Wildly inaccurate floor count

[RoxysDad]

I upgraded my Alta HR to the new charge 3 because I wanted a waterproof device however in the last 7 days it has recorded a total of 160 floors ascended!!  And in one day alone I apparently climbed a total of 49 floors. 

Now I’ve seen the explanation about the altimeter using barometric pressure to “count” floors climbed but let’s put this into perspective. 

• I live in a single storey house
• i walk my dog twice a day on a level route
• i swim for 30 minutes twice a day in my lap pool

There are no stairs... however a change in atmospheric pressure from 0 (sea level) to 500ft (152m) is about 0.2psi or 2 kPa, whereas when I’m swimming, the change in pressure can be as much as 1.45psi or 10kPa which is 5 times more than when out of the pool. 

I don’t think Fitbit have factored this into the equation as my stairclimb always increases when I swim. 

[Mikechem]

My weekly floors range from 400 - 800+ a week. 

They changed the Fitbit once and it worked for about two weeks before being pointless and an embarrassment to the device. 

More importantly went through the help desk and gave all my details and then was cut off. Not attempt from them to reconnect. 

The lady could also not tell me if the false floors also affects my calories, steps or health reading. 

Annoyed you can’t turn off the function that clearly by the number of complaints is not reliable enough to be part of a device that has a primary function of informing the user of their health stats. 

[Mikechem]

telling someone to manually remove floors on a daily basis is a rubbish response to a fundamental flaw in your product. You might as well tell them to keep a note pad with them. How you can offer a function that tracked me doing 208 floor in one day shows your technology if not robust enough. To then not be able to deactivate the function or give a straight answer as to how such a reading would impact my other stats is also very, very poor. 

Moderator edit: format

[SunsetRunner]

---

@MrMarv wrote:

@Av84fun - 

I propose that on a walk that lasts over an hour on a "basically level" course that you may be able to raise up and down enough to have your devices accumulate enough 10-foot increases which is what Fitbit defines as a floor. What Fitbit hasn't said is how long they allow that 10 feet to be reached before resetting the counter, they only say they only count elevation gains - walking down a flight of stairs doesn't count.

---

Hi @MrMarv,

My suggestion to Fitbit is to put some intelligence behind it. If I am logging an activity, such as a hike or a bicycle ride, the floor count should not be included. Elevation is then part of the activity and calories burned are on acount of the activity/heartrate combo. If I don't log an activity and it is just daily routine, it should count floors when in one and the same location (no distance, no active movement of the wrist).

It would be good if Fitbit at least tells us whether the elevation/floor count is or is not part of any calculation, or if it is "just" an indicator of elevation.

[Mello3]

Honestly I dont just want to 'delete' stairs off my device. I would love for a way to manually track it. If I was working on levels then maybe I could set the device like the other exercises and then at that time when you press on stairs it would record the levels instead of just every day all the time. 

[Mikechem]

Their literature speaks of a continuous climb. Agreed, no time is given, but the implication is a purposeful stairs derived action. 

I propose that all these ridiculous floor counts come from a broken sensor or a break in the link to the algorithm. The result being any movement is tracked as altitude without a time reset or accounting for downward movement. 

This is backed up by looking at the breakdown of the floors climbed. In a 15 minute period I climbed 14 floors. I was in a pool playing with my young son lifting him onto a lilo and general gentle swimming strokes. 

I just find the lack of any definite response from Fitbit as to what is happening so very poor. 

[20thcenturyfox]

I get that manually removing floor counts must seem archaic to modern urban warriors.  But as a former pilot, very familiar with the operation and limitations of an altimeter--including the need for constant adjustments to account for atmospheric pressure changes & ambient temperature changes--I shake my head at everyone's expectations that some little gizmo based on an altimeter plus accelerometer is going to keep perfect track of such a fine movement as a stair step without regard for air pressure and temperature.  

