time-nuts@lists.febo.com

Discussion of precise time and frequency measurement

View all threads

Striking change in iPhone time accuracy with 8.2

AG
Anthony G. Atkielski
Wed, Apr 1, 2015 3:59 AM

Has anyone else noticed a dramatic improvement in the accuracy of time
of day on iPhones and iPads since the release of iOS 8.2? The accuracy
used to be only plus or minus 2 or 3 seconds, now it is about 100
times better, usually a few tens of milliseconds. I figure Apple might
have finally paid some attention to accurate time of day with 8.2,
possibly because of the Apple Watch. It's a pleasing improvement, I
hope it's permanent.

--
Anthony

Has anyone else noticed a dramatic improvement in the accuracy of time of day on iPhones and iPads since the release of iOS 8.2? The accuracy used to be only plus or minus 2 or 3 seconds, now it is about 100 times better, usually a few tens of milliseconds. I figure Apple might have finally paid some attention to accurate time of day with 8.2, possibly because of the Apple Watch. It's a pleasing improvement, I hope it's permanent. -- Anthony
DJ
David J Taylor
Wed, Apr 1, 2015 7:01 AM

Has anyone else noticed a dramatic improvement in the accuracy of time
of day on iPhones and iPads since the release of iOS 8.2? The accuracy
used to be only plus or minus 2 or 3 seconds, now it is about 100
times better, usually a few tens of milliseconds. I figure Apple might
have finally paid some attention to accurate time of day with 8.2,
possibly because of the Apple Watch. It's a pleasing improvement, I
hope it's permanent.

--
Anthony

---===

I noticed something similar on my iPad, but hadn't correlated the change
with iOS 8.2.  I thought it was just chance, but you could be right.  On the
other hand, I just checked now and it's 1.9 seconds off.  I compare with
Emerald Time, which includes my one stratum-1 NTP server in its sources.

https://itunes.apple.com/gb/app/emerald-time/id290384375?mt=8

Cheers,
David

SatSignal Software - Quality software written to your requirements
Web: http://www.satsignal.eu
Email: david-taylor@blueyonder.co.uk

Has anyone else noticed a dramatic improvement in the accuracy of time of day on iPhones and iPads since the release of iOS 8.2? The accuracy used to be only plus or minus 2 or 3 seconds, now it is about 100 times better, usually a few tens of milliseconds. I figure Apple might have finally paid some attention to accurate time of day with 8.2, possibly because of the Apple Watch. It's a pleasing improvement, I hope it's permanent. -- Anthony ==================================== I noticed something similar on my iPad, but hadn't correlated the change with iOS 8.2. I thought it was just chance, but you could be right. On the other hand, I just checked now and it's 1.9 seconds off. I compare with Emerald Time, which includes my one stratum-1 NTP server in its sources. https://itunes.apple.com/gb/app/emerald-time/id290384375?mt=8 Cheers, David -- SatSignal Software - Quality software written to your requirements Web: http://www.satsignal.eu Email: david-taylor@blueyonder.co.uk
R
Raj
Wed, Apr 1, 2015 12:03 PM

Now that you mention it!

I use the app Emerald Sequoia watch that gives me time from 4 servers but it does not change the phone time.
Seconds to one decimal place.

Raj, vu2zap

At 01/04/2015, you wrote:

Has anyone else noticed a dramatic improvement in the accuracy of time
of day on iPhones and iPads since the release of iOS 8.2? The accuracy
used to be only plus or minus 2 or 3 seconds, now it is about 100
times better, usually a few tens of milliseconds. I figure Apple might
have finally paid some attention to accurate time of day with 8.2,
possibly because of the Apple Watch. It's a pleasing improvement, I
hope it's permanent.

--
Anthony

Now that you mention it! I use the app Emerald Sequoia watch that gives me time from 4 servers but it does not change the phone time. Seconds to one decimal place. Raj, vu2zap At 01/04/2015, you wrote: >Has anyone else noticed a dramatic improvement in the accuracy of time >of day on iPhones and iPads since the release of iOS 8.2? The accuracy >used to be only plus or minus 2 or 3 seconds, now it is about 100 >times better, usually a few tens of milliseconds. I figure Apple might >have finally paid some attention to accurate time of day with 8.2, >possibly because of the Apple Watch. It's a pleasing improvement, I >hope it's permanent. > >-- >Anthony
DM
Dave Martindale
Wed, Apr 1, 2015 1:18 PM

Hmm.  I'd say that the time setting accuracy may have improved, but the
timekeeping accuracy still isn't wonderful.

