Accuracy of Radio Controlled Clocks

[watson1]

I have a couple of radio controlled clocks (RCC). They keep in step with each other and I use them to set watches and clocks to the correct time.

I have just ordered a cheap chinese timegrapher and when it arrives I want to check and if necessary adjust it's timebase.

 

Q1.)

how accurate is the rate of a radio controlled clock?

Quote

Most RCCs synchronize themselves with a time broadcast signal once a day, at night, although some check themselves every few hours. Generally, that gives them an accuracy of better than plus or minus a half second (±0.5s) a day. Another advantage is that they automatically correct themselves for daylight-saving time, leap years, months with different numbers of days, and so on.

Q2,)

is there a way to identify if my RCC are adjusting their time periodically?

Q3.)

can I rely on the rate of the 1second tick being accurate?

[oldhippy]

Type this into google and download the pdf file.

how accurate is the rate of a radio controlled clock

[watson1]

 

Quote

 However, a much larger time error is
likely to accumulate during the interval between
synchronizations, as will be seen in the answer to question 4.

and fig6 shows that the guessimated "rate" is not anywhere near zero.

I conclude that a synchronised RCC cannot be used as an accurate timebase for a timegrapher.

 

whereas this seems to say that a disciplined RCC corrects it's rate by trimming it's own oscillator.

Quote

One of the first radio clocks was offered by Heathkit in late 1983. Their model GC-1000 "Most Accurate Clock" received shortwave time signals from radio station WWV in Fort Collins, Colorado. It automatically switched between WWV's 5, 10, and 15 MHz frequencies to find the strongest signal as conditions changed through the day and year. It kept time during periods of poor reception with a quartz-crystal oscillator. This oscillator was disciplined, meaning that the microprocessor-based clock used the highly accurate time signal received from WWV to trim the crystal oscillator. The timekeeping between updates was thus considerably more accurate than the crystal alone could have achieved. Time down to the tenth of a second was shown on an LED display. The GC-1000 originally sold for US$250 in kit form and US$400 preassembled, and was considered impressive at the time. Heath Company was granted a patent for its design.^[19][20]

In the 2000s (decade) radio-based "atomic clocks" became common in retail stores; as of 2010 prices start at around US$15 in many countries.^[21] Clocks may have other features such as indoor thermometers and weather station functionality. These use signals transmitted by the appropriate transmitter for the country in which they are to be used. Depending upon signal strength they may require placement in a location with a relatively unobstructed path to the transmitter and need fair to good atmospheric conditions to successfully update the time. Inexpensive clocks keep track of the time between updates, or in their absence, with a non-disciplined quartz-crystal clock, with the accuracy typical of non-radio-controlled quartz timepieces. Some clocks include indicators to alert users to possible inaccuracy when synchronization has not been recently successful.

 

 

How do I tell them apart?

Which takes me back to my Q2 & 3

Quote

Q2,)

is there a way to identify if my RCC are adjusting their time periodically?

Q3.)

can I rely on the rate of the 1second tick being accurate?

 

[jdm]

9 hours ago, watson1 said:

I have just ordered a cheap chinese timegrapher and when it arrives I want to check and if necessary adjust it's timebase.

Can't do that. For one, the calibrating procedure of these machines is not user accessible, or even documented. And then one can't use just another clock or watch as reference, no matter how accurate it is.

Be reassured that the timegrapher will be accurate enough for the practical job it's intended (which is not only about precise regulation), especially for non professional use.

[vinn3]

good post !  I remember  the "time tick" on the radio dial.  Greenwich meant time?  vin

[watson1]

7 hours ago, jdm said:

Can't do that. For one, the calibrating procedure of these machines is not user accessible, or even documented. And then one can't use just another clock or watch as reference, no matter how accurate it is.

Be reassured that the timegrapher will be accurate enough for the practical job it's intended for, epscially for non professional use.

I don't agree.

These simpler timegrapher claim to report rate errors down to 1second/day. That is 1 part in 86400 or approximately 11.6ppm

Any good measuring instrument needs to have an accuracy/tolerance at least half the lowest measured resolution and preferably approaching one tenth of the measurement resolution.

That puts the timegrapher's required accuracy into the 5ppm to 1ppm.

If one wants to check that, then one requires an instrument that gives better than 2ppm and preferably 100ppb

Any of the time standards are thousands of times better than that.

Now back to the adjustment, if the (cheap chinese) timegrapher is found to be out of tolerance.

I read/watched one of our watch builders adjust his timegrapher ( a real Witschi version), but he used a PC for his time standard. That drew a few adverse comments from others that said "no way a PC can give the accuracy required".

