New Year's partygoers wearing Apple Watch will have 'most accurate watch in the room'

AppleInsider · December 30, 2015 6:42PM

Apple's attention to detail encompasses the accuracy of the Apple Watch as a timepiece, thanks to a complex proprietary system of servers, satellites and more, making it one of the most accurate watches in the world.

The accuracy of the Apple Watch time was highlighted by Apple vice president Kevin Lynch in an interview with Mashable published Wednesday. In the interview, Lynch provided a number of details on how Apple ensures accuracy of its wearable lineup, including:

Apple has 15 Network Time Servers around the world
Each building connects to GPS satellites that gather time info from the U.S. Naval Observatory
A user's iPhone connects to the servers, which then sends time data to the watch
Unique hardware in the Apple Watch makes it four times more accurate at telling time than the iPhone
The Apple Watch also compensates for delays in sending data
In testing, Apple uses high-speed cameras that look for latency in a watch face's second hand

"With New Year's coming, those who have the Apple Watch will be the most accurate watch in the room," Lynch told Mashable.

Some existing traditional timepieces on the market also sync with world atomic clocks, accomplishing this through methods like GPS or other wireless technologies. By using the iPhone as a conduit, the Apple Watch can potentially sync the current time more frequently, allowing for consistency and accuracy at levels not seen in most traditional watches.

dickprinter · December 30, 2015 6:50PM

Aren't my friends and family going to be envious of me when I'm the one screaming "HAPPY NEW YEAR!!!" at the exact, appropriate moment compared to everyone else in the room.

dachar · December 30, 2015 7:11PM

I have been wondering about Apple Watch time. I have a "radio controlled" clock that uses an atomic time signal sent via radio waves to keep it correct, I use the "pips" on BBC Radio 4 that mark the hour and now I have my Apple Watch. None of them have the hour at exactly the same time. I have not been able to properly work out the differences accurately yet but I think its about 5-10 seconds between first and last. Does anyone know why radio waves for the BBC Radio and the atomic controlled clock appear to have a different time to the internet+WiFi used by the Apple Watch?

patsu · December 30, 2015 7:26PM

dachar said:

I have been wondering about Apple Watch time. I have a "radio controlled" clock that uses an atomic time signal sent via radio waves to keep it correct, I use the "pips" on BBC Radio 4 that mark the hour and now I have my Apple Watch. None of them have the hour at exactly the same time. I have not been able to properly work out the differences accurately yet but I think its about 5-10 seconds between first and last. Does anyone know why radio waves for the BBC Radio and the atomic controlled clock appear to have a different time to the internet+WiFi used by the Apple Watch?

You will need to look at how both are implemented.

It is insufficient to say 'my clock syncs to an atomic clock, therefore it must be equally accurate'.

How often does the sync take place ? What happens in between syncs ? What is the latency during syncing ?

Even though smartphones may sync to the same NTP servers, they may not sync as often to preserve battery life. The Watch's primary purpose is time keeping, naturally Apple will spend more run-time resources to finesse it.

For your radio wave approach, you'll have to factor in the network + device latency, and the sync interval. Syncing to an atomic clock just means you get a more accurate source (that does not drift) during the query. It does not mean you'll be as accurate as that atomic clock unless that device is engineered to minimize all the latencies, and has an accurate clock itself.

edited December 2015

thewhitefalcon · December 30, 2015 8:10PM

My Casio atomic is rated to +-30ms per month without a sync. So either the Apple Watch is pretty darn accurate or he's blowing smoke.

justin e · December 30, 2015 8:14PM

Wouldnt the iphone have the same accurate time? or even more so?

cpsro · December 30, 2015 8:26PM

justin e said:

Wouldnt the iphone have the same accurate time? or even more so?

Close. The iPhone would have all but the last two features. But the iPhone connects more directly to the Internet--maybe one less hop than for the Watch--so compensation for communications delays on the iPhone should yield better initial accuracy than the Watch. Over time (through the course of 24 hours?), if the iPhone/Watch doesn't re-synchronize with time servers, the Watch apparently has a 4x more accurate oscillator and will become more accurate. If the Watch synchronizes directly over wifi without going through the iPhone, it still may be less accurate initially because its processor is slower than the iPhone 5s/6/6s.