Oh well, I guess Fitbit is partly responsible for creating these crazy expectations people seem to have.  I was fascinated and impressed that the technology had developed to the point where it could estimate a floor count in real time with any degree of reliability.  But the first time I saw it showing the floor count climbing over 100 in the first hour of the day, I looked out the window and realized immediately what was probably happening, googled "wildly inaccurate floor count", and voila! I arrived here.

Now it just provides some ongoing comic relief to hear everyone's outrage that there must be a "defect" in the device.  There is no defect.  But it would be helpful if Fitbit did more to educate its customers about the predictable limitations of the altimeter.

Trust me, there are no old pilots who think it's too much of a bother to keep adjusting their altimeters!   

[Mikechem]

I agree. I’d like to remove from my dash board but the Fitbit team won’t tell me if the 100+ floors I “do” a day count to my fitness rating, my calories burned or my steps. 

If I knew even a rounded number of x I could just do the quick sum in my head. Until the give an answer, I can’t remove the thing from my dash board in confidence. 

[DCFitness]

I think the expectation was created by many months of the counts working just fine and as expected (on my device at least). Not perfect I’m sure, but neither off by a factor of 5 or 10 as they have been for these past couple of months.

It sounds like an equipment failure to me, but they are claiming it’s software. I guess we’ll see. I ordered a new versa 2 and if it does the same thing from the gate, I’ll return it.

[RoxysDad]

Since it’s winter here in Australia, I’m not swimming at the moment, and I have updated my Charge 3 to the latest upgrade, I can confirm that my stairclimb shows reasonable numbers.
The question is. Will it revert to the wildly inaccurate numbers next month when I get back in the pool.
Watch this space.

[RoxysDad]

Read my post regarding water pressure

[SunsetRunner]

I posted this yesterday, but a moderator deleted it, without indicating why this message should not have been posted. I believe it is an honest response to another community member about the altimeter in the Charge 3 and it outlines why Charge 3 users are complaining. My question to Fitbit and the moderators of this forum is to give an answer, instead of a workaround. 

Hi @20thcenturyfox,

---

Oh well, I guess Fitbit is partly responsible for creating these crazy expectations people seem to have. (...) But it would be helpful if Fitbit did more to educate its customers about the predictable limitations of the altimeter.

---

Indeed. That is the whole point. Most people that complain about it, switched from a Charge 2 to a Charge 3 device. With the "upgrade" came the expectation that it would give more and better and faster. It used to give a fairly consistent and relatively correct value for floors/stairs count and people relied on that. But at some point, the figures started to change drastically. And Fitbit did not come back with a plausible reason, nor an explanation of the limitations of an altimeter. Their answer is simply: negate the problem by manually adjusting your logs, using a tedious workaround. If it would have been this buggy from the start, people would dismiss it. But it has been one of the features that Fitbit actively promoted as achievement and to earn badges with (https://blog.fitbit.com/fitbit-badges/).

As a "non-pilot", I didn't know and I wasn't looking to find out how accurate an altimeter is, or how it works exactly. I simply looked at the value, thought back about the day, and said: "yes, that's about right". And I did that for over two years. Walked up and down stairs, climbed mountains, drove a car, took elevators and escalators, rode my bike... And never did I think I got a weird value. Well, once (one time, 1, uno, een, eins), when I did a steep mountain hike and the Charge 2 classified it as stairs.

And only after having changed to the Charge 3, and only after a couple of months of using it, suddenly it gave out weird readings. Simple thought: something changed, and it wasn't me. Options: hardware failure or software bug. Or simply a design flaw. Whatever it is, I just want Fitbit to explain to me what it is and whether they think they can fix it, so I understand the implications.

Happy user with consistent floor count, until Charge 3 in 2019.

[20thcenturyfox]

@SunsetRunner   I responded to your post last night, and strangely my response also seems to have disappeared...even though I would have thought it might be more helpful than critical to Fitbit.  It relates directly to why these problems might be more acute in the summer.  Here is what I wrote:

I realize no one is interested in how and why pilots have to make adjustments to their altimeters to keep the ground from coming up and smacking them in the fusilage.  But one old expression comes to mind when people here report on Fitbit floorcount anomalies related to swimming or driving in the car on a warm day, and that is "High to low, look out below," reminding us that the ground will be closer than indicated when moving from areas of high to low ambient temperature, as well as atmospheric pressure.   

