How TR Anticipates a power change in erg mode... with different trainers.. help & advice please

Doing structured workouts in erg mode with a largish power change, eg 130w to 350W and I have hit a a problem. I have experienced a delay BEFORE the power starts to transition. This happened with my Direto smart trainer in both Zwift workouts and with the Elite MyEtraining software. The delay occurs both when power rises and when it is due to fall. I have found it massively disruptive to trying to do a structured session with short sharp power changes.

It did not happen with TrainerRoad. I thought this was excellent. I am a convert. I was also curious why this was the case.

I have been tipped off by GPLama and other sources, that the reason you do not have this problem with your structured workouts is that you anticipate the change and so send the power change signal a few seconds early so that the delay is minimised. Nice one. Neat solution guys. Love it. and it works. It is much better with TR and the Direto.

However, here is my problemā€¦ I am using an Elite Direto which seems to have a delay of 3-4 seconds and a slow transition. I am frustrated with this device. So, I am considering swapping to a Kickr Core. I am told by those who understand this problem that the delay before the power starts to change with a kickr is more like 1-2 seconds. And the transition is quicker. That is great. I am almost sold on thisā€¦ butā€¦

ā€¦my question is, if you apply a consistent adjustment to all trainers, does this end up being too early for the kickr core? Or do you adjust the anticipation of the power change depending upon the turbo (or brand) or whatever.

Or to put it another way, am I going to hit a different problem - that of the transition starting early? Or is this a known issue that you manage successfully from trainer to trainer.

I hope this makes sense? Cheers Phil

In general the interval duration will be consistent regardless of the responsiveness of the trainer.

They send the action to increase power at the same lead time that they send the action to decrease power.

All that matters is that the trainer responds consistently at the start and stop of the interval.

If they send their command 5 seconds early and the trainer takes 3 seconds to respond it would be something like this for a 60 second interval

0:00 - command sent
0:03 - you are at target power
0:05 - interval officially starts
1:00 - command sent
1:03 - you are at rest target power
1:05 - interval officially ends

So you spent 60 seconds at the target power, it just didnā€™t align perfectly with the command.

you only will run into problems when the response time of the trainer is different at the start of the interval than it was at the end of the interval

Yup, the only real difference is that the ā€œintervalā€ as defined by the workout and your ā€œactual time in zoneā€ may differ by the time shift that occurs as the start and end of the interval.

If you look purely at the interval set, you think you are missing out. But in reality, with the lag at the start and end, you get the same effective loading the body.

The only time this is debatable is on super short intervals around 15 seconds or less. Some trainers arenā€™t able to hit a peak resistance in the narrow time window. The common recommendation for that is to use ā€œResistanceā€ mode on the TR app, and shift to hit those power targets, just like you would on a standard (dumb) trainer.

As to the Elite vs Wahoo, and the TR 2 second pre-send command, I know my Wahoo Kickr 2017 and CycleOps Hammer hit the peak at the right time, so that jump in time is about perfect. For those that differ, you just get the scenario I described above, which is not ā€œbadā€ when you see what really happens in the workout.

1 Like

OK, I get that there is a delay at each end, so the interval is the same overall. Thanks.

The issue for me is knowing what the marker is to increase power, especially if I want to keep a similar cadence for the higher power piece. Is it when the segment reaches zero, 2 second after, 4 seconds after, when I feel it come inā€¦ I was being fooled by not appreciating the delay, getting to the segment start and the power jump, starting to increase power and just spinning up until the resistance eventually came in.

OK, so I read your response that all the segments in TR have a 2 seconds pre-send command. That answers my question, between kickr & Direto. People are telling me the kickr delay is around 1-2 seconds. I am experiencing 3-4 seconds on the Direto, before it starts its transition. So now I know what is going on. That explains your 2 seconds Pre-send.

Chad, on the swap to resistance mode. Do you do that part way through a workout, eg the Spanish needles one, (or even Baird) when you reach the short peaky bits. Then convert back. Is that the trick?

Cheers (By the way Chad I am impressed with how much time you devote to answering all these questions. :slight_smile: do you ever sleep/train?)

1 Like

Any excuse to get a KICKR Core is legit in my eyes.

3 Likes

It can depend on the workout profile and frequency of the tricky intervals. With something like Spanish Needle, I would tend to just switch to Resistance mode for the entire workout and shift manually, along with cadence changes, to get the best performance.

But some may work better by swapping modes.

  • If you are on PC, you can quickly swap between modes by using the ā€˜Mā€™ key to cycle through.
  • If you are on Mobile or Mac, itā€™s more of a pain right now. You have to go into the trainer in Devices and swap there. Not fast or easy, so those would be a full swap for the whole workout or I would just live with the shortcoming of ERG if there werenā€™t that many to handle.
  • LOL, I do train about 6-10 hours a week depending on the plan I follow at any point in time. I am able to squeeze in forum time on breaks, lunches, and while recovering. I also do some work on them during easier Z1-2 workouts. Itā€™s fun and I like helping where I can.
  • This is an amazing group with incredible knowledge and willingness to support others. Iā€™m happy to be a small part of that :smiley:

For short intervals like for 15 seconds on/off as in Spanish Needle, the average power may be a few percent lower but as they are 150% FTP intervals, doing them at 145-147% doesnt make much difference and you are still in the ballpark of the system being targeted.

