• Upcoming updates
    Over the next couple of weeks we will be performing software updates on the forum. These will be completed in small steps as we upgrade individual software addons. You might occasionally see a maintenance message that will last a few minutes at most.

    If we anticipate an update will take more than a few minutes, we'll put up a notice with estimated time.

    Thank you!

High/very high accuracy quartz clocks.

timw

Registered User
Jan 21, 2023
9
6
3
74
Country
Most quartz clock movements operate at 32768 HZ (32.738 KHZ) - and there are 86,400 seconds in a day. Easy to see, then, that even a good quartz movement, off by one part in 32,000, is off by two seconds per day. In fact most are good for 15-30 seconds per month (spm), which, while most of us take that for granted, is still pretty amazing. But not good enough for navigation.

For maybe forty years, until GPS positioning became common, the marine chronometer was either - or both - an electric clock or a high-accuracy battery-powered quartz clock modified to operate at 4.19 MHZ, more than 100xs faster, easily accurate to better than one minute/year. Such clocks, made by Seiko and European companies, were expensive. I think at the moment only the Chinese are producing such clocks, the Yantai Daxin CZ-05, and the Chijiu Clock Co. CJCD-2T, either for a little more than USD$200, maybe $300 delivered to the US.

But these days there are cheap oscillators at up to 25 MHZ and higher. A quartz clock operating at that rate would be truly a set-and-forget clock, accurate to a few seconds per year. And it would seem it should be neither too difficult to build one nor too expensive. And that there might even be kits! But if there are, I cannot find them.

The approach the chronometer makers appear to have used was to take a 32768 HZ common (not expensive) quartz clock movement and wire it to the faster oscillator, with other pieces to count and tell the movement when to "tick." The trick for someone like me is find out how to do that.

Cost should be not-much for an old clock, whether or not originally quartz, eBay maybe $25. A good Seiko or Takane movement, $20. The oscillator, $25-50. Other electrical parts, for a guess, $20. Total $100, give or take. For a clock that would need a battery once or twice per year, an occasional time check, and that would just plain tell the time.

Anyone interested in collaborating in making such a clock? For my part I would need simple and explicit direction on how to build/wire the electrics.
 

Robert Gift

Registered User
Nov 12, 2012
569
74
28
Denver, Colorado
Country
Region
As a teenager manyears ago I would listen to WWV at 5.0 MHz. I considered making a crystal-controlled clock and occasionally comparing how much it drifted from NBS WWV.
Not wortheffort.
But I precisely set a synchronous motor clock'second hand to WWV. Was amazed that after a week the second hand will still right on!
 

timw

Registered User
Jan 21, 2023
9
6
3
74
Country
As a teenager manyears ago I would listen to WWV at 5.0 MHz. I considered making a crystal-controlled clock and occasionally comparing how much it drifted from NBS WWV.
Not wortheffort.
But I precisely set a synchronous motor clock'second hand to WWV. Was amazed that after a week the second hand will still right on!
I'm guessing that has to do with the consistently precise flow of electricity in the USA. I have read that in Europe it is not that way, and so keeping electric clocks accurate is not-so-easy. Can anyone else add to this? For example, I have a cheap 30+-year-old GE electric bedside radio alarm clock. It only displays minutes and not seconds, but I have never had to adjust it for either gaining or losing time.
 

Roy Gardner

Ticker Stopper
NAWCC Member
Jun 14, 2016
79
56
18
Cameron Park, California
Country
Region
LeapSecond Home Page is an interesting website about extreme amateur timekeeping, but it's weighted toward atomic clocks. The author of the website gave a presentation at an NAWCC symposium in 2013. His site has a page on the stability of power mains frequency.

I work at a place where I help build tools to test prototype PC motherboards and I once tried to figure out from the documents how I could adjust the onboard Real Time Clock to be more accurate. It seems the clock is inexpensive and its rate isn't adjustable because most PCs are on the internet and their clock automatically jumps to match an online time server several times per week.