I just checked my iPhone, using the Emerald Time app to display the
difference between iOS time and NTP time.  The local time was 850 ms fast.
Then I went to Settings->General->Date and Time and turned "Set
Automatically" off, then back on again.  This seems to have forced the
phone to resync its local time, and now the error is zero (less than 1 ms).

I also have an iPad which had been configured not to set the time
automatically.  It was 2.5 seconds fast.  When I turned automatic setting
on, it reset its local time to within 2 ms of correct.

So the setting appears to be nice and accurate.  But the iPhone, which has
been in "automatic time set" mode ever since the iOS update 3 weeks ago,
was nearly one second off.  I don't know how often iOS does a time sync, so
I don't know how long it took to drift by that much.  This amount of error
does suggest that it only periodically resets itself to the correct time,
without trying to correct the local oscillator to provide continuously
accurate time (like using NTP would do).

  • Dave

On Tue, Mar 31, 2015 at 11:59 PM, Anthony G. Atkielski <
anthony@atkielski.com> wrote:

Has anyone else noticed a dramatic improvement in the accuracy of time
of day on iPhones and iPads since the release of iOS 8.2? The accuracy
used to be only plus or minus 2 or 3 seconds, now it is about 100
times better, usually a few tens of milliseconds. I figure Apple might
have finally paid some attention to accurate time of day with 8.2,
possibly because of the Apple Watch. It's a pleasing improvement, I
hope it's permanent.

--
Anthony


time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Hmm. I'd say that the time setting accuracy may have improved, but the timekeeping accuracy still isn't wonderful. I just checked my iPhone, using the Emerald Time app to display the difference between iOS time and NTP time. The local time was 850 ms fast. Then I went to Settings->General->Date and Time and turned "Set Automatically" off, then back on again. This seems to have forced the phone to resync its local time, and now the error is zero (less than 1 ms). I also have an iPad which had been configured not to set the time automatically. It was 2.5 seconds fast. When I turned automatic setting on, it reset its local time to within 2 ms of correct. So the setting appears to be nice and accurate. But the iPhone, which has been in "automatic time set" mode ever since the iOS update 3 weeks ago, was nearly one second off. I don't know how often iOS does a time sync, so I don't know how long it took to drift by that much. This amount of error does suggest that it only periodically resets itself to the correct time, without trying to correct the local oscillator to provide continuously accurate time (like using NTP would do). - Dave On Tue, Mar 31, 2015 at 11:59 PM, Anthony G. Atkielski < anthony@atkielski.com> wrote: > Has anyone else noticed a dramatic improvement in the accuracy of time > of day on iPhones and iPads since the release of iOS 8.2? The accuracy > used to be only plus or minus 2 or 3 seconds, now it is about 100 > times better, usually a few tens of milliseconds. I figure Apple might > have finally paid some attention to accurate time of day with 8.2, > possibly because of the Apple Watch. It's a pleasing improvement, I > hope it's permanent. > > -- > Anthony > > _______________________________________________ > time-nuts mailing list -- time-nuts@febo.com > To unsubscribe, go to > https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there. >
RO
Reid Oda
Wed, Apr 1, 2015 4:33 PM

Hi,

First email to the list here. I've been curious about this for a little
while: does the iPhone get sub-second timing info from GPS?

I did an experiment the other day where my friend and my iPhone clocks had
a fairly large offset, of perhaps 0.3 seconds. We were away from any
windows.

We went to a window, and watched as the Google maps location estimate got
to its most accurate setting. I assume the estimate accuracy is based on
the number of satellites the phone can see. Once we got this lock, the
clocks were then synchronized to within a few ms of each other. I estimate
8 ms, but our method (listening to ticks from a homemade app) was not ultra
accurate.

This seems to imply that the iPhone does get sub-second timing info from
GPS. Can anyone confirm/deny this?

On Wed, Apr 1, 2015 at 8:03 AM, Raj vu2zap@gmail.com wrote:

Now that you mention it!

I use the app Emerald Sequoia watch that gives me time from 4 servers but
it does not change the phone time.
Seconds to one decimal place.

Raj, vu2zap

At 01/04/2015, you wrote:

Has anyone else noticed a dramatic improvement in the accuracy of time
of day on iPhones and iPads since the release of iOS 8.2? The accuracy
used to be only plus or minus 2 or 3 seconds, now it is about 100
times better, usually a few tens of milliseconds. I figure Apple might
have finally paid some attention to accurate time of day with 8.2,
possibly because of the Apple Watch. It's a pleasing improvement, I
hope it's permanent.

