Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
If it was hardware, wouldn't all the 5s's show the same reading or at least the ones from the same batch?

I think it's software only because I've been able to get the phone 'stuck' at weird angles such as 32° while rotating it quickly then have it finally settle to an incorrect 4° when placed on a flat surface.
 
I can't believe all the wishful thinking in this thread. This is a HARDWARE issue. They went to a crappy part this time. Its been discussed. Apple is NEVER going to admit that the 900billion phones sold are defective. They are trying to figure out how to hack the code to make it look like it works but are having a tough time because the new part is messed up in variations. They are already fixing in production but that takes weeks to get to your store so take you phones back now and wait, that's what I did.
 
Okay if that is what he has explained then it is mfg issue and a big deal IMO.

It's really not. Software traditionally does, and should, compensate for these expected differences.

That's right, expected.

For a manufacturing process to conform to zero tolerance:

A) has never been achieved
B) would not result in a device affordable to any consumer.

I am a hardware and software engineer with extensive experience in the semiconductor industry.

Discount my opinion as you wish.
 
It's really not. Software traditionally does, and should, compensate for these expected differences.

That's right, expected.

For a manufacturing process to conform to zero tolerance:

A) has never been achieved
B) would not result in a device affordable to any consumer.

I am a hardware and software engineer with extensive experience in the semiconductor industry.

Discount my opinion as you wish.

What's your informed opinion on the original 16-bit sensor being downgraded to a cheaper 6-bit one with higher noise? What kind of accuracy hit would you expect this to incur? Any other complications?
 
I can't believe all the wishful thinking in this thread. This is a HARDWARE issue. They went to a crappy part this time. Its been discussed. Apple is NEVER going to admit that the 900billion phones sold are defective. They are trying to figure out how to hack the code to make it look like it works but are having a tough time because the new part is messed up in variations. They are already fixing in production but that takes weeks to get to your store so take you phones back now and wait, that's what I did.

900billion phones sold??? Wow, I can't wait to see my next Apple dividend check! :p

But seriously, I would add to a couple of your statements, as follows:

"They are [RUMORED TO BE] trying to figure out how to hack the code..."
"They are [RUMORED TO BE] already fixing it in production..."

The fact that some people are reporting that unnamed "techs", "senior techs", "executive contacts", and "supervisors" (LOL) have made various off-the-record statements to individual customers, some of which contradict each other by the way, does not confirm that Apple is doing ANYTHING AT ALL.

I hope they are. I suspect they probably are. But I also have some concern they may treat this just like the Iphone4 antenna problem, meaning they are just preparing to say it is "within specs" and tell everyone to live with it or give them a some $15 consolation prize instead of fixing the problem.
 
I can't believe all the wishful thinking in this thread. This is a HARDWARE issue. They went to a crappy part this time. Its been discussed. Apple is NEVER going to admit that the 900billion phones sold are defective. They are trying to figure out how to hack the code to make it look like it works but are having a tough time because the new part is messed up in variations. They are already fixing in production but that takes weeks to get to your store so take you phones back now and wait, that's what I did.

Even so, if a particular phone is always registering "off" by the same amount, it can be fixed by software, even if it requires some manual calibration.
 
What's your informed opinion on the original 16-bit sensor being downgraded to a cheaper 6-bit one with higher noise? What kind of accuracy hit would you expect this to incur? Any other complications?

What are the two part numbers so I can lookup spec sheets
 
So another update. So since my replacement had same issues just a bit better than my old iPhone 5s. I called in and I talked to senior advisor. He says since I've had my phone was replaced and had same issue. He would take it up with engineering. And said he thinks it may be a hardware issue since my gfs iPhone 5s is perfectly leveled and the fact that all other iPhone 5s gyroscope ranges are different. He told me to wait for a software update and see if it fixes the issue. So have to wait now. He gave me his contact info so I can directly call him with an issue.
 
To all the software and hardware engineers

What do you think of this new Bosch designed chip that Apple presumably built into the iPhone 5s:
BMA 220

http://newsletter.spezial.de/home/article/146/digitale-beschleunigungs-sensoren

Could it be part of the problem or is really just a software issue.

As I mentioned earlier in a post - I am not a technician but dropping the resolution from 16bit (iPhone5) to 6bit (iPhone 5s) sounds to a layman as pretty drastic choice.
I am working professionally with photos and 8 bit images are significantly lower in quality than 16 bit. But I dunno if I am allowed to make such comparisons.

I am wondering if they did it
a) for monetary reasons only (and took a chance for a lesser perfomance) ? OR
b) because the other sensors together with the M7 compensate for its (presumed) deficiencies ?

NOTE
a poster called gs28 before me said the iPhone 5/5C uses the the Bosch BMA180 -> to my knowledge that is not correct.
The iPhone 5 uses a chip from another manufacturer (forgot its name) but it had 16bit resolution.
But it is said that the 5s uses the Bosch BMA220 which is the cheapest chip on the link that I also posted here.

So the questions still stands if resolution could make this kind of a difference in the measurements ?!
 

First, let me just point out that resolution is not equal to accuracy.

Second, the BMA220 has a higher bandwidth (lower noise peaks) and selectable bandwidth (flexibility of software/firmware to config for an application.)

I can't find Bosch spec sheets. All links I have found are dead.