I used to work at a place that designed a clock accurate enough to be the NTSC television genlock time source for TV studios. The quartz oscillator had to be in a little temperature-controlled oven to help keep its rate constant. Sort of like the COSC requirement for quartz chronometers to be temperature compensated.

Somewhere on the internet, I bet there's a schematic and bill of materials for the type of clock timw would like to build. Perhaps a multi-megahertz crystal divided down to 32,768 Hz to drive a standard quartz clock, sort of like the Bathys company using a chip scale atomic clock module to drive a standard quartz movement to make a true atomic wristwatch, their model Cesium 133.
 
Last edited:

Robert Gift

Registered User
Nov 12, 2012
569
74
28
Denver, Colorado
Country
Region
I'm guessing that has to do with the consistently precise flow of electricity in the USA. I have read that in Europe it is not that way, and so keeping electric clocks accurate is not-so-easy. Can anyone else add to this? For example, I have a cheap 30+-year-old GE electric bedside radio alarm clock. It only displays minutes and not seconds, but I have never had to adjust it for either gaining or losing time.
Yes. Amazing at howell our electricompanys keep 60Hz AC precisely 60 Hz. Buthrough our huge power grid they are allocked in with one another.
Somewhere I found a tuned vibrating-reed frequency indicator. Its 11 reeds range from 55 to 65 , I recall. Have only seen the 60 Hz reed vibrating.
On this desk is an old synchronous motor clock keeping perfectime running backward. Now at 5:36 it is reading 7:24
 
  • Like
Reactions: Roy Gardner

wisty

Registered User
Dec 24, 2014
130
76
28
Country
Why? Radio synchronized quartz movements that keep time to better than 1 second per decade are <$20 on eBay. If you are planning to be out of range of one of the many time signals, then a GPS module that can can be connected to an Arduino or other suitable device is <$10. It will tell you where you are as well as the time.
 

SteveC1964

Registered User
Dec 11, 2009
50
6
8
76
Houston, Texas
Country
Region
What about the solar powered Patek Philippe quartz clocks?

When Patek got into Quartz timepieces, they built their own crystals. I have been told they discarded 99%, keeping only the best 1%.

Decades back I saw a Patek exhibit. They displayed the Graves watch, and the Caliber 89. They also exhibited some solar powered clocks. When they tested the clocks, they fully charged them. They put them in cave. The cave was sealed. Six months later, they removed the clocks. None had lost a second. Fascinating clocks but $4-6,000 was a bit out of my price range.
 
  • Like
Reactions: Incroyable

timw

Registered User
Jan 21, 2023
9
6
3
74
Country
Why indeed to any of it? We need only look at our phones, or time.gov. I myself just like the idea of a clock on my shelf that tells time - all by itself with the help of a battery. I know how to build an ordinary quartz clock - a child can do that - but I do not know how to build the more accurate version, nor can find anything like instructions.
 

wisty

Registered User
Dec 24, 2014
130
76
28
Country
To get the stand alone ( no radio or GPS help) quartz accuracy you are looking for the only way I have seen that is sensible is to use a small microprocessor.

The quartz crystal is used as an input to the microprocessor. The microprocessor counts the input pulses and delivers the output needed. It can be a one second pulse, drive a stepper motor for hands or even drive a digital display directly.
If you measure the crystal accurately against a known time standard, you can adjust the counting within the microprocessor to compensate for the difference between the nominal and actual frequency of the crystal.

One stage further is to get the microprocessor to measure the temperature of the crystal, and add compensation in software - but again you would need to measure the temperature behaviour of the crystal and program in the compensation.

I have three quartz timekeeping devices like this. Two don't compensate for temperature, but have had the crystals calibrated - both are in governors for my Eureka clocks; one from Bryan Mumford one from Frank Roesky. The third is a Longines VHP wristwatch which has a second crystal that acts a the temperature sensor.

Making something similar using an Arduino would not be difficult, but I doubt you will find step by step instructions for doing so. To keep power demand down you might want to use a smaller processor - but the difficulty factor of building and programming that goes up.

On the other hand I also have a couple of commercial stand alone battery driven, radio synchronised clocks that do just what you want.
 
