iPhone 5s accelerometer/gyroscope/compass values incorrect

[SPoon222222]

Why did you swap it when there is no evidence that they changed anything in the hardware since the first one you got?

I changed it before the article so I do not know where you are coming from. I am merely stating what the article states and not my opinion. The new device I received is -3° flat compared to -6° flat on the first one I bought. The long and short sides also exhibited slight improvements compared to the original phone I purchased.

 

[dmystic1]

I changed it before the article so I do not know where you are coming from. I am merely stating what the article states and not my opinion. The new device I received had -3° flat compared to -6° flat on the first one I bought. The long and short sides also exhibited slight improvements compared to the original phone I purchased.

I just read the article after my post. But if after reading the post here and I saw that all of a sudden people started receiving phones with no problem then I would run to exchange for a newer build date. That has not been the case. I get that you spend a lot for an iPhone, but price does not mean better or perfect. Too many people are just getting carried away with exchanges and in the long run consumers will be affected.

 

[Morac]

I am tempted to go back and purchase another 5s next week (hopefully a build week 42) and see if the problem has been fixed via an hardware upgrade.

There's no point as Apple can't fix this via hardware without doing a redesign, which they aren't going to do. There's two ways to fix this:

1. Replace the new Bosch sensor with the old STMicroelectronics. I don't think this is likely to happen and there's some evidence that iPhone 5Cs may use the new Bosch sensor as many are also having the problem.
2. Add a PROPER calibration feature in software. This is a lot cheaper to do and it will fix all existing units without requiring hardware swaps.

 

[gCloud]

The high bias of the new Bosch part means your best bet is to constantly return your phone for replacement until you get one that happens to have an offset of zero.

I got loads abuse when i suggested the same, be careful.
Its time to move on i think, this issue will never be solved, if u let you return period run out you are stuck with a duff phone or a refurb. If u get one with good gyro u'll probably get a screen with bleed or interlace or dead pixels, or a unit with a duff battery, or a faulty touchId sensor it just a pain in the ass with a price tag of $1200. Enough is enough. My honeymoon with iPhone started with the 3G had every iphone since, honeymoon ends here.
Nexus 5 is out soon for half the price wind with ... Wait for it ... A file system user and apps can access, it may not be as smooth, may have its own faults, will need yo purchase loads of apps again, but hey it will be $700 cheaper.
Bring on the abuse..

 

[DCIFRTHS]

I just read the article after my post. But if after reading the post here and I saw that all of a sudden people started receiving phones with no problem then I would run to exchange for a newer build date. That has not been the case. I get that you spend a lot for an iPhone, but price does not mean better or perfect. Too many people are just getting carried away with exchanges and in the long run consumers will be affected.

Communication is key, and as usual, Apple has dropped the ball

...you can point that finger of blame (for the returns) directly at Apple. They deserve to lose money because of this situation.

 

[Juanefny]

After reading gizmodo..

I figured I would try going and seeing if I'd get lucky and get a replacement that didn't have the issue. So I made an appointment with the "genius" and when I went in he took it to the back and came back telling me that it is a known issue and that it will be fixed via software... He also said that they would not replace at this time as this is a widespread issue and the replacement would have this problem too.
Make of that whatever you wish...

 

[Nolander07]

I will all work out in the end.

 

[alFR]

Its time to move on i think, this issue will never be solved

I agree with the first part of that but not the second. Even the guy in the Gizmodo article says it could be fixed with software (and an increasing number of people are being told that by Apple), so I'm not sure why everyone is suddenly so sure it won't be. That said, I agree that if people are worried that it won't and don't want to take the chance they should return it, get a refund and wait and see. Likewise, if you need file system access etc. Android is definitely a better fit for you. Enjoy your Nexus.

 

[DMoggo]

I got loads abuse when i suggested the same, be careful.
Its time to move on i think, this issue will never be solved, if u let you return period run out you are stuck with a duff phone or a refurb. If u get one with good gyro u'll probably get a screen with bleed or interlace or dead pixels, or a unit with a duff battery, or a faulty touchId sensor it just a pain in the ass with a price tag of $1200. Enough is enough. My honeymoon with iPhone started with the 3G had every iphone since, honeymoon ends here.
Nexus 5 is out soon for half the price wind with ... Wait for it ... A file system user and apps can access, it may not be as smooth, may have its own faults, will need yo purchase loads of apps again, but hey it will be $700 cheaper.
Bring on the abuse..

I'm beginning to think this way too sadly.

 

[nanocustoms]

A larger offset equals more resolution. I think the issue is a software one and that the offset values are only going to +/- 20 instead of +/-95.