Some even suggested using their mobile phones as their preferred time standard.

youtube.com/watch?v=qgzzf6y5TSk

 

 

[watson1]

3 hours ago, vinn3 said:

good post !  I remember  the "time tick" on the radio dial.  Greenwich meant time?  vin

I must be slow.

It took me a re-read of you post to realise "vin" was not a comment but a signature !

[jdm]

14 minutes ago, watson1 said:

I don't agree.

These simpler timegrapher claim to report rate errors down to 1second/day. That is 1 part in 86400 or approximately 11.6ppm

I see that you're not familiar with using one. A min 1 sec/d is just the instantaneous display figure during the latest measurement interval. But if you change (from menu) the measurement interval to be longer, and then observe the line drawn, you will see if it is absolutely flat (unlikely), or gaining / slowing slightly.

Anyway, measuring timekeeping to a resolution better of 1 sec/day would be pointless, as even the best mechanical watches realistically aim for +/- 2 or 3 secs / day.

I suggest that on your learning path you focus on other subjects rather than an absolutely exact timegrapher calibration.

[watson1]

My next project is a pendulum clock. My challenge is to better 1second/day and see how far I can go with adjustments.

If the timing accuracy of the timegrapher is out by 1 or 2seconds/day, then that makes a nonsense of trying to adjust for better than that.

Edited February 7, 2018 by watson1

[jdm]

16 minutes ago, watson1 said:

My next project is a pendulum clock. My challenge is to better 1second/day and see how far I can go with adjustments.

If the timing accuracy of the timegrapher is out by 1 or 2seconds/day, then that makes a nonsense of trying to adjust for better than that.

I have explained above how to can observe accuracy better than 1 sec a day, here again:

But if you change (from menu) the measurement interval to be longer, and then observe the line drawn, you will see if it is absolutely flat (unlikely), or gaining / slowing slightly.

That being said, bench timegraphers are made to measure wrist or pocket watches with a Swiss lever. I don't know if it would work at all on a clock.

 

 

 

[vinn3]

1 hour ago, watson1 said:

I don't agree.

These simpler timegrapher claim to report rate errors down to 1second/day. That is 1 part in 86400 or approximately 11.6ppm

Any good measuring instrument needs to have an accuracy/tolerance at least half the lowest measured resolution and preferably approaching one tenth of the measurement resolution.

That puts the timegrapher's required accuracy into the 5ppm to 1ppm.

If one wants to check that, then one requires an instrument that gives better than 2ppm and preferably 100ppb

Any of the time standards are thousands of times better than that.

Now back to the adjustment, if the (cheap chinese) timegrapher is found to be out of tolerance.

I read/watched one of our watch builders adjust his timegrapher ( a real Witschi version), but he used a PC for his time standard. That drew a few adverse comments from others that said "no way a PC can give the accuracy required".

Some even suggested using their mobile phones as their preferred time standard.

youtube.com/watch?v=qgzzf6y5TSk

 

 

welcome to the forum.   why do some diferent cell phones have have a few minute time diferent times?  vin

[jdm]

14 hours ago, vinn3 said:

why do some diferent cell phones have have a few minute time diferent times?  vin

Because the owner hasn't set it to get time from mobile network, or from the Internet. 

[oldhippy]

4 hours ago, watson1 said:

My next project is a pendulum clock. My challenge is to better 1second/day and see how far I can go with adjustments.

Ha Ha Ha.

 

[Ishima]

I mean this as kindly as possible, but I think your expectations are what need calibrating, not your timegrapher. 
A timegrapher result is only an indication anyway, it's not uncommon to find a mechanical watch, worn will keep noticeably (if only noticeable by a pedant whos measured the time loss/gain against a more accurate device) different time than a timegrapher suggests it should.

I don't know anything about clocks, but I do know that as a mechanical timekeeping device, their timekeeping will never be perfect either, and a change in the temperature of the room (which presumably will always be changing during the day and season to season) the clock is in will affect its accuracy.

There's a never-ending list of things you can learn in horology, perfect regulation isn't one of them, so my advice is to not get held up on it.

 

[oldhippy]

I don't know anything about clocks, but I do know that as a mechanical timekeeping device, their timekeeping will never be perfect either, and a change in the temperature of the room (which presumably will always be changing during the day and season to season) the clock is in will affect its accuracy.

 

You can have a clock with a compensated pendulum. You also need a movement with maintaining power, unless you have 1-year duration which they are quite rear. There are too many factors to go into as to why this is impossible in a manual clock. It has been proven that a quartz movement has more advantages in keeping correct time.  