  • Like
Reactions: Bernhard J.

Schatznut

NAWCC Member
Sep 26, 2020
1,417
735
113
SoCal
Country
Region
The tempco and aging of the crystal are the predominating factors, not the period of oscillation per se. Tempco is measured in PPM/C and is the easiest to control. Aging can occur due to many different factors in manufacturing and use, but typically it is relatively minor, especially if the temperature is held constant and not excessively high. So an oscillator trimmed at a particular temperature is going to be as accurate as that calibration at that temperature. Temperature fluctuates, so will the oscillator. Easiest bet is to build an oven around the oscillator circuit and regulate the oven's temperature at a few degrees above the maximum ambient temperature. It's easy to build one that will hold a temperature of 0.1C or better. Then trim the oscillator to its fundamental frequency when the oven is at the setpoint and equilibrium has been achieved. Remember that the outcome will be heavily dependent upon both the accuracy and the precision of the frequency counter used to trim it.

The reason the power grid runs at 60Hz with high accuracy is not intuitive. Watch Robert Sapolsky's lecture on emergence on YouTube and it will become clear.
 
  • Like
Reactions: Roy Gardner

timw

Registered User
Jan 21, 2023
9
6
3
74
Country
Hi. I read the last two posts as saying that building a high-accuracy clock may not be as easy as I'd like to think. If so I need to know that and I appreciate the information. Here at eBay is the sort of thing gets me salivating a little.
And other similar offerings.
 
  • Like
Reactions: Roy Gardner

Schatznut

NAWCC Member
Sep 26, 2020
1,417
735
113
SoCal
Country
Region
This is a Temperature-Controlled Crystal Oscillator (TCXO) which has very good thermal stability - +/- 1PPM over the range -10 to +60C. Does it say what its initial accuracy is? You need both accuracy (nominal frequency) and thermal stability.

What I was suggesting above is that if you're a hard-core do-it-yourselfer, you could build your own TCXO. But if you can buy one for a few bucks, why go to the effort?
 

timw

Registered User
Jan 21, 2023
9
6
3
74
Country
"you could build your own TCXO. But if you can buy one for a few bucks, why go to the effort?"

Exactly so! I cannot help asking: when you see that eBay listing do you have a strong temptation to reach for your wallet?
 

Schatznut

NAWCC Member
Sep 26, 2020
1,417
735
113
SoCal
Country
Region
Two different approaches to the same end goal - a highly accurate time base.

In the case of the TCXO, this is a local open-loop system; that is, it depends on holding its frequency without active feedback. The eBay listing indicates (from the photo) its frequency is 24.000000 MHz. If we accept that means that only the last digit varies by +/- 1 when the frequency is set at the factory, that would indicate that it would have an accuracy of +/- 1.3 seconds in one year. For twenty bucks that isn't bad at all, if we believe the claim of accuracy.

In the case of the receiver, it's taking its time signal from the GPS network, which, as I understand it, is derived from earth-bound atomic clock sources, which are the ultimate manifestation of stable and accurate timebases. Buying access to an atomic timebase for about eight bucks is absolutely astonishing.

In both cases, developing a readout for the timebase would cost several times the price of the timebase itself. However, that's still not a big swinger in the days of Arduino and Raspberry Pi computers. Having been obsessed with accurate time since as a kid I built a WWV receiver, I am blown away by the incredible technical times we live in.
 

Robert Gift

Registered User
Nov 12, 2012
569
74
28
Denver, Colorado
Country
Region
... Having been obsessed with accurate time since as a kid I built a WWV receiver, I am blown away by the incredible technical times we live in.
I was also Bobsessed with time.
In 1971?, I made a pilgrimage to see the atomiclock!
George C. Hahn, government employee athe National Bureau of Standards, very kindly took me to see the atomiclock in Boulder, Colorado.
(I wondered if my presence in its room affected it in any way.)