--
Anthony


time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Hi, First email to the list here. I've been curious about this for a little while: does the iPhone get sub-second timing info from GPS? I did an experiment the other day where my friend and my iPhone clocks had a fairly large offset, of perhaps 0.3 seconds. We were away from any windows. We went to a window, and watched as the Google maps location estimate got to its most accurate setting. I assume the estimate accuracy is based on the number of satellites the phone can see. Once we got this lock, the clocks were then synchronized to within a few ms of each other. I estimate 8 ms, but our method (listening to ticks from a homemade app) was not ultra accurate. This seems to imply that the iPhone does get sub-second timing info from GPS. Can anyone confirm/deny this? On Wed, Apr 1, 2015 at 8:03 AM, Raj <vu2zap@gmail.com> wrote: > Now that you mention it! > > I use the app Emerald Sequoia watch that gives me time from 4 servers but > it does not change the phone time. > Seconds to one decimal place. > > Raj, vu2zap > > At 01/04/2015, you wrote: > >Has anyone else noticed a dramatic improvement in the accuracy of time > >of day on iPhones and iPads since the release of iOS 8.2? The accuracy > >used to be only plus or minus 2 or 3 seconds, now it is about 100 > >times better, usually a few tens of milliseconds. I figure Apple might > >have finally paid some attention to accurate time of day with 8.2, > >possibly because of the Apple Watch. It's a pleasing improvement, I > >hope it's permanent. > > > >-- > >Anthony > > _______________________________________________ > time-nuts mailing list -- time-nuts@febo.com > To unsubscribe, go to > https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there. >
CA
Chris Albertson
Wed, Apr 1, 2015 6:29 PM

On Wed, Apr 1, 2015 at 9:33 AM, Reid Oda reid.oda@gmail.com wrote:
... Once we got this lock, the

clocks were then synchronized to within a few ms of each other. I estimate
8 ms, but our method (listening to ticks from a homemade app) was not ultra
accurate.

Sound travels at about 1 foot per ms.  So you can effectively delay
the click on one phone my placing it 8 feet farther away then the
other  phone.  Adjust the distance until  the clicks seem to occur at
the same time.  That said, I doubt your ears are sensitive enough to
work at the single digit ms level.  Some musicians have learned to
keep in time and can hear a 20ms difference.  Most people can hear a
50 ms delay.    But as the clicks get close we hear them as one sound

If you have a computer with a microphone yo can record the clicks at
some high sample rate and then look at the spike on the audio
waveform.  This WILL sub-ms clicks.

--

Chris Albertson
Redondo Beach, California

On Wed, Apr 1, 2015 at 9:33 AM, Reid Oda <reid.oda@gmail.com> wrote: ... Once we got this lock, the > clocks were then synchronized to within a few ms of each other. I estimate > 8 ms, but our method (listening to ticks from a homemade app) was not ultra > accurate. Sound travels at about 1 foot per ms. So you can effectively delay the click on one phone my placing it 8 feet farther away then the other phone. Adjust the distance until the clicks seem to occur at the same time. That said, I doubt your ears are sensitive enough to work at the single digit ms level. Some musicians have learned to keep in time and can hear a 20ms difference. Most people can hear a 50 ms delay. But as the clicks get close we hear them as one sound If you have a computer with a microphone yo can record the clicks at some high sample rate and then look at the spike on the audio waveform. This WILL sub-ms clicks. -- Chris Albertson Redondo Beach, California
P
Paul
Wed, Apr 1, 2015 7:31 PM

On Wed, Apr 1, 2015 at 12:33 PM, Reid Oda reid.oda@gmail.com wrote:

This seems to imply that the iPhone does get sub-second timing info from
GPS. Can anyone confirm/deny this?

As of iOS N where I believe N == 5 it does the equivalent of calling
ntpdate every few hours if the network is available.  I assume it uses the
mobile system if the network is unavailable and you have a cellular radio
because the devices (can) have a lot of drift.

On Wed, Apr 1, 2015 at 12:33 PM, Reid Oda <reid.oda@gmail.com> wrote: > This seems to imply that the iPhone does get sub-second timing info from > GPS. Can anyone confirm/deny this? > As of iOS N where I believe N == 5 it does the equivalent of calling ntpdate every few hours if the network is available. I assume it uses the mobile system if the network is unavailable and you have a cellular radio because the devices (can) have a lot of drift.
RO
Reid Oda
Wed, Apr 1, 2015 8:29 PM

On Wed, Apr 1, 2015 at 2:29 PM, Chris Albertson albertson.chris@gmail.com
wrote:

d

Thanks for the idea of using distance to determine time offset. That's
pretty clever. :)

As for perceiving ms-order offsets of audio events (if you're interested):
it depends on the sound. The sharper the tick the easier it is to perceive
a timing offset. The tick I use is extremely sharp and I can (and I bet you
could too) be able to tell the difference between a 2 and 8 ms offset. 2 ms
sounds like a slightly phasey version of a single tick while at 8 ms you
can clearly hear 2 sounds.

