adjust quartz for accuracy

[JohnR725]

4 minutes ago, GPrideaux said:

Unless the test equipment is calibrated and certified by an accredited measurement lab you'd have no way of knowing how much of any difference measured was from error in the watch or in the test equipment itself.

You could always buy something like this to calibrate your equipment.

https://www.witschi.com/en/additional-products/gps-receiver/

 

[praezis]

7 hours ago, JohnR725 said:

always buy something like this to calibrate

Not always, John. Some buy just a GPS receiver, using the 1-pps signal directly (<< $50) without paying W… prices .

Link (German)

Frank

[JohnR725]

7 hours ago, praezis said:

Some buy just a GPS receiver, using the 1-pps signal directly (<< $50) without paying W… prices

 yes you could buy a GPS with a one pulse per second signal void paying the ripoff price a witschi but?  Is the normal navigation GPS going to be good enough? Yes the witschi is a ripoff price but it's an example of the GPS designed for timekeeping purposes and has a much better timekeeping system. So if you really want ultimate you buy a better GPS just not from witschi and not the typical navigation GPS is it just isn't going to be good enough if you're obsessed with timekeeping.

 

 

 

 

[praezis]

You have to use the 1-pps signal that most receivers provide. Often they just use it for blinking the led, without dedicated output.

Witschi cannot improve that signal. 

Accuracy of time pulse signal……. 60 ns
says the data sheet of the low cost receiver of my link. 
That means: error after 1sec measuring time is < 0.005 s/day -
after 10 sec measuring time it is < 0.0005 s/day 

Frank

[Kimmoth]

On 12/6/2024 at 12:21 AM, praezis said:

trimmers disappeared many decades ago

What about the groups of tiny jumpers, where one or two pairs of pads are soldered together? Would this be a factory calibration measure, and if so, I suppose a tool like the subject of this thread would allow one to reverse engineer its logic?

Hm, wait a minute - this PCB has a trimmer and the jumpers. Any idea what else they might be for?

 

On 12/6/2024 at 7:46 AM, JohnR725 said:

Then whatever you're comparing You are quartz watch to has to keep time better than the quartz watch by quite a bit.

Doesn't my phone keep amazing time because it has to do GPS? Its own inaccuracy would be corrected probably at least once a minute, wouldn't it?

Edited December 8, 2024 by Kimmoth
Added pic

[tomh207]

7 hours ago, Kimmoth said:

What about the groups of tiny jumpers, where one or two pairs of pads are soldered together? Would this be a factory calibration measure, and if so, I suppose a tool like the subject of this thread would allow one to reverse engineer its logic?

Hm, wait a minute - this PCB has a trimmer and the jumpers. Any idea what else they might be for?

 

Doesn't my phone keep amazing time because it has to do GPS? Its own inaccuracy would be corrected probably at least once a minute, wouldn't it?

What are you referring to as jumpers?

 

phones have not always had a gps receiver, their time reference would be “network “ time from the cell radio service. The cellular network would reference three or more time servers within the provider network and they in turn would reference master time servers usually referenced to the international atomic clock network. Some ISP’s and companies now use specialised gps receivers within their networks for the time reference, especially relevant if timekeeping is critical.

 

Tom

[Kimmoth]

These are the jumpers. 

As for the phone, you've detoured from the actual situation here; my phone does have GPS, like any almost any phone made in the last ten years, and thus keeps ridiculously accurate time, which meets the criterion stipulated earlier: far more accurate than the thing we want to check the accuracy of.

Edited December 9, 2024 by Kimmoth

[JohnR725]

26 minutes ago, tomh207 said:

What are you referring to as jumpers?

I think what he means are the items in the image that I've circled. I don't think these are actually for timing there probably to select functions or their test points. A lot of times you might have one circuit board for several types of watches perhaps?

the problem with slowly doing an answer is I see that already clarified but I will leave this here as it is still a valid answer.

29 minutes ago, tomh207 said:

specialised gps receivers within their networks for the time reference

by the way that brings up a problem with what is quoted below?

On 12/7/2024 at 12:11 PM, praezis said:

Witschi cannot improve that signal. 

Accuracy of time pulse signal……. 60 ns
says the data sheet of the low cost receiver of my link. 
That means: error after 1sec measuring time is < 0.005 s/day -
after 10 sec measuring time it is < 0.0005 s/day 

the 60 nano seconds is achievable 99% of the time and the GPS can do better according to the spec sheet below.

let's look at the specifications sheet and we have a minor problem

notice in the features timing is blank? let me quote myself?