In 1965 a classmate claimed that he.ard WWV on his walky-talky. "Impossible!", I assured.
Later discovered that another friend's father, electrical engineer and HAM radioperator, would key his walky-talky athe radio receiver andisten to his other walky-talky to set his three grandmother clocks thoughouthe house.
Located near Washington DC?, WWV broadcast Eastern Standard Time. I was displeased that it was not Greenwich Mean Time.

When did NBS establish 303 499-7111 to listen to WWVA?
Now Universal Coordinated Time - the lastime I listened.
(Since there was no leap second this year, I did not listen the afternoon of December 31st as I usuallyvould.)
 
  • Like
Reactions: Roy Gardner

Tim Orr

NAWCC Star Fellow
NAWCC Member
Donor
Sep 27, 2008
1,930
425
83
Boulder CO
Country
Region
Good afternoon, all!

One small proviso regarding "radio-controlled" "atomic" clocks: I have been told (and it seems logical, though I have not been able to either observe or verify it) that such clocks are deliberately set up to run "ever so slightly" fast. The rationale is that it is far easier to "slow down" a mechanical clock than it is to "speed it up." This is probably not true of such clocks as have only a digital display.

Anybody else heard this, or can anyone verify?

Best regards!

Tim Orr
 

timw

Registered User
Jan 21, 2023
9
6
3
74
Country
Two different approaches to the same end goal - a highly accurate time base.
Thought provoking, and after a little (what could go wrong with that?) it seems to me the distinction pivots on distinguishing clocks from time (and from people who trouble to build their own: in a significant sense they become the clock). Perhaps also a distinction between the how and the what. My radio-controlled and GPS "clock" friends want to know what time it is, exactly. And to achieve that give up on personal clocks in favour of a receiver and display. To be sure, their devices also keep time between "messages," but that function is usually performed by a simple quartz clock. If it works as it should, with several time checks per day, then never more than a small fraction of a second off.

On the other side is people who want a clock. That is, a free-standing independently operating clock, with the further proviso that it keep time about as accurately as anyone could reasonably want - my notion of "reasonable" being what works for marine/celestial navigation. Of course real ship owners don't really care about clocks or time (in this context), they just want to know where they are!

I suppose the grown-up choice wrt time is the GPS clock, and Seiko is out with a battery-powered GPS clock radio, GP501, for about $125. Not on Amazon yet, however.

And the traditional language of time systems makes this distinction most sharply, that between master and slave. The master being the clock, the slave merely displaying the time. No doubt the virtues of either in terms of ownership are purely derivative and of little substance, but for the time being I prefer an analog clock.
 
  • Like
Reactions: Roy Gardner

wisty

Registered User
Dec 24, 2014
130
76
28
Country
I'm not sure about celestial/astronomical clocks, I have no knowledg of how the very high precision clocks were used.
Navigation clocks were not required to keep "accurate" time over long periods. What was expected of them was to keep a highly consistent "rate" under all conditions. It was too hard/expensive to get the rate perfect.
A ship's chronometer that gained or lost a known number of seconds each 24 hours was perfectly acceptable as long as it always gained or lost those same number of seconds each 24 hours. The navigator could simply factor in the rate factor when doing the navigation calculations.
When chronometers were serviced, they were observed for a considerable period, and the rate established. When issued to a ship they would be accompanied by a certificate stating the rate. Once set to time the hands would not be touched again until the end of the voyage or the next service, the correct time would be found by calculation.
 

timw

Registered User
Jan 21, 2023
9
6
3
74
Country
Navigation clocks were not required to keep "accurate" time over long periods.
Understood, but a remark pertaining mainly to the history of marine chronometers and not-so-much the present. Clocks then could be depended on to be materially inaccurate, the saving grace then being consistent, "precise," error that could be bookkept. But it has been a long time since any marine chronometer worthy of name would be expected to be that inaccurate. And the practical problems of taking an accurate sight at sea mean that a few seconds are usually not going to matter too much. And even without GPS, a radio time check is enough to check the clock. Now of course there are a number of wristwatches, not yet cheap, that claim accuracy even to within ten seconds per year or better. And commercial ships, with GPS navigation systems, know exactly where they are moment-by-moment all day long. Bottom line, with the possibilities of modern accuracy, precision is less important, if for no other reason that even a precise clock is going to have accuracy problems even within its range of precision.
 