[jdm]

We agree that aiming for perfect accuracy on a mechanical timepiece seems a bit futile, especially in an hobbyist's context. On the other hand such devices have existed since about 300 years (yuck!), in the form of Marine Chronometers. Watchmaking has made extraordinary progress is the last 15 years, so I think it can only get better. I for one would be very pleased to have a mechanical clock at home that is beautiful and accurate.

[oldhippy]

Marine Chronometers as you know have a balance. Our friend mentioned a pendulum clock.

[Marc]

For those who think this is a pipe dream have a read.

http://wornandwound.com/the-240-year-old-pendulum-clock-thats-more-accurate-than-your-watch/

@watson1 I admire you ambitions and wish you the very best of luck. I hope that you might document your progress here.

[vinn3]

Quote

1 hour ago, Marc said:

For those who think this is a pipe dream have a read.

http://wornandwound.com/the-240-year-old-pendulum-clock-thats-more-accurate-than-your-watch/

@watson1 I admire you ambitions and wish you the very best of luck. I hope that you might document your progress here.

I recently bought a used modern pendulum clock.  it has a knob on the weight to adjust the accuracy.   it seems to maintain + -   5 min..  which is fine with me.   vin

 

[oldhippy]

This clock has only been tested at 100 days. Our friend is talking about 1 year.

I suppose it still could be possible. Until its achieved we will have to wait.

[Marc]

27 minutes ago, oldhippy said:

This clock has only been tested at 100 days.

After only 100 days this clock lost a mere 3/8 of a second which translates to only 1.37 seconds in a year, far exceeding the OP's target of bettering 1 second per day.

 

On 07/02/2018 at 6:15 PM, watson1 said:

My challenge is to better 1second/day and see how far I can go with adjustments.

 

[jdm]

Today i was fiddling with my 1900 timegrapher and found the screen below, with the picture following  it shows the 99 setting enabled..

Which confirms what I have been saying regarding the ability of regulating to better than 1 sec/day without having to mess with the instrument in any way.
And for the model 1000 which doesn't have the 999 / 99 setting, the same can be achieved by just lengthening the measurement interval.

All that just for completeness, as I still think that these timegraphers can't be used for clock regulation, and that is extremely unlikely that anyone, especially a beginner, can get extraordinary precision from a common mechanical clock.

 

[oldhippy]

A timegrapher can be used for certain types of clocks the ones that are fitted with a lever escapement. You should be able to get it within a few seconds a day. I could with the older type of timing machine, it came with a lead and clip that you attached to the movement. 

[measuretwice]

One thing I noticed about Weishi's claimed accuracy is the temperature range.  afaik accurate crystal performance needs a much narrow range than -10 - + 50C.  Its the why behind a crystal watch keeping different time on and off the wrist, right?  Perhaps I'm missing something on it.

I attended a lecture recently by an amateur Foucault pendulum maker.  The accuracy levels these guys achieve is incredible as is the thought and technology put into the pendulum.  It transcends the practical, but for anyone with an interest in science and engineering its fascinating stuff.  For example, a source of error is unequal stress in the wire itself that affect how it bends during an oscillation and they go to lengths to compensate for that and myriad of other things.  He conducts experiments when he goes away for a month on holidays, as, despite being mounted on the basement slab, walking around the house will mess it up.  Over the period you can see on the chart when a truck goes down the road.  Anyway, I'm going a bit OT, but those guys are really into super accurate pendulums.

As for measuring, having a pendulum a break a light or IR beam is easy to set up.  You can pick up a HP 5335a for a few hundred dollars and if it has option three should be good to 1ppm (a constant temp oven for the crystal).  Heck for that matter, you can get a 1ppb oscillator for $20 which could be the base for an accurate counter if you have some electronics ability.

Computers do a poor job of measuring time as their is no need for their crystal to be tuned to any super accurate level.  Your super fast computer @ 3GHz, might really be 3.001 GHz for example, doesn't matter for computing but it sure does for timing.  Furthermore temperature changes will affect the clock speed over a session.  However we now have these GPS dongles - coming from the need for accurate synchronization of international networks and securities trading.   I believe they have extremely accurate output however haven't used one so can't quote chapter and verse....but if does make make it accurate enough, its appealing given the computers ability to capture and manipulation data.

There is nothing in my life than depends on 1/s/d accuracy, but I also get the fun of pursuing a challenge so please post your progress

Edited February 11, 2018 by measuretwice

[jdm]

1 hour ago, measuretwice said:

One thing I noticed about Weishi's claimed accuracy is the temperature range.  afaik accurate crystal performance needs a much narrow range than -10 - + 50C.  Its the why behind a crystal watch keeping different time on and off the wrist, right?  Perhaps I'm missing something on it.

Note they mention the Crystal Oscillator to be Temperature Compensated. That is the same technology used on HA (High Accuracy) quartz watches.