correct result in this example would be to delay the averaging until 30-seconds in but stop right at the e/o the interval.

As I envisioned things, that's what it should do. Not having written the code myself, though, I can't say that it actually does do it that way.

acoggan said:

As I envisioned things, that's what it should do. Not having written the code myself, though, I can't say that it actually does do it that way.

err ... it seems at least the latest software version actually calculates exactly in this manner. I imported an L5 sample workout.csv -pure 450W/200W ON/OFF.

When I highlight a section of interest, AP is calculated as soon the interval begins but NP is marked n/a until it becomes >30-seconds wide. From there until the e/o the ON interval NP indicated exactly 450W.

I'll have to try that at home on the older version to see if it's the same but it appears I'm barking up the wrong tree :o

rmur

cassette player said:

what does AP stand for

Average Power