  • Like
Reactions: Roy Gardner

Schatznut

NAWCC Member
Sep 26, 2020
1,417
735
113
SoCal
Country
Region
Here's the difference between accuracy and precision. We define a distance of 1.0. We have two measuring sticks. One is accurate but not precise and the other is precise but not accurate. In the time domain, which would we rather have?
Accuracy vs Precision.jpg
 
  • Like
Reactions: Roy Gardner

Robert Gift

Registered User
Nov 12, 2012
569
74
28
Denver, Colorado
Country
Region
Here's the difference between accuracy and precision. We define a distance of 1.0. We have two measuring sticks. One is accurate but not precise and the other is precise but not accurate. In the time domain, which would we rather have?
View attachment 746742
Makes no sense. If the lowere precise, it would be 1.000 and stilbe the same length as the upper.
 

Schatznut

NAWCC Member
Sep 26, 2020
1,417
735
113
SoCal
Country
Region
You're missing the point, Robert. A measurement can be accurate but not precise (the upper one) or it can be precise but not accurate (the lower one). Accuracy and precision are totally unrelated to each other.
 

Mike Mall

NAWCC Member
Oct 27, 2021
486
167
43
Country
You're missing the point, Robert. A measurement can be accurate but not precise (the upper one) or it can be precise but not accurate (the lower one). Accuracy and precision are totally unrelated to each other.
So the TCXO controlled clock would be highly precise, but not necessarily accurate.(depending on the initial setting)
And the GPS controlled clock would be accurate, but not necessarily precise. (in between updates)
Am I following you here?
 

timw

Registered User
Jan 21, 2023
9
6
3
74
Country
So the TCXO controlled clock.... And the GPS controlled clock....
Lots of ways to think about this. Here's my best. Imagine a clock that is set to the wrong time and that cannot be reset. Observing it for a while, you realize that while it is wrong and measures the passage of time imperfectly, its rate of error is exactly constant. Compare that with a COSC certified mechanical watch accurate to within -4/+6 seconds per day. While the clock is precise, but not accurate at all, with it you can still figure out the exact time. And while the watch is accurate, never far off the mark, with it you will never, ever know the exact time. As it happens you have some task that requires knowing the exact time. Which do you choose, the accurate or the precise?

But we're living through an - the - historical period in which technology renders the distinction moot. Another generation or two and most folks will think of mechanical clocks and watches as quaint and eccentric, looking instead at their cell phones for the exact time. Oh, wait! That's now!
 
  • Like
Reactions: Mike Mall

Schatznut

NAWCC Member
Sep 26, 2020
1,417
735
113
SoCal
Country
Region
So the TCXO controlled clock would be highly precise, but not necessarily accurate.(depending on the initial setting)
And the GPS controlled clock would be accurate, but not necessarily precise. (in between updates)
Am I following you here?
Yes, that was the point I was trying to convey. For my purposes I'd rather have accuracy than precision. I'd rather be sure that my clock was correct when it said 12:18:45 (when that really was the time) than having had it wander off and tell me that it's 12:16:45.0001 when it's really 12:18:45.1. To explore this further we have to get into a discussion of random versus systematic error.
 
  • Like
Reactions: Mike Mall

Schatznut

NAWCC Member
Sep 26, 2020
1,417
735
113
SoCal
Country
Region
My understanding is that the 60 Hz is only precise over the long term, over the short term it can drift. If drift is detected then the power companies compensate for the drift by temporarily running at higher or lower than 60 Hz.
Actually, there's a bit more to that and it's not intuitive. The principle of emergence tells us why the power grid maintains its frequency so well. Robert Sapolsky's lecture at Stanford gives the best explanation I've ever seen, although he doesn't talk about the power grid at all. His "guess the weight of the ox" story gives the answer.
 

Cheezhead

Registered User
Dec 30, 2010
268
34
28
Wisconsin
Country
Region
Perhaps a moderator would move this thread to the Quartz clock forum to encourage its use.