On 12/7/2024 at 9:50 AM, JohnR725 said:

Yes the witschi is a ripoff price but it's an example of the GPS designed for timekeeping purposes and has a much better timekeeping system.

so the witschi GPS is extremely expensive because who exactly needs one anyway? Most watchmakers never get there timing machines calibrated the vintage paper tape machines existed in shops for basically forever never being calibrated and nobody really ever cared even the modern witschi timing machines how may people ever get them calibrated probably a very tiny percentage and then they send them someplace so the quantity of witschi GPS calibrators are sold is probably extremely tiny which is why it's extremely expensive. But it's still a GPS designed for timekeeping purposes not navigation not that I can't be used for navigation.

let's snip out something from the witschi GPS users guide. Amusingly of course they're going to pick and choose and leave out details like the normal GPS would be just fine. At least for normal stuff including normal timing. But they do get things right the timing GPS needs more GPS signals has better circuitry just produces a much nicer timing signal.

now let's go backwards in time

let's drop back one generation version of Frank talks about and go to the version 6 and snip out image showing the various versions that were available. although if you do look on eBay the version 6 still seems be available even though they considered obsolete. You will note one version is specifically for timing.

then let's see what it says about the timing GPS

I have a PDF on timekeeping GPS and another one on problems of not having a clean signal.

On 12/7/2024 at 2:07 AM, praezis said:

Some buy just a GPS receiver, using the 1-pps signal directly (<< $50) without paying W… prices .

then we get the whole amusement of this silly discussion. Yes there are special GPS receivers for timekeeping purposes capable producing really nice signals consistently or there is the GPS module you can buy on eBay for considerably less than $50 and will work 99.9% of the time. then let me quote freight again worst-case with the particular module he was recommending is outstanding as quoted below.

On 12/7/2024 at 12:11 PM, praezis said:

That means: error after 1sec measuring time is < 0.005 s/day -
after 10 sec measuring time it is < 0.0005 s/day 

 

https://www.u-blox.com/en/product/neo-7-series

TI_GPS_PPS_Timing_AppNote.pdf Timing_AppNote_(GPS.G6-X-11007).pdf

[praezis]

Great research, I am impressed!

The Wit... info is a bit outdated, I suspect. John, did you find out when their calibrating device appeared?

Development of those receivers is ultra fast.  Some years ago there were those providing the GPS messages only (text) but no time pulse.

The text messages appear each second, too, but accuracy is lower ("0.01 to 0.001 s").
Wit... does not say that they refer just to accuracy of the message-1 s. That will frighten users and make them want the "right" device with super-duper accuracy, available only from Wit... 

I don't know why that table above misses a dot at "Timing" but my receiver is providing the 1-pps time pulse with already mentioned accuracy. Is it the missing 1% ("99%")? No idea, but not important at all for me.

Frank

[JohnR725]

6 hours ago, praezis said:

did you find out when their calibrating device appeared?

I was looking through my technical data for GPS stuff and 2019 but that conflicts with what were both thinking of. So a little more digging and the earliest one as I have is 2011 I will attach both user guides. 

 

Witschi GPS Receiver GPS timimg machine User Manual EN 2011.pdf Witschi GPS RECEIVER 19.91D35e_3.1.1.pdf

[LittleWatchShop]

Hey, what time is it?

[Kimmoth]

Depends - Einstein

[LittleWatchShop]

I guess it is fun to ponder ways to calibrate a quartz watch down to 0.1ppm, but, I don't know...beyond the fun, what is the point?

When I was a senior in college studying EE, my senior project was a precision temperature compensated oscillator for the purpose of calibrating quartz watches.  This was back at the beginning of the quartz era (I am old!). My dad was a watchmaker and I was doing this project for him.  Anyway, it did not turn out all that well.

In my recent estate buy out, I acquired several machines that measure quartz watch frequency.  At least one works.  No idea how accurate though.  You dont pull crystals in modern quartz watches.  Instead, the calibration is done digitally. So, for a modern watch the tool is useless.

It is 8:23 am here.

[praezis]

48 minutes ago, LittleWatchShop said:

I guess it is fun to ponder ways to calibrate a quartz watch down to 0.1ppm, but …

No, it would be no fun - it is not even possible: 0.1 ppm = 0.009 s/day.

Topic was devices that can and are appopriate to test quartz watch accuracy. Adjustment would be another issue/topic.

53 minutes ago, LittleWatchShop said:

At least one works.  No idea how accurate though. 

Here the topic „GPS time pulse“ that John and me were discussing, comes into play - for those who are interested.
 

57 minutes ago, LittleWatchShop said:

So, for a modern watch the tool is useless.

Useless for adjustments yes, but not absolute. It still will show if the quartz oscillator is dead or alive, if the coil is ok or broken, if the rate is out of specs. Enough? 

In fact I don‘t use my quartz watch tester for watches. I use it for (digitally) adjusting the 32k quartz oscillators of my electronics.

Frank