Running, cadence, and too little steps. Please share your data

[suzzlo]

@juhis70 just my last run

Distance: 16k
Time: 1:34h
Cadence: 79

94min x 79 x 2 = 14852. The watch (+stryd) reported 15249

I hope this helps

[juhis70]

@suzzlo Thank you! I guess Stryd gives very accurate step count (and it overrides the step count from wrist). I think the ~400 step difference is not much at all. That is much more valid data than I get from my watch, although you have the Stryd which may affect things.

[LGoSo]

@juhis70 I already reported a similar problem and I think it is, at least, a discrepancy (https://forum.suunto.com/topic/6306/stride-frequency-discrepancy). I am always running with two watches: S5 (with Stryd) and Coros Pace (with pod). My cadence values from the Stryd (on S5) and from my Coros are always the same. The cadence value estimated from the steps on S5 divided by time are always lower than the value from Stryd and Coros.
My last run:
Stryd (on PowerCenter): 41:29h (moving time 40:50h) 6.81km 173spm
S5 (on SA): 41:28h 6.81km 87rpm (174spm) 6915steps
Coros: 41:29h 6.82km 173spm

With the steps from S5, the cadence is 167spm or 169spm depending if you take the moving time or wall clock time. Anyway it is lower and, for this run, it is quite acceptable (~4spm difference). But the difference can be higher as showed in my other post (10 spm difference and even higher on other runs). And always lower. But for suzzlo, the cadence from the steps is higher than the one reported in SA.

[Dimitrios Kanellopoulos]

Hi guys. Steps are coming from a different source than cadence. Unfortunately. Let’s seperate those concerns.

[suzzlo]

@dimitrios-kanellopoulos are not both comming from our feets?

(sorry)

[Dimitrios Kanellopoulos]

@suzzlo s5 for example has a context engine that detects what you are doing. Will count less steps ie if dishwashing etc.

[lorenzo12375]

Under Armour machina 2

[juhis70]

@suzzlo s5 for example has a context engine that detects what you are doing. Will count less steps ie if dishwashing etc.

@dimitrios-kanellopoulos So in my case the context engine fails to detect I’m running, although I’ve told the watch I’m running and the watch detects my pace is < 6/km. Does watch “tell” (pass, whatever) the context engine that I’m running, and/or my speed/pace?

Because if I would really run 10.9 km with only 6000 steps, my one step length would be 1.8 meters, and stride length would be 3.6 meters! That is totally unreal!

We don’t have to mix cadence and steps, but if another of them gives unreal results, then the watch could use better values from the other.

I’m also starting to believe that also this second watch is somewhat broken, as I get so weird (wrong) values. From apparently two separate systems (cadence and steps). And I would just like to run, not “fix my watch”…

[Egika]

@juhis70 said in Running, cadence, and too little steps. Please share your data:.

I’m also starting to believe that also this second watch is somewhat broken, as I get so weird (wrong) values. From apparently two separate systems (cadence and steps). And I would just like to run, not “fix my watch”…

I think this is an interesting topic and good finding.
Am following.
The good thing is, that you don’t need a watch to just run.

[Dimitrios Kanellopoulos]

@juhis70 well what I mean is the cadence uses the raw sensor data (frequency) to produce cadence and steps come from another input.

[juhis70]

For the last weeks I’ve focused on running. Ran over 200 km in June, that’s my new one month record.

OK, the cadence AND steps counter are broken, even if I did have strange arm movement. And I can assure you it is not that strange, normal balanced running when I’ve watched video of my running. We’ve all seen a running person with funny arm movements, but I’m not that.

Steps

All runs report nearly only half of the steps, ranging from 51 % to 57 %. This error is not linked to cadence error, their relation/ratio is not constant. Example 11.4 km run 2.7. (July 2nd): Suunto 5 steps 5860, correct steps more like 11400 steps.

Also when the watch is in my hand not in excercise mode, it misses a lot of steps. Thousands a day.

For example, yesterday I ran 3.6 km and walked 6.6 km. Me running 3.6k is around 3.6k steps. Walking 1 km equals at least 1200 steps. After 3.6 km + 6.6 km (and daily activity, which produces 1000-3000 steps at least) step count was around 9500, should have been at least 11500 steps (3.6k+6.6k x 1.2). Plus daily activity steps.

Running cadence

I’ve had two Suunto 5 watches. With the first one I noticed, that the cadence value “jumps” between correct values (rpm) and exactly half of the correct values. My running cadence is usually between 80 and 100, average for a run is around 87 rpm. So the values jump between 40…50 and 80…100. There are no random values, or “between” values (of course some “noise” in the data).