For an inexpensive, accurate quartz clock there is a YouTube video showing a twisted pair of thin enamel insulated wires soldered to the quartz crystal connections to make a capacitor. The twisted pair was shortened to bring a fast-running clock into near perfect accuracy.

How to adjust "unadjustable" quartz clock.
 
  • Like
Reactions: timw

TQ60

Registered User
Sep 15, 2016
258
14
18
Madera CA
Country
Lots of stuff here.

Will repeat some as we have not read closely all posts.

The OP was relating the higher clock speed to better accuracy.

Others properly suggested temperature control and txco.

As a radio tech, the tcxo is very common.

There are different types and most are adjustable.

The ones with an oven are simple, they maintain a temperature well above ambient so no chance of varying temperature.

They have other stability additions.



The cold ones have other components that also react to ambient temperature in different ways than the crystal so as a package the frequency is stable over temperature.

Better ones are good to 1 PPM, 1 Part per MILLION.

does not matter what frequency is used, a percentage is a percentage.

The other part is the counter or divider.

This must count or divide consistently.

With most communications equipment now being programmable there are many versions of master clock oscillators that are very stable in small packages.

With all of the existing quartz clocks the rest is common.

One could go old school to be decorative and find an oven from maybe 800 mhz Micor radio as a source then use a collection of TTL chips to create a busy looking board to divide or count the clock down to 1 PPS Pulse Per Second.

The Micor oven is better than 1 PPM, divided down to 1 PPS would still retain the same accuracy.

Adding logic to light of LED displays to show time.

A skeleton clock quartz version.

Regarding things to buy, there needs to be a market for items.

The clock in your cellphone is controlled by GPS, every phone on any given carrier will tick at the exact same time.

A GPS controlled clock would be the ultimate.

It could have a supercapacitor as the power storage.

Solar cell for power source.

Lcd display for minimum power drain and be cast in clear resin so ther are no controls or connectors.

GPS uses an almanac to keep track of general location, this is for fast recovery from power outages.

To conserve energy the almanac is used to confirm the unit is still where it was if it restarts.

If it determines it moves the "brain" could wake up to determine by location the time zone and DST adjustment then turn off.

The clock would know where it is so it can adjust to GPS time to local time with zero interaction other than giving it some light.

Would not cost much but not many would want one.
 
  • Like
Reactions: Roy Gardner

Schatznut

NAWCC Member
Sep 26, 2020
1,417
735
113
SoCal
Country
Region
I'm just crazy enough to build one of these to drive a set of Nixies.
 

Roy Gardner

Ticker Stopper
NAWCC Member
Jun 14, 2016
79
56
18
Cameron Park, California
Country
Region
My understanding is that the 60 Hz is only precise over the long term, over the short term it can drift. If drift is detected then the power companies compensate for the drift by temporarily running at higher or lower than 60 Hz.
I read once that audiophiles with synchronous turntables didn't like it when the power companies made their daily correction to the total number of cycles in a day. Perhaps nowadays the correction is made over a long enough period that golden ears wouldn't notice.
 
Last edited:

Schatznut

NAWCC Member
Sep 26, 2020
1,417
735
113
SoCal
Country
Region
I read once that audiophiles with synchronous turntables didn't like it when the power companies made their daily correction to the total number of cycles in a day. Perhaps nowadays the correction is made over a long enough period that golden ears wouldn't notice.
Heck, the "Golden Ears" hear stuff that isn't even there! :chuckling: So do I but it probably has a lot to do with being around aircraft engines a lot.
 

TQ60

Registered User
Sep 15, 2016
258
14
18
Madera CA
Country
Forgot about the nixie tubes, that would look nice.

We saw a video couple weeks ago where someone is making the tubes and clocks.
 

Forum statistics

Threads
178,823
Messages
1,568,443
Members
53,992
Latest member
Trafxprt
Encyclopedia Pages
909
Total wiki contributions
3,077
Last edit
E. Howard & Co. by Steven Thornberry