My Casio g shock keeps extraordinary time. I did open it up and tune it,
but still I'd expect it to drift.

After 6 months untouched I still can't separate by eye the second from UTC.

Also, with regard to the video's query about all the clocks running slow -
they have been tuned to run at the temperature of a person's wrist.

On Sunday, 28 June 2015, John Stuart j.w.stuart@comcast.net wrote:

Very interesting observation. I wonder if Casio do the same long term analysis?
I don’t have as large a sample size of the same watch but I have had a small number of Casio quartz watches over the years, including the F-91W,and have noticed that they have all run fast, even from new and whether or not they are on the wrist. I suspect that this is a deliberate design feature, as the Japanese, along with some others, including myself, have a horror of being late for an appointment. Having a watch which designed to run just a little fast would help prevent lateness.

"Ceux qui sont prêts à abandonner une liberté essentielle pour obtenir une petite et provisoire sécurité, ne méritent ni liberté ni sécurité."
Benjimin Franklin

John,

Thanks for posting that. I will have to look this person up.

Given typical tempco plots for 32 kHz crystals I would assume the lost time is due to temperature: wall or room temperature is lower than wrist temperature. Someone on the list with that model watch should do a quick test to confirm. Also a measure of voltco would be interesting.

Note the days of quartz being adjusted with a trimmer capacitor are long gone. It's done digitally now. See, for example, paragraph 8.3 of http://www.nxp.com/documents/data_sheet/PCA2003.pdf

It's much more convenient to periodically delete pulses than have to add them so that's why wristwatch oscillators are designed to run fast rather than slow.

/tvb

More info:

32 kHz watch circuit with programmable output period and pulse width
http://www.nxp.com/documents/data_sheet/PCA2002.pdf

The Principle of Digital Trimming
http://www.emmicroelectronic.com/sites/default/files/public/products/datasheets/v3022_ds_0.pdf

Digital trimming for frequency adjustment
http://www.google.com/patents/US5530407

Method for digital frequency trimming an oscillator in an electronic timepiece
http://www.google.com/patents/US4282595

AVR4100: Selecting and testing 32kHz crystal oscillators for Atmel AVR microcontrollers
http://www.atmel.com/images/doc8333.pdf

Quartz Crystal Theory
http://www.jauchusa.com/ablage/med_00000818_1327049076_Quartz%20Crystal%20Theory%202007.pdf

Crystal Theory
http://www.euroquartz.co.uk/portals/0/pdf/tech-notes.pdf

MSP430 32-kHz Crystal Oscillators
http://www.ti.com/lit/an/slaa322b/slaa322b.pdf

Introduction to the quartz tuning fork
http://www.d.umn.edu/~jmaps/phys5061/labs/projects/qtf/ajp_qtf.pdf

The Accuracy and Stability of Quartz Watches
http://tf.nist.gov/general/pdf/2276.pdf

Hi

Just to clarify:

In the “art” the watches all ran fast rather than slow. They would have run slow if the
room temperature / skin temperature delta was an issue. Since they did not, one assumes
that Casio digitally compensates this model (and probably all their watches).  The typical
watch tuning fork will shift more than the observed delta when run in a cold court house if
un-compensated.

By far the best explanation is the “set to deliberately run fast” one.

Bob

But wouldn't normal watch wear just balance itself over time, one wears their watch for say 12 hours and the rest it sits on a counter at a much colder temperature. So wonder if Casio would actually go to such lengths to compensate. Maybe, interesting though.

-=Bryan=-

Back in the day when mechanical escapement pocket watches, and wrist
watches were state of the art, the jeweler would adjust the watch to
run at a normal rate, and give them a daily wind.  Everything looked
nice in the display case.

When a customer bought a watch, the jeweler would set the watch to
his shop clock, and instruct the customer to wear, and wind the watch
normally for two weeks, but do not set it.  At the end of the two
weeks, bring the watch back to the shop for a check up...

When the watch came back, the jeweler would calculate the number of days
the watch had been worn, note the difference from his shop clock, and
calculate the daily rate of the watch.  He would then set his timing
machine for the the inverse of that rate, and set the watch to match.

Now, when the customer wore his watch, the watch would seem to always
be right on because it was adjusted for a rate that compensated for
the customer's patterns of wearing the watch...his "personal error".

This trick had an added advantage because the customer got to see
how so-so his brand new watch behaved during those two weeks, and
got to be dazzled by his jeweler's rare ability to make the new
watch perform so much better than the factory could!

If this was normal back at the turn of the 20th century, why wouldn't
Casio, and others at least do as well?  Especially now that all
electronic watches have a microprocessor built in... complete with
temperature sensing diodes, battery monitors, and other nifty gadgets.

