Speed Display BUG: Head unit shows GPS speed, although the fit profile records sensor speed.

ahh, it is CH, the surface of the roads is perfect and on that bridge too, btw i found a dropout in the FIT file, before the bridge, after the pizzeria, at the Laufohrstasse 4 , there the speed was coming from GPS for 7seconds.

yes, the 1sec recording is recommended, You should get much better speed chart with that because the smart recording is using 5-8sec recording interval mostly and the speed chart wont be smooth (on that bridge the gaps between recording positions were 6-7secs long)

As the front sensor works with the earth magnetic field, crossing bridges with a metal structure, can affect the sensor.

I had this too, on bridges with mostly metallic structure.

I switched to 1 sec recording, but the displayed speed still fluctuated for a few seconds on that bridge in all four rides. I checked the 1 sec recorded data by Garmin BaseCamp, and strange enough, the recorded speed points are very smooth where I observed the fluctuations, but . The only difference made by 1 sec / smart recording is the file size. 1 sec recording makes the file over 2X larger, while I didn't feel any difference on the displayed speed. One of the fit files if you are interesed.

https://drive.google.com/file/d/1hofKEL4lr8sHVu7Q75dW-3rOhC2LatQP/view?usp=sharing

i checked, the overall speed chart is much better (more datapoints=much better speed chart, but it needs a visually good viewer app too like the Strava's charts), i dont use basecamp but i suppose that has no good details. but yes, on the bridge the speed values are wrong again, maybe Zico's tip is right and the metal structure is the problem. Dont worry about the file size, the 830 has "unlimited" space 

This is quite strange, as data points are listed as a table instead of a chart by double clicking the file in basecamp, and it gives me a series of smooth speed points on the bridge. I have no idea what trick basecamp is using because it should show the raw data list by double clicking. The metal structure may be a problem.

I have multiple magnets on the front wheels of my bicycles, because I need them to calculate short term acceleration. Background: I use a Velocomp/iBike Newton+, and the more dense speed data Newton gets, the more acceleration data it can calculate, thus the grade reading is more accurate.

(note: it calculates instantaneous grade using NOT barometer as a primary source, but other internal sensors, gyroscopes and accelerometers, and the data of those are cleaned  with the acceleration of the bicycle itself)

Eg. now I have 4 magnets on my trekking bike with 32 spockets, and certainly the wheel circumference is set to 542 mm everywhere, since the real one is appr. 2168 mm.

I use Newton, Edge 530 (same sw as a 830) and a Bryton 410 when riding.

Officiallly all of them prefer speed sensor data to gps speed data. In case of Newton it is evident, because it is a cycling computer without gps receiver:-), in case of Bryton there is a setting to choose the primary source of speed, and Edge is simply said by everyone to prefer the sensor data.

And I can confirm that there is a much bigger speed fluctuation in  case of Edge vs Bryton/Newton.

Evidently there cannot be any problem with the placement of one or more magnets, because in this case the sensor would send incorrect data to all the three cycling computers.

And the extent of fluctuation is less than 25% of the actual speed, so we can exclude that there is a problem with one or more magnets from the four.

It is like Newton or Bryton show 17.5 km/, 17.8 km/h, 17.6 km/h etc, and at the same time Edge 17.4 km/h, 16.1 km/h, 17.9 km/h.

Sometimes I think that it is the problem with the preference order of gps vs sensor, sometimes I feel that it is simply a question of the power of CPU. Rolling at 15.6 m/h means 4.33 m/s, so 8 data per second (4.33 divided 0.542), while at 23.4 km/h means 12 data per sec.

Thus even at low speed there are pretty frequent data, and this frequency would be possible only above 62 km/h in case of one magnet.

I dont know who calculates speed using the raw data, the speed sensor itself, or the head unit, but I tend to think that the cycling computers, otherwise the speed data on the computers would be really the same, maybe with just some tiny time lag.

I shared my story only to underpin that somethin is wrong with 530/830 as regards displaying actual speed.