My 0.02 is that its easily fixed by expanding the range of offsets for the 5S/5C and subsequent hardware revisions that feature this sensor.

The Bosch MA220 offers power consumption savings and is has up to 5 times the sensitivity. the issue now is that the software must be using -20 to 20 instead of -95 to 95 during calibration (which should be when the offsets are recorded), which is why we are consistently seeing 3-4 degree variances.

Cheers

Chris

 

[slowturn]

Chris, you may want to reread the article again.
http://gizmodo.com/heres-why-the-iphone-5s-accelerometer-is-so-screwed-up-1445966306

A larger offset (variance) means it is not as accurate.
The old sensor measures within +/- 1 degree while the new one measures within +/- 5 degrees.

Can't believe Apple went with an inaccurate accelerometer to reduce power usage.

A larger offset equals more resolution. I think the issue is a software one and that the offset values are only going to +/- 20 instead of +/-95.

My 0.02 is that its easily fixed by expanding the range of offsets for the 5S/5C and subsequent hardware revisions that feature this sensor.

The Bosch MA220 offers power consumption savings and is has up to 5 times the sensitivity. the issue now is that the software must be using -20 to 20 instead of -95 to 95 during calibration (which should be when the offsets are recorded), which is why we are consistently seeing 3-4 degree variances.

Cheers

Chris

 

[Morac]

Chris, you may want to reread the article again.
http://gizmodo.com/heres-why-the-iphone-5s-accelerometer-is-so-screwed-up-1445966306

A larger offset (variance) means it is not as accurate.
The old sensor measures within +/- 1 degree while the new one measures within +/- 5 degrees.

Can't believe Apple went with an inaccurate accelerometer to reduce power usage.

You might want to reread it. The BIAS is not as accurate. The variance is similar to what it was, so this can be fixed in software by correcting for the bias (i.e. calibration).

 

[Khedron]

Why did you swap it when there is no evidence that they changed anything in the hardware since the first one you got?

Because of the large variations in bias if you are lucky you can get one that has an offset close to zero.

If the bias is uniformly distributed then you have approximately a 10%-20% chance of one that gives the right readings (the new part bias is 5 times the old).

 

[slowturn]

You are correct. variance is similar.

You might want to reread it. The BIAS is not as accurate. The variance is similar to what it was, so this can be fixed in software by correcting for the bias (i.e. calibration).

 

[Khedron]

Can anyone reply me?

Probably the horizon will be wrong and stars and other objects will be in the wrong position, i.e. the app has no technological benefit over an old school paper star map.

 

[Robster3]

Just realised my $500 watch has a compass

 

[Khedron]

You might want to reread it. The BIAS is not as accurate. The variance is similar to what it was, so this can be fixed in software by correcting for the bias (i.e. calibration).

Since Apple refuses to comment on the issue the prospect of waiting for every app to be updated with calibration and then every one being individually calibrated is not appealing.

It also means M7 data is useless because it cannot be interpreted "live" and any active processing of the data means the M7 low-power savings are negated.

 

[nanocustoms]

I'll be reading more into it, but if the scenario below applies, then I feel the problem is software.

------

Bias seems to be a range of +/- from the true value. Due to manufacturing variances, the chip must obtain calibration values by running calibration procedures.

Assuming the sensitivitys are equal, the range -20 to 20 offers 40 calibration points, where the range -95 to 95 offers 190.

IF the software is only spitting out -20 to 20, it would explain why few of us can calibrate correctly.

Cheers

Lets solve this MR!!!

 

[Mr. Luigi]

Astronomy/Planetarium Apps will be affected

Can anyone reply me?

All Astronomy/Planetarium apps (SkySafari, SkySafari+, StarWalk, GoSkyWatch, GoSatWatch, SunSurveyor, etc. etc. etc.) require the level sensors and compass to align the screen with the sky. If those sensors are feeding the apps compromised data so too will the apps functionality be compromised. The screen will not align properly with the sky and could/will jump around. I assure you that the developers of these apps are throwing a piss fit over this issue as it renders their apps much less attractive to buyers.

As an astronomy/physics instructor for over 35 years it is my opinion that these apps have added much to the enjoyment of studying the heavens...especially for those new to sky watching. The good old paper maps and plastic planispheres (Philips' is probably best known) still do a good job. But some do find the learning curve for them more steep.

 

[iBighouse]

I find it hard to believe that Apple's design crew and engineering team would mismatch a couple of sensors that could result in getting bad data from them. Surely they are more adept than that, so I'm hoping they come out with a software solution in an update that makes the whole "mismatched sensor" issue meaningless. If the sensors were so poorly matched, someone somewhere along the way, be it from Apple's engineering team, Bosch's, or the other sensor manufacturer would have said something to someone.