Being a little bit below for me is preferable than shifting up and down manually in resistance mode for such short but intense intervals. What a pain in the balls (pardon my french) shifting up and down would be, it would ruin my vital rest intervals!

1 Like

My trick for that is using the big-small ring swap. I can usually find a spot on the rear that works with the big for the ON, and small for the OFF. Makes for quick changes compared to rear shifting.

2 Likes

I have a Tacx NEO and I have a similar issue. Not bothering too much, but maybe I should ?

Basically, on short sprints, like the one I did today - Fuji, 20 secs at 200% FTP, I notice that:

  • the resistance seems to adjust immediately
  • the yellow line (actual power) is adjusting only 3-4 seconds after (itā€™s gradually going up to 200%)
  • it seems to ease up 1 second before the interval end (this one is strange ?!)
  • all this causes my average power to miss the target by 20 watts on average

Frankly, I think itā€™s not that important, but it sure is a bit frustrating.

I have a Tacx Neo, too. Intervals consistently start 2 seconds early (which sometimes still surprises me) and end 2 seconds early. The latter also surprises me, but usually in a positive way. :slight_smile:

Especially at the start of each interval the target power is way off (too high or too low) and will only slowly move to the desired target.

Nothing that frustrates me much but seems like an odd QA glitch in an otherwise flawless product.

1 Like

Looking at your 1st interval zoomed in:
image
Looking at the yellow ā€˜Powerā€™ line:

  1. You can see the slight delay as the trainer ramps up resistance.
  2. It overshoots, which is common, and then settles into a stable finish.
  3. Then it extends just a few seconds past the blue interval block and drops.

All of this is pretty much intended and acceptable. The shifts we see are nearly impossible to remove entirely.

  • It reduces resistance early for the reason I listed above (because it starts early), do the overall ā€œeffective resistanceā€ is done for the same overall time period.
  • What you experience on the Neo may well differ from other trainers, because it uses a ā€œvirtualā€ flywheel compared to a real flywheel on every other trainer.
  • That flywheel difference may be the outlier compared to the more consistent (and possibly larger) delay we see on real flywheel units.

Yeah. But the yellow line on the screenshot seems to be going down 1 second late, while in reality the resistance is definitely easing 1 second before the end of interval.

Now, Iā€™m really not worried about it - as @tigloo it also feels like the resistance goes up a bit before the interval start, although we donā€™t really see it in the yellow line. So all in all, the work done is probably good, just that Iā€™m a geek - I like when the numbers match :slight_smile:

2 Likes

ANT+ / BLE limitations/lag/delays? Refresh rate of power meters/trainers isnā€™t all that quick. For the 2INPower from Rotor to give proper pedal analysis for their Torque360 software it has to switch to their own 50Hz mode.

1 Like

Here are two screenshots from yesterday.

First screenshot, green line starts early and I can feel the resistance increase at that moment. Cadence drops momentarily, as I try to catch up, commanded power is way higher than target power, then converges.

Second screenshot, same workout, one interval later: I try to anticipate the power increase by early increase in cadence. This however results in commanded power to stay a lot lower for some seconds until both converge again.

@PawelKozela, had the same thought, ā€žshouldnā€™t those numbers match?ā€œ and then ā€žoh crap, thereā€™s work to do, letā€™s keep those legs moving!ā€œ

Repeated at every interval. :wink:

Agreed. Over the course of the workout, this is insignificant.

What I do think though, is that TrainerRoad should have a option to teach your trainer how to interpret the data coming from the trainer/power source. Ride a special trainer workout that then has the software interpret how best to interact with your specific trainer.

Other than that we just embrace the small discrepancies and look to the average.

I have the cheap Tacx Flow Smart and it does the same thing.

I do this and it works to great effect. Additionally, on the Flow Smart it is a ā€œmustā€ in order to combat the wattage floors and ceilings.

2 Likes

Perhaps the below screenshot will help, I have a Kickr Core and I have my power smoothing on 0 sec.
I did Ebbetts in ERG mode the other day; "Ebbetts is 4x8-minute intervals between 88-94% FTP with 5-second high-power tags between 150-180% FTP.

4-minute recoveries separate the intervals."

image

By this point I was already quite tired so I might not have been able to hit the target powers perfectly anymore, but if you are better at hitting the power targets than me (which isnā€™t that hard haha) then you might see smoother lining.

The below screenshot is from the second interval, that looks far smoother:

Edit: I ride by the way with the middle ring up front (got three upfront) and middle at the back.

I have a Tacx Vortex and have the same issues. Often the whole workout will be slightly under the target watts causing calories, TSS and IF score to be a tad lower than prescribed. I found adjusting the total workout to 101% seemed to even things out.

For clarity, you are using 0 seconds smoothing in the TR app, but it is obvious by the artificially smoother power line, that the Smoothing in ERG is enabled in the Wahoo Core.

It is on but default for all Wahoo trainers, and makes overly clean power lines, as shown in your pic. You can use it that way, but be warned that if you switch to a different brand trainer or get a real power meter, the data will look VERY different (nowhere near that smooth, even with smoothing turned on in the TR app).

Bottom option.

1 Like

Ahh! Thanks for letting me know, still blue in the smart trainer stuff.

I think Iā€™ll put it on off so that it really shows how my workout has gone.

1 Like