My research is about audio synchronization, which is what got me interested
in clocks.

On Wed, Apr 1, 2015 at 2:29 PM, Chris Albertson <albertson.chris@gmail.com> wrote: > d Thanks for the idea of using distance to determine time offset. That's pretty clever. :) As for perceiving ms-order offsets of audio events (if you're interested): it depends on the sound. The sharper the tick the easier it is to perceive a timing offset. The tick I use is extremely sharp and I can (and I bet you could too) be able to tell the difference between a 2 and 8 ms offset. 2 ms sounds like a slightly phasey version of a single tick while at 8 ms you can clearly hear 2 sounds. My research is about audio synchronization, which is what got me interested in clocks.
O
Oz-in-DFW
Wed, Apr 1, 2015 8:41 PM

On 4/1/2015 2:31 PM, Paul wrote:

On Wed, Apr 1, 2015 at 12:33 PM, Reid Oda reid.oda@gmail.com wrote:

This seems to imply that the iPhone does get sub-second timing info from
GPS. Can anyone confirm/deny this?

As of iOS N where I believe N == 5 it does the equivalent of calling
ntpdate every few hours if the network is available.  I assume it uses the
mobile system if the network is unavailable and you have a cellular radio
because the devices (can) have a lot of drift.


time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

The phone has to keep synch within a few microseconds of the network.
There was a time when operators were very sloppy about clock time and
really only worried about network frequency, but most operators are now
maintaining 50 ns or so at the base stations and have to maintain within
5 microseconds to meet the LTE specs.  How much of this is preserved
through to the user interface is anyone's guess.

--
mailto:oz@ozindfw.net
Oz
POB 93167
Southlake, TX 76092 (Near DFW Airport)

On 4/1/2015 2:31 PM, Paul wrote: > On Wed, Apr 1, 2015 at 12:33 PM, Reid Oda <reid.oda@gmail.com> wrote: > >> This seems to imply that the iPhone does get sub-second timing info from >> GPS. Can anyone confirm/deny this? >> > As of iOS N where I believe N == 5 it does the equivalent of calling > ntpdate every few hours if the network is available. I assume it uses the > mobile system if the network is unavailable and you have a cellular radio > because the devices (can) have a lot of drift. > _______________________________________________ > time-nuts mailing list -- time-nuts@febo.com > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there. The phone has to keep synch within a few microseconds of the network. There was a time when operators were very sloppy about clock time and really only worried about network frequency, but most operators are now maintaining 50 ns or so at the base stations and have to maintain within 5 microseconds to meet the LTE specs. How much of this is preserved through to the user interface is anyone's guess. -- mailto:oz@ozindfw.net Oz POB 93167 Southlake, TX 76092 (Near DFW Airport)
DG
David Glover-Aoki
Wed, Apr 1, 2015 9:10 PM

On Apr 1, 2015, at 1:29 PM, Reid Oda reid.oda@gmail.com wrote:

As for perceiving ms-order offsets of audio events (if you're interested):
it depends on the sound. The sharper the tick the easier it is to perceive
a timing offset.

I use an iPhone app called "Atomic Clock". It syncs with an NTP server of your choice, but it also has a particularly sharp "tick" on the seconds, and a different-toned tick at the top of the minute.

The clock also has an outer circle that can be configured to complete one rotation every second, which is great for comparing between clocks.

--
David Glover-Aoki | https://david.gloveraoki.net/contact
PGP key 5518C7DE  | Amateur Radio KJ6TLX

On Apr 1, 2015, at 1:29 PM, Reid Oda <reid.oda@gmail.com> wrote: > As for perceiving ms-order offsets of audio events (if you're interested): > it depends on the sound. The sharper the tick the easier it is to perceive > a timing offset. I use an iPhone app called "Atomic Clock". It syncs with an NTP server of your choice, but it also has a particularly sharp "tick" on the seconds, and a different-toned tick at the top of the minute. The clock also has an outer circle that can be configured to complete one rotation every second, which is great for comparing between clocks. -- David Glover-Aoki | https://david.gloveraoki.net/contact PGP key 5518C7DE | Amateur Radio KJ6TLX