# Strange varying time offset in exported FIT files

**URL:** <https://www.trainerroad.com/forum/t/strange-varying-time-offset-in-exported-fit-files/52813>\
**Category:** TrainerRoad Software\
**Tags:** bug\
**Created:** [February 9, 2021, 7:40pm UTC](https://www.trainerroad.com/forum/t/strange-varying-time-offset-in-exported-fit-files/52813 "2021-02-09T19:40:54Z")\
**Posts on this page:** 8\
**Page:** 1

<div class="post-metadata">

**Author:** ![willvousden](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/willvousden/32/63978_2.png) [@willvousden](https://www.trainerroad.com/forum/u/willvousden)\
**Post date:** [February 9, 2021, 7:40pm UTC](https://www.trainerroad.com/forum/t/strange-varying-time-offset-in-exported-fit-files/52813/1 "2021-02-09T19:40:54Z")

</div>

While investigating [another problem](https://www.trainerroad.com/forum/t/kickr-v5-power-discrepancies/52709), I noticed that the timeseries in the FIT files exported from TR have some sort of drifting time offset when compared with the data I get from Zwift and my ELEMNT for the same session.

Here’s an example: [DC Rainmaker Analyzer](https://analyze.dcrainmaker.com/#/public/b77f0194-6788-4bc4-434c-e3e1a5d8aee5)

Easiest to see by looking at the power. In this case I have three power readings: TR and Zwift – both from my KICKR – and an independent one from my power2max (connected to my ELEMNT head unit). You can see that everything starts off fine, but an offset between TR and the other two gradually develops. After around 10 minutes, the TR power reading is **leading** the other two by around 20 seconds, and by the end of the workout the difference is about 30 seconds!

In this experiment TR was controlling the trainer in ERG mode and Zwift was just receiving the power reading (both running on my Mac). However, I’ve noticed the same thing in all the comparisons I’ve done of TR against other data sources.

What gives? Is this a bug in TR?

---

<div class="post-metadata">

**Author:** ![rkoswald](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/rkoswald/32/28273_2.png) [@rkoswald](https://www.trainerroad.com/forum/u/rkoswald)\
**Post date:** [February 9, 2021, 7:55pm UTC](https://www.trainerroad.com/forum/t/strange-varying-time-offset-in-exported-fit-files/52813/2 "2021-02-09T19:55:40Z")

</div>

Did you pause the TR workout or stop pedaling for 20-30 seconds? Were there dropouts on the TR connection to the computer?

---

<div class="post-metadata">

**Author:** ![willvousden](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/willvousden/32/63978_2.png) [@willvousden](https://www.trainerroad.com/forum/u/willvousden)\
**Post date:** [February 9, 2021, 8:17pm UTC](https://www.trainerroad.com/forum/t/strange-varying-time-offset-in-exported-fit-files/52813/3 "2021-02-09T20:17:36Z")

</div>

Aha, auto-pausing. That’s it. I’ve been doing speed work and ITL recently and both of these cause TR to auto-pause. I knew I was setting myself up for embarrassment by mentioning the B word.

Still, shouldn’t this pause show up as a gap in TR’s data? If I pause my ELEMNT mid-workout, I can see it quite clearly in the data: [DC Rainmaker Analyzer](https://analyze.dcrainmaker.com/#/public/031e3b23-6a35-4397-5543-9719c5b23f24)

It seems like TR is falsifying the timestamps on subsequent datapoints to conceal the pause in the workout.

---

<div class="post-metadata">

**Author:** ![rkoswald](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/rkoswald/32/28273_2.png) [@rkoswald](https://www.trainerroad.com/forum/u/rkoswald)\
**Post date:** [February 9, 2021, 9:17pm UTC](https://www.trainerroad.com/forum/t/strange-varying-time-offset-in-exported-fit-files/52813/4 "2021-02-09T21:17:14Z")

</div>

I’ve never analyzed TR data to this level of detail but when you pause and restart a workout, all of the graphics make it appear continuous. Looks like they are adjusting the timestamps after a pause to make that happen.

---

<div class="post-metadata">

**Author:** ![bbarrera](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/bbarrera/32/14734_2.png) [@bbarrera](https://www.trainerroad.com/forum/u/bbarrera)\
**Post date:** [February 9, 2021, 10:44pm UTC](https://www.trainerroad.com/forum/t/strange-varying-time-offset-in-exported-fit-files/52813/5 "2021-02-09T22:44:29Z")

</div>

> [@willvousden](#):
>
> Aha, auto-pausing. That’s it.

Yes. I consider it a bug that workout timestamps are not using a free running clock. Pausing the workout should not pause the clock.

The only workaround is to disable “Pedal to Resume” on settings. Or use a bike computer to record the workout.

---

<div class="post-metadata">

**Author:** ![IvyAudrain](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/ivyaudrain/32/47471_2.png) [@IvyAudrain](https://www.trainerroad.com/forum/u/IvyAudrain)\
**Post date:** [February 10, 2021, 8:09pm UTC](https://www.trainerroad.com/forum/t/strange-varying-time-offset-in-exported-fit-files/52813/6 "2021-02-10T20:09:59Z")

</div>

Wouldn’t hurt to touch base with the team at [support@trainerroad.com](mailto:support@trainerroad.com) so we can take a look, provide a more in-depth explanation based upon your ride log, and perhaps write a bug-issue if we think we need to!

---

<div class="post-metadata">

**Author:** ![willvousden](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/willvousden/32/63978_2.png) [@willvousden](https://www.trainerroad.com/forum/u/willvousden)\
**Post date:** [February 10, 2021, 9:02pm UTC](https://www.trainerroad.com/forum/t/strange-varying-time-offset-in-exported-fit-files/52813/7 "2021-02-10T21:02:32Z")

</div>

Thanks for the tip – I’ve shot them an email 👍

---

<div class="post-metadata">

**Author:** ![bbarrera](https://sea2.discourse-cdn.com/trainerroad/forum/user_avatar/www.trainerroad.com/bbarrera/32/14734_2.png) [@bbarrera](https://www.trainerroad.com/forum/u/bbarrera)\
**Post date:** [February 10, 2021, 9:07pm UTC](https://www.trainerroad.com/forum/t/strange-varying-time-offset-in-exported-fit-files/52813/8 "2021-02-10T21:07:57Z")

</div>

Let us know what you find out. There has been at least one improvement, but I don’t recall the details now. I turned off auto-pause a long time ago.