-Chuck Harris

Bryan _ wrote:

and that magic timing machine was the vibrograph,  more abot it here:
https://www.google.com/search?q=vibrograph&client=firefox-a&hs=Kyh&rls=org.mozilla:en-US:official&channel=np&tbm=isch&tbo=u&source=univ&sa=X&ei=iK6RVeX4FJTUoASf2KeIBQ&ved=0CB4QsAQ&biw=1024&bih=507
we had one
73
KJ6UHN

On 6/29/2015 12:16 PM, Chuck Harris wrote:

Hi

Well there are a few possibilities:

1. The person wears their watch 24 hours a day
2. They take it off for random periods of time
3. The person swims marathons weekly in cold water with their watch on
4. The watch sites in a cabinet for 10 years

The further down that list the colder the watch runs.

Cassio and pretty much everybody else do temperature compensate their watches for the nominal temperature characteristic of the crystal they use. It is easy to do digitally and thus costs nothing (either in silicon area or power). They also adjust the watch on frequency digitally rather than mechanically. Compared to the way we used to do it (40 years ago), it’s a lot cheaper to do it the “new way”. It is interesting that
the nominal set point is still “fast” though.

Bob

Hi

Most don’t, but some do. The ones that do care talk a lot. When they go online
with a “this watch is junk” post (with lots of data) it impacts the market.

That’s been done since the original quartz watch designs. It’s done with a robot
these days and is part of the checkout process on the module. It actually saves
money. The precision of the crystal (and other parts) used can be lower if you
do the calibration. Lower precision => better yield => lower cost.

The chip in the watch likely goes into just about every watch they make. The
size of the die is dictated by their ability to dice the wafer (roughly 1 mm square).
Anything smaller than that makes no sense. However many devices you can
pack into that much area, you get for free. With a modern process, that’s a lot
of devices.

One might think that running a much older process would be cheaper. That would only
work if you made more watch chips (by wafer lots) than you make of everything else.
That’s just not true. The answer is (as with everything else) that you keep up fairly
close to a modern process. The process parameters for an ultra low power watch
chip will be a bit different, but the geometry is dictated by the basic gear on the line.
They all rumble down the same basic line at one of many (massively large) foundaries

An accurate watch “with all the testing” sells for < $10. The market for an much
less accurate watch would (presumably) be at an even lower sell point….

Bob

Hi Andy,

I think you would be guessing wrongly.

The vast majority of watch owners don't want to ever have to set,
wind, adjust the calendar, or in anyway think, or fiddle with
their watch's time.  They want it to just be right.  In other
words, their disinterest makes them the "anti-time-nut".

The engineering team that designed the watch is where the concerns
about the watch always being "right" get turned into hardware.  It
costs them only a few days of an engineer's time to put, a temperature
sensor, battery voltage sensor, and a table describing the nominal
performance of the crystal at a normal range of temperatures and
voltages, into the watch.  It costs nothing on a per watch basis,
as it is a trivial amount of additional silicon... silicon that
resides within the necessary spaces between the pads used to connect
the silicon die to the circuit board.

As to per unit testing, the only testing required (after the initial
design phase) is to program an offset into the watch that makes up
for minor frequency variations in the crystal at nominal room
temperature.  Crystal manufacture is an imperfect process, so they
have to do that anyway.

-Chuck Harris

Andy wrote:

Hi

Even back in the 1970’s we used computing counters to read the 32 KHz. You
get 7 digits in about 10 ms these days. It will take you longer to power up the
module and stabilize it than the frequency reading takes.

The semiconductor process for the watch chips has always been a bit odd. They
never need anything faster than 32 KHz for a clock. They are after very low
leakage. Back in the old days having it run at 1 V was considered strange. It’s
pretty common today.

They use the same tricks to dump an analog function into a digital process as the
MCU people. It’s not a great analog sub system, but it’s good enough for the purpose.
They may even calibrate the temperature and voltage when the set the frequency. That
would let them get away with a lot of slop on those sub systems.

If they digitize the temperature range from 0 to 50 C, that’s plenty for any normal
environment the watch will see. A 5 bit ADC would be good enough for that task
(with the proper full scale input).

The same thing likely applies to the voltage. The circuit is unlikely to function below
half voltage on the battery. Again a few bits of ADC will tell you everything you
need to know to drive the table.

I’d bet that they run a sigma delta and are quite happy with information at a “once
a minute” sort of rate. There’s not a lot of “analog” stuff in that case. It would be
pretty easy to dump onto the die.

Bob