I wonder about the Nav Clock iPhone app, though, which communicates "directly" with GPS and might be even more accurate than using network time servers.

edited December 2015

patsu · December 30, 2015 8:37PM

justin e said:

Wouldnt the iphone have the same accurate time? or even more so?

Not necessarily. According to the interview, the Watch has a crystal temperature controlled oscillator.

In addition, iPhone and Macs don't have to keep precise real world time. They have other clocking mechanisms for internal use (e.g., host clock).

The iPhone helps the Watch relay NTP requests but they do not need to apply the sync on themselves every time. They just don't have that stringent time piece requirement.

edited December 2015

patsu · December 30, 2015 8:53PM

sog35 said:

So this is what Tim Cook's Apple has become......

A watch that tells accurate time. You gotta be kidding me. Steve Jobs would never allow the time accuracy of a product be one of its pillar highlights.

Its getting more apparent that Tim Cook really has no idea how to run an elite company.

I see your concern and point. There are room for improvement for messaging Apple's vision.

But that vision may be orthogonal to what the short term investors hope for though.
In addition, if they are negative on Apple and have invested money to short it, they will just find other reasons to hate Apple's approach.

IMHO, it's more prudent to invest in the fundamentals. Build up their platforms. Focus even more on the customers.

Over time, the short term investors who are unhappy will go away. But the people who understand Apple's approach will stay.

However, this is not to say Apple shouldn't improve their messaging. It's something they have to work on regardless of the stock price.

Jobs is known to put pressure on the advertising folks all the time too.

edited December 2015

hodar · December 30, 2015 8:54PM

patsu said:

justin e said:

Wouldnt the iphone have the same accurate time? or even more so?

Not necessarily. According to the interview, the Watch has a crystal temperature controlled oscillator.

In addition, iPhone and Macs don't have to keep precise real world time. They have other clocking mechanisms for internal use (e.g., host clock).

The iPhone helps the Watch relay NTP requests but they do not need to apply the sync on themselves every time. They just don't have that stringent time piece requirement.

A temperature-COMPENSATED (not controlled) oscillator allows for fluctuations in the frequency based on fractional degrees temperature change. These are usually in the ±~50ppm range/°F. For the practical purposes, practically nothing, considering that nominal is taken at 72°F.

patsu · December 30, 2015 8:57PM

hodar said:

patsu said:

Not necessarily. According to the interview, the Watch has a crystal temperature controlled oscillator.

In addition, iPhone and Macs don't have to keep precise real world time. They have other clocking mechanisms for internal use (e.g., host clock).

The iPhone helps the Watch relay NTP requests but they do not need to apply the sync on themselves every time. They just don't have that stringent time piece requirement.

A temperature-COMPENSATED (not controlled) oscillator allows for fluctuations in the frequency based on fractional degrees temperature change. These are usually in the ±~50ppm range/°F. For the practical purposes, practically nothing, considering that nominal is taken at 72°F.

Yes, but that oscillator would also be more precise than the ones used on iPhone and Macs under regular conditions if the Watch is 4x as accurate.

wiggin · December 30, 2015 9:01PM

AppleInsider said:

Apple has 15 Network Time Servers around the world

Each building connects to GPS satellites that gather time info from the U.S. Naval Observatory

A user's iPhone connects to the servers, which then sends time data to the watch

Unique hardware in the Apple Watch makes it four times more accurate at telling time than the iPhone

The Apple Watch also compensates for delays in sending data

In testing, Apple uses high-speed cameras that look for latency in a watch face's second hand

It would seem bullets 1, 2, 3, and 5 are quite Rube-Goldberg-eque. The iPhone that the Apple watch is synced to has access to the same GPS timing signal that their NTSs have. Sure, you aren't going to have a GPS signal indoors, but just how often is the phone hitting the NTS anyway? And pretty much all of the delays that bullet 5 is compensating for is due to 1-3, which should be entirely unnecessary!

patsu · December 30, 2015 9:04PM

wiggin said:

AppleInsider said:

Apple has 15 Network Time Servers around the world

Each building connects to GPS satellites that gather time info from the U.S. Naval Observatory

A user's iPhone connects to the servers, which then sends time data to the watch

Unique hardware in the Apple Watch makes it four times more accurate at telling time than the iPhone

The Apple Watch also compensates for delays in sending data

In testing, Apple uses high-speed cameras that look for latency in a watch face's second hand
It would seem bullets 1, 2, 3, and 5 are quite Rube-Goldberg-eque. The iPhone that the Apple watch is synced to has access to the same GPS timing signal that their NTSs have. Sure, you aren't going to have a GPS signal indoors, but just how often is the phone hitting the NTS anyway? And pretty much all of the delays that bullet 5 is compensating for is due to 1-3, which should be entirely unnecessary!