This of course ruins the cadence graphs, and average cadences. The more the cadence locks to ½ value, the lower the avg cadence. I’ve had avg cadence values 60–71 with the first watch, and 59–65 with the second watch. So the value is useless.

After about 7 runs I sent the first watch to service. They replaced it with a refurbished one. That did not make the problems go away. Actually, this second watch locks more to half cadence than the right value compared the the first watch, so the avg cadence values are lower. Also the step counting is worse (fields “step x” and “cad x” in the data, the higher the worse)

The Data

Maybe it speaks for itself

Graphs

To prove the watch is really alternating between correct cadence and half cadence, and not just throwing some random cadence values, I can reconstruct good cadence graphs from excercises .FIT file by going through the cadences, and just multiplying the half cadences by two. (I have an OpenOffice calc sheet with macros to automate this.)

Here’s 25.6. run, cadence graph from Suunto App:

And here is (from .FIT) reconstructed cadence graph, green line is the corrected cadence, and blue line is a moving average that cuts the biggest spikes:

Suunto 5 avg cadence 63 rpm, correct avg cadence 87 rpm.

My right arm has maybe “sharper” movements, so the watch picks the right cadence and cadence graph is better, but not perfect. Here’s a 2.4 km test run in 28.6. (the longest run with watch in my right hand):

Suunto 5 avg cadence 82 rpm, correct 87. That’s almost good. But the steps, Suunto 5 counted 1338 steps, should be around 2400 steps

[juhis70]

“Funny” thing about the bad steps counter in my watch: while walking, if I go to the watch screen that shows the steps, the step count increases every step (sometimes there is a lag at beginning, then it adds several steps). It does not any skip any steps, even when I hold my arm in front of me looking at the watch (i.e. not waving arm). That is nice, almost like magic.

But when the watch is not in the “steps screen”; it apparently doses off occassionally and forgets to count the steps while I’m walking. This is not nice.

[Aaurakkio]

hei @juhis70 ,

wierd thing you are facing there.

On my watch the measures correspond quite nicely and the steps from watch are just slightly below the calculated by cadence (see 5 random activies from last 6 weeks):

FYI, I keep my S5 on inside left wrist and constantly on the default view (during the activities).

Given the information, I wonder whether your issue could be either:

1. something defective in your specific watch; or
2. how you hold the watch and how you move your arm.

[juhis70]

@aaurakkio Hi, thanks for your reply and data. Your values are really accurate, I’d “buy” that, unlike my tens of percents wrong step counts. I don’t think my arm movement is very uncommon, my previous activity bracelet counted (can still count) my steps very precicely. And S5 can count steps even when I hold the watch in front of me and look at the step counter. It starts failing only when I’m not observing it…

[Harald Värs]

My wife bought the S5 few years ago and end up frustrated because she felt that no matter how much she moved during normal day she could barely get the activity/steps for the day. That felt quite odd and we changed watches (my polar to shes S5). I experienced same problems but didn’t have time to investigate any further. Now I had a change to compare directly 2 devices at the same time with Oura ring and picture shows that our earlier experiences might have some truth behind them. Now I fount this thread after searching similar cases online and feel like there might be some correlation here with @juhis70.

[Egika]

@harald-värs maybe no correlation. Your numbers should rather be the interpretation of steps.
Oura counting every single foot and Suunto the cadence (left+right foot)

[Harald Värs]

@egika Fair point, thank you.
Does anyone know then if Suunto is one of few that counts steps like this, because I’ve used to measure healthy active day around 10k mark on steps and what devices I’ve used in past are somewhat on par with data that Oura gives here. Seems just little weird for me, that’s all.
Sorry going little off topic here.

[juhis70]

@harald-värs Thanks for your reply. Steps are single foot steps in Suunto’s world, too. Every step is/should be counted.

When talking about cadence, Suunto uses revolutions (right + left foot) per minute (rpm), while many others use steps per minute (spm). Some think that ideal running cadence is 180 spm, or 90 rpm.

If Oura counts the steps correctly, there’s definitely something wrong with also your S5’s steps. At about same ratio as in mine, my S5 counts around 60 % of my steps (running and walking).

[Dimitrios Kanellopoulos]

@harald-värs Suunto 5 and Suunto 3 only have a context engine that doesn’t over count steps from washing dishes or tapping on the floor with your legs or typing. Others will. Including oura.

[Harald Värs]

@dimitrios-kanellopoulos
OK, thanks. Things start to make more sense now on my part.