Something like "That's a great sensor for really basic stuff, but it'll conflict with your other sensors and give erroneous info based on the mismatched bias/bits/whatever."

 

[bitslap47]

This is an easy software fix.

The BIAS tolerance is at issue. That relates to the consistency of what output is given by the chip for "0" from chip to chip.

Before it was a 1 degree variance and so they made a single calibration at apple to be used for all phones since the worst deviation would be 1 degree at level.

Now it can be 5. They can no longer use a pre-set calibration.

I am betting using the other sensors on-board they can extrapolate what zero is. Worst case a one time user interaction on first run like the other set ups done i.e. Do you want to use iCloud, location services, etc.

They could even have the OS ask a user to calibrate the first time an app using the sensors requests data, like the allow access to camera roll prompt. This method would require no changes by developers.

In any case I have been able to compensate easily in my code, so Apple surely can as well.

 

[benzn]

I'll be reading more into it, but if the scenario below applies, then I feel the problem is software.

------

Bias seems to be a range of +/- from the true value. Due to manufacturing variances, the chip must obtain calibration values by running calibration procedures.

Assuming the sensitivitys are equal, the range -20 to 20 offers 40 calibration points, where the range -95 to 95 offers 190.

IF the software is only spitting out -20 to 20, it would explain why few of us can calibrate correctly.

Cheers

Lets solve this MR!!!

Nope, still wrong. If you read the datasheets, the range of the accelerometer is something like +/- 2G (2000mG), i.e. an accelerometer is pointless if you were only able to measure <0.1G. Furthermore, (from the datasheet), the resolution/sensitivity of the old sensor is small (something like 1.9mG IIRC) and the resolution of the new sensor is 62.5mg. Now, additionally, for each individual sensor you're going to add in the bias (which can be the ranges you specified) and you'll end up with the final accelerometer readings. If you look back into my comment history, you can find a more mathematical breakdown.

 

[munakib]

If there was an "easy" software fix, Apple would have fixed it by 7.0.1, i am sure they didnt need to see macrumors 70 page long thread to get wind of something being wrong.

I am confident a software fix will come in the long run but my feeling is at worst case scenario apple will ask/make it mandatory for app developers (apps which require gyro/accelerometers) to include re-calibration tools at the setup of the apps.

There is no way Tim Cook will "recall" their gorgeous iPhone5s and tarnish his already fragile leadership reputation at Apple.

This is an easy software fix.

The BIAS tolerance is at issue. That relates to the consistency of what output is given by the chip for "0" from chip to chip.

Before it was a 1 degree variance and so they made a single calibration at apple to be used for all phones since the worst deviation would be 1 degree at level.

Now it can be 5. They can no longer use a pre-set calibration.

I am betting using the other sensors on-board they can extrapolate what zero is. Worst case a one time user interaction on first run like the other set ups done i.e. Do you want to use iCloud, location services, etc.

They could even have the OS ask a user to calibrate the first time an app using the sensors requests data, like the allow access to camera roll prompt. This method would require no changes by developers.

In any case I have been able to compensate easily in my code, so Apple surely can as well.

 

[Morac]

iPhone 5s accelerometer/gyroscope/compass values incorrect

For people who are confused on variance vs bias, here's an analogy.

Say you need to measure something in inches (or centimeters), but you only have a ruler that has foot (meter) markers. That would be variance which is pretty much unfixable as it's not possible to get an accurate measurement. Now say you need to measure the same thing and you have a ruler that has inch markers, but it starts at say 6 instead of 0. You can still accurately measure, if you know to subtract 6 from whatever result you get. That's bias.

Now say you have millions of rulers and each start at a different number. The rulers can all be fixed by adjusting for the offset, but the since the offsets are not the same, the rulers will need to be calibrated individually.

That's a simplified version of the position Apple is in. Except instead of having one ruler per device, there's 5. Apple just needs to build an interface that allows individual calibration. It really isn't that hard to do, they just have to actually do it without admitting there's actually a problem.

 

[bitslap47]

If there was an "easy" software fix, Apple would have fixed it by 7.0.1, i am sure they didnt need to see macrumors 70 page long thread to get wind of something being wrong.

I am confident a software fix will come in the long run but my feeling is at worst case scenario apple will ask/make it mandatory for app developers (apps which require gyro/accelerometers) to include re-calibration tools at the setup of the apps.

There is no way Tim Cook will "recall" their gorgeous iPhone5s and tarnish his already fragile leadership reputation at Apple.

It's an easy fix. They are exercising caution as they should.