It sounds like their margin is 50ms. So they will have to do it before the clocks drifts apart under all conditions.

GPS is slow and eats battery. That's why people complain about Maps slowness on the Watch.

Accurate NTP sync is important because if they compensate it wrong, the Watch can be off by > 50ms until the next good sync (Even if the oscillator is more precise than iPhone's). If they get that next sync wrong again, then it will continue to be wrong, so on and so forth.

Similarly if GPS sync takes too long, then they probably can't guarantee the 50ms error margin in all cases.

That high speed camera test probably attempts to detect rendering latency too. The Watch is a low power device, but has to keep rendering the clock regardless of the watch face and complications used.

edited December 2015

djsherly · December 30, 2015 9:57PM

Hmm - the watch needs to be accurate but short of running the space program it's not clear to me why this is a 'feature'. No one cares if your own time is a second or two out.

I would focus on the performance of the watch as a tool. Yes it tells time, but it's almost incidental as a feature. Any watch can do that well enough.

FWIW - I just picked up a second hand Apple watch and so far I'm OK with it. I was hoping to replace my myriad fitness wearables with it but so far it's been a little underwhelming. The app maturity is just not there yet for me. Other aspects I do like though. I might flip it and wait for v2 to come along.

patsu · December 30, 2015 10:09PM

djsherly said:

Hmm - the watch needs to be accurate but short of running the space program it's not clear to me why this is a 'feature'. No one cares if your own time is a second or two out.

I would focus on the performance of the watch as a tool. Yes it tells time, but it's almost incidental as a feature. Any watch can do that well enough.

FWIW - I just picked up a second hand Apple watch and so far I'm OK with it. I was hoping to replace my myriad fitness wearables with it but so far it's been a little underwhelming. The app maturity is just not there yet for me. Other aspects I do like though. I might flip it and wait for v2 to come along.

I thought so at first.

But now I think it implies that the Watch has a pretty good real-time execution mechanism (with 50ms margin) despite its lower power requirements; all the way from network servers to rendering data on screen in a timely fashion.

I expect its app ecosystem to focus more and more on process control and sensor measurements. Apple should come up with smart bands that integrate into this real-time framework.

spheric · December 30, 2015 11:42PM

dachar said:

I have been wondering about Apple Watch time. I have a "radio controlled" clock that uses an atomic time signal sent via radio waves to keep it correct, I use the "pips" on BBC Radio 4 that mark the hour and now I have my Apple Watch. None of them have the hour at exactly the same time. I have not been able to properly work out the differences accurately yet but I think its about 5-10 seconds between first and last. Does anyone know why radio waves for the BBC Radio and the atomic controlled clock appear to have a different time to the internet+WiFi used by the Apple Watch?

You're probably listening to the BBC via internet radio. That usually has a delay of easily five seconds IME (when my preferred terrestrial radio station in the car drops out of reception, I have enough time to hit the iPhone, launch the app, and start the stream via cellular network, and after all that it will pick up pretty much exactly where the radio dropped off). I presume this is due to transcoding, network speed, and buffering delays.

spheric · December 30, 2015 11:44PM

sog35 said:

So this is what Tim Cook's Apple has become......

A watch that tells accurate time. You gotta be kidding me. Steve Jobs would never allow the time accuracy of a product be one of its pillar highlights.

Its getting more apparent that Tim Cook really has no idea how to run an elite company.

You're seriously arguing that the single most important function of a TIMEPIECE being ridiculously accurate beyond even the most arcane usage scenario imaginable would have been something Jobs wouldn't have proudly and happily expounded upon at a keynote? Or are you actually being sarcastic?

spheric · December 30, 2015 11:49PM

djsherly said:

Hmm - the watch needs to be accurate but short of running the space program it's not clear to me why this is a 'feature'. No one cares if your own time is a second or two out.