My guess is the BMA220 is actually better suited for how it is being used, but of course I don't have access to the iOS source to see how it is interfaced, or the spec sheets to see what integral filters, if any, are present on each.

There are a lot of good articles out there describing what each spec contributes to suitability and accuracy for a particular application.
 
Links to iPhone 5s "ingredients"

http://www.chipworks.com/en/technical-competitive-analysis/resources/blog/inside-the-iphone-5s/

here is an extract from the site:
bildschirmfoto20.png


of course I cannot garantuee these guys are right. ;)
 
First, let me just point out that resolution is not equal to accuracy.

Second, the BMA220 has a higher bandwidth (lower noise peaks) and selectable bandwidth (flexibility of software/firmware to config for an application.)

I can't find Bosch spec sheets. All links I have found are dead.

My guess is the BMA220 is actually better suited for how it is being used, but of course I don't have access to the iOS source to see how it is interfaced, or the spec sheets to see what integral filters, if any, are present on each.

There are a lot of good articles out there describing what each spec contributes to suitability and accuracy for a particular application.

Hey look what I found (thank god... Didn't feel like typing all of this:)

http://www.lionprecision.com/tech-library/technotes/article-0010-sensor-resolution.html
 
I doubt this calibration method would work on my phone as I have different offset values for each side of the phone in addition to the back.

If I put my phone with the home button on the bottom in the level app and tap the screen and the rotate is 90 degrees clockwise so the home button is on the left, the display reads 85 degrees. It's not accurately determining the rotation.

As such it would need 5 separate calibrations.

No, it'll work, I promise. The accelerometer only gives you three numbers. With the phone flat on its back you can read the acceleration along two of the axis which have no gravity force along them, and when you tilt the phone to the side, you then calibrate the axis coming out the back of the phone. If you introduce calibration to these raw values then the resulting orientation is correct in all orientations of the device.
 
I can't believe all the wishful thinking in this thread. This is a HARDWARE issue. They went to a crappy part this time. Its been discussed. Apple is NEVER going to admit that the 900billion phones sold are defective. They are trying to figure out how to hack the code to make it look like it works but are having a tough time because the new part is messed up in variations. They are already fixing in production but that takes weeks to get to your store so take you phones back now and wait, that's what I did.

Here...$.02...you dropped this.

Why wouldn't you just wait to take your phone back instead? Wow.
 
Here's an example of why this can't be fixed in software with a blanket fix. Notice that the difference between 90 degrees as reported by the phone is 96 degrees.

http://www.youtube.com/watch?v=xGbhB75bHew

this video is pointless. moves it way to fast...doesn't make it level since he/she is resting it on the buttons...its in a case.

The real test is, does the phone register the same reading on both long edges and both short edges? It does on mine.
 
iPhone 5s accelerometer/gyroscope/compass values incorrect

The real test is, does the phone register the same reading on both long edges and both short edges? It does on mine.

On mine it does not. The values are 4 off from each other taking into account the buttons.

The top and left have the same values and the right and bottom have the same values.

And you can pause the video.
 

That's what I needed. No doc for the other sensor to compare to but not necessary.

This is a well suited sensor. It even has a sample configuration in the document for gaming application specifically.

This document also potentially explains the glitch/jump I see on the live wallpaper, parallax on the home screen, and in sensor data on app launch.

There is a likely different settle time on wake than the other chips (or the other chip did not sleep and so had no settle after power up,) and also the self test bit may be getting set in error for one or all axis, possibly related to 64-bit vars, and is not being compensated for since the data is imposed on top of the test signal.

I would lean farther toward driver/software interface being the issue.
 
That's what I needed. No doc for the other sensor to compare to but not necessary.

This is a well suited sensor. It even has a sample configuration in the document for gaming application specifically.

This document also potentially explains the glitch/jump I see on the live wallpaper, parallax on the home screen, and in sensor data on app launch.

There is a likely different settle time on wake than the other chips (or the other chip did not sleep and so had no settle after power up,) and also the self test bit may be getting set in error for one or all axis, possibly related to 64-bit vars, and is not being compensated for since the data is imposed on top of the test signal.

I would lean farther toward driver/software interface being the issue.

ok...:D
I think I understood the last sentence..

So do you think it is something that can be fixed with a software update ?
 
I doubt this calibration method would work on my phone as I have different offset values for each side of the phone in addition to the back.

If I put my phone with the home button on the bottom in the level app and tap the screen and the rotate is 90 degrees clockwise so the home button is on the left, the display reads 85 degrees. It's not accurately determining the rotation.

As such it would need 5 separate calibrations.

You stated "the level app and tap the screen", what level app? If you're talking about the built in compass level and when you tap the screen it goes to 0, don't tap the screen for it will only offset to 0 and the difference will be off more on the other sides.
 
I can't believe all the wishful thinking in this thread. This is a HARDWARE issue. They went to a crappy part this time. Its been discussed. Apple is NEVER going to admit that the 900billion phones sold are defective. They are trying to figure out how to hack the code to make it look like it works but are having a tough time because the new part is messed up in variations. They are already fixing in production but that takes weeks to get to your store so take you phones back now and wait, that's what I did.

What in the world are you talking about? You don't know crap and you're not helping the cause. Apple employees don't even know what's going on so why do you feel that you know it all.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.