In localities not normally prone to windy conditions with wild pressure swings, I wonder if there could be a temperature aspect to these excessive floor count readings, as when one is moving from a hot sidewalk into an air conditioned car or building on a hot day, or one's arm is moving through warm air and then plunging into cold water.  

Does a sudden drop in temperature--just like a sudden drop in pressure-- fool the Fitbit into registering a "floor?"  For those of you experiencing this, does this seem like a plausible factor or explanation?  

Since currently the Fitbit has no way of sensing temperature changes, I'm not sure how they would adjust for this now, but a fuller understanding of the phenomenon will surely lead to improvements in time. 

[SunsetRunner]

Hi @20thcenturyfox,

Thanks! I re-read what I wrote and I may have sounded a bit unappreciative, which was not what I intended. Your feedback is very welcome! I am interested in the why and how, but I just wasn't considering it when I complained about the misreadings of the Charge 3. Your explanation makes very good sense.

I wrote something similar in another thread (https://community.fitbit.com/t5/Charge-3/Incorrect-floors-being-tracked/m-p/3632167/highlight/true#M...).

Looking back, I tried to match the air pressure and the temperature in an attempt to understand it. I made some screenshots and added both hPa and Celsius in the captions. What you say makes sense to me. Sitting in a car with the airco on (cold air flow), while the temp/pressure outside is higher, may trigger this behaviour. Before the summer time, it seemed more stable.

The biggest differences between the Charge 2 and 3 seem to be the SpO2 sensor and the waterproof case. So maybe, the hardware reacts differently and the software algorithms have not been tweaked to compensate. And maybe, Fitbit understands that the design may have an unwanted side-effect in abnormal floor counts.

If you say that pilots need to re-adjust all the time and that this is a common thing, I would assume this is also known to the people incorporating an altimeter into a device to offer a stairs/floors count feature. Would we have a tech spec's list with which to compare Charge 2 and 3 sensors and build features, we could deduct what is the most plausible cause. But then: I think Fitbit already knows, and decided it can't say anything about it, because it may possibly have users put in a claim. And moderators can't answer until a real fix is possible. Which may be never, if the design simply won't allow for an upgrade that solves the issue.

If a moderator deletes a comment, or threads of comments, I assume they have read it. Which pops up the question whether what we're writing here is seen as helpful or rather as unwanted. I just can't understand why nobody at Fitbit is reaching out then to really help us out in understanding. Which to me is taking a customer seriously. I see that as good customer service with an open and transparent attitude.

[SunsetRunner]

For reference:

One and a half hour into my day. 22 floors already. And, now that I’m looking at it, also 2km. Hmmmm. Seems a bit off. 

18c and 1017hPa. 

 

[RoxysDad]

Have you upgraded your firmware??
I’m not swimming at this time of year and since the upgrade, only seeing 1 or 2 floors a day.

[Mikechem]

I’ve done it to my phone. I’ll need to link it with my device today. On a mere 72 floors today...

Sent from my iPhone

[SunsetRunner]

---

@RoxysDad wrote:
Have you upgraded your firmware??
I’m not swimming at this time of year and since the upgrade, only seeing 1 or 2 floors a day.

---

If you refer to version 20001.63.5: it is running on my phone for quite some time already. It does not seem to solve the issue. The release notes don't really explain what has been changed.

[SunsetRunner]

 

[DCFitness]

My new Versa 2 came in the mail today. I’m mostly underwhelmed with the new features, but will keep everyone on this thread up to date as to stairs accuracy in week 1. 

[DCFitness]

 Day 2 with the new device and already a very clear difference. My old Versa had a broken altimeter despite what customer service tried to tell me. The new Versa 2 works just fine. I'll confirm this trend in a week and then be back in the spring when the pool opens again to see if the new Versa 2 altimeter craps out like the other one did (at 1 year and 1 day).