This is actually crucial to the watch market. You can't sell a watch that "is a second or two out" for more than $50. Yes, this thing is a lot more, but it is first and foremost, not least in name, a watch.

dasanman69 · December 31, 2015 3:08AM

How about Apple Watch wearing non-partygoers?

wiggin · December 31, 2015 3:37AM

patsu said:

wiggin said:

It would seem bullets 1, 2, 3, and 5 are quite Rube-Goldberg-eque. The iPhone that the Apple watch is synced to has access to the same GPS timing signal that their NTSs have. Sure, you aren't going to have a GPS signal indoors, but just how often is the phone hitting the NTS anyway? And pretty much all of the delays that bullet 5 is compensating for is due to 1-3, which should be entirely unnecessary!

It sounds like their margin is 50ms. So they will have to do it before the clocks drifts apart under all conditions.

GPS is slow and eats battery. That's why people complain about Maps slowness on the Watch.

Accurate NTP sync is important because if they compensate it wrong, the Watch can be off by > 50ms until the next good sync (Even if the oscillator is more precise than iPhone's). If they get that next sync wrong again, then it will continue to be wrong, so on and so forth.

Similarly if GPS sync takes too long, then they probably can't guarantee the 50ms error margin in all cases.

That high speed camera test probably attempts to detect rendering latency too. The Watch is a low power device, but has to keep rendering the clock regardless of the watch face and complications used.

But the GPS doesn't have to be on all the time. And getting the timing signal is a much more trivial task than getting the full set of satellite ephemeris data and triangulating position (i.e., when using Maps). That is the part that causes the long "warm up" period for GPS (especially non-assisted GPS devices) and eats up battery. Just having Location Services turned on, but not actively updating satellite positions and calculating your location has little impact on your battery. Just how often does the Apple watch need to sync its time? It supposedly has a much more accurate crystal oscillator, so it shouldn't really need to sync up that often. If it only updated once a day (or even 10 times a day), the slowness and battery drain you are worried about would be complete non-issues.

Granted, if the user didn't have Location Services turned on they would need the NTS as a backup plan. But to do that by default and then try to back out the delay caused by the round trip request traveling over a completely unpredictable route over the internet through untold numbers of servers and routers seems a tad silly.

patsu · December 31, 2015 3:51AM

wiggin said:

patsu said:

It sounds like their margin is 50ms. So they will have to do it before the clocks drifts apart under all conditions.

GPS is slow and eats battery. That's why people complain about Maps slowness on the Watch.

Accurate NTP sync is important because if they compensate it wrong, the Watch can be off by > 50ms until the next good sync (Even if the oscillator is more precise than iPhone's). If they get that next sync wrong again, then it will continue to be wrong, so on and so forth.

Similarly if GPS sync takes too long, then they probably can't guarantee the 50ms error margin in all cases.

That high speed camera test probably attempts to detect rendering latency too. The Watch is a low power device, but has to keep rendering the clock regardless of the watch face and complications used.

But the GPS doesn't have to be on all the time. And getting the timing signal is a much more trivial task than getting the full set of satellite ephemeris data and triangulating position (i.e., when using Maps). That is the part that causes the long "warm up" period for GPS (especially non-assisted GPS devices) and eats up battery. Just having Location Services turned on, but not actively updating satellite positions and calculating your location has little impact on your battery. Just how often does the Apple watch need to sync its time? It supposedly has a much more accurate crystal oscillator, so it shouldn't really need to sync up that often. If it only updated once a day (or even 10 times a day), the slowness and battery drain you are worried about would be complete non-issues.

Granted, if the user didn't have Location Services turned on they would need the NTS as a backup plan. But to do that by default and then try to back out the delay caused by the round trip request traveling over a completely unpredictable route over the internet through untold numbers of servers and routers seems a tad silly.

As I understand, GPS takes a long time to warm up if it's not active yet.
If it's already active, then its response is faster.

Apple has been working on this for a while now. If a direct GPS call can solve the issue with little impact, I doubt Apple will go through the trouble of doing complex NTP sync. And yes GPS is not available if location service is off, or when the user is indoor.

edited December 2015

New Year's partygoers wearing Apple Watch will have 'most accurate watch in the room'

Comments