Is a Gray code counter more energy efficient?...

[Ricky]

On Monday, June 19, 2023 at 1:53:59 PM UTC-4, whit3rd wrote:

On Monday, June 19, 2023 at 12:03:12 AM UTC-7, Ricky wrote:
On Monday, June 19, 2023 at 2:41:21 AM UTC-4, whit3rd wrote:
On Sunday, June 18, 2023 at 11:03:50 PM UTC-7, Ricky wrote:
On Monday, June 19, 2023 at 12:45:39 AM UTC-4, whit3rd wrote:
On Sunday, June 18, 2023 at 9:36:03 PM UTC-7, Ricky wrote:
On Monday, June 19, 2023 at 12:21:23 AM UTC-4, whit3rd wrote:
On Sunday, June 18, 2023 at 8:47:48 PM UTC-7, Ricky wrote:

Uh, how do you count down without using flipflops?

Shift register, gates determine the inshifted bit... there\'s latched bits (flipflops, if you will)
but not \'toggle\' control of them. The pseudorandom sequence generated only goes
to all-ZERO outputs once per cycle, but cannot do all-ONE outputs without stalling.
So, it cannot do divide-by-2^N, but at most divide-by-((2^N) -1).

Sorry, I have no idea what you are trying to say about this. So they use FFs. Glad we have that clear. I\'m familiar with pseudo random sequences, but what advantage to they provide here?
...
I have no idea what you are talking about. Sorry.
The point of a clocked timer is that it indicates when it has run a full many-clocks
sequence. Both an array of flipflops, and a pseudorandom sequence generator, take
many clocks to complete a sequence. Both are clocked, and can indicate a time at the
end of the sequence. The fact that \'pseuorandom sequence generator\' makes a
sequence is interesting, but it is NOT the only function that that particular kind of
hardware can serve. It also divides down a fast clock, if you note the occurrences of
repetition of its state.

Sorry, you just are not explaining yourself. You don\'t need to explain counters to me. I just need to understand what you are doing. You seem to flail all around with no sense of purpose.

Oh, BTW, you seem to ignore the need for an N bit state detector for whatever is making your sequence generator. That\'s even more logic.

--

Rick C.

-+-+ Get 1,000 miles of free Supercharging
-+-+ Tesla referral code - https://ts.la/richard11209

 

[whit3rd]

On Monday, June 19, 2023 at 11:34:04 AM UTC-7, Ricky wrote:
> On Monday, June 19, 2023 at 1:52:11 PM UTC-4, whit3rd wrote:

[about a wristwatch with crystal clock and divider...]

...The cascade of flipflops that eventually
creates the one minute pulse for minute-hand incremental movement includes
lots of faster-than-a-second values, which can drive multiple phases of subsecond
period. An N-phase clock delivers N pulses to move an M-pole rotor by 1/M of
a revolution, in a synchronous AC motor such as the electrostatic drive of the second hand.

Thus, to achieve 1.0 revolution of the second hand, in one minute, smoothly,
you want a frequency F that satisfies
1 minute= 60 s = M/(F/N)
where F would, most conveniently, be a power-of-two multiple...

Why would a power of 2 be \"most\" convenient? I fail to see the significance. Dividing by any integer is trivial. Most effective is a loadable counter with a carry chain. When the counter reaches all 1s, the counter is loaded with the appropriate preset.

A multiplicity of powers of 2 are already present, in the (presumably) ripple-counter
that divides down a 32.nnnn kHz rock to one minute, each stage being a divide-by-two flipflop.

of the one-per-minute pulse that drives
the minute hand. That implies that one can conveniently set M * N =2^(any small integer) to achieve
the speed one wants.

Only if you are willing to work with an arbitrary F.

The crystal frequency most convenient is what one would use, of course.
It isn\'t arbitrary, it\'s specified.

It\'s not clear that a clock division by 60 results in a useful divisor for minutes or seconds or hours.

It\'s clear to me. What part do you not understand? Dividing 1 by 60 gives 1/60 which is the period of a minute. Is that not clear?

Dividing a second clock by 60 does that, but there IS NO SECOND CLOCK.
The powers-of-two clock division goes down to MINUTES to form a minute clock,
and 1/64 minute is the closest item in the division chain to a second.
We want uniform revolution of the second hand to complete one cycle
in one minute; the seconds aren\'t electrical events, they\'re markings under the second
hand on the watch face.

 

[Ricky]

On Monday, June 19, 2023 at 9:55:11 PM UTC-4, whit3rd wrote:

On Monday, June 19, 2023 at 11:34:04 AM UTC-7, Ricky wrote:
On Monday, June 19, 2023 at 1:52:11 PM UTC-4, whit3rd wrote:
[about a wristwatch with crystal clock and divider...]

...The cascade of flipflops that eventually
creates the one minute pulse for minute-hand incremental movement includes
lots of faster-than-a-second values, which can drive multiple phases of subsecond
period. An N-phase clock delivers N pulses to move an M-pole rotor by 1/M of
a revolution, in a synchronous AC motor such as the electrostatic drive of the second hand.

Thus, to achieve 1.0 revolution of the second hand, in one minute, smoothly,
you want a frequency F that satisfies
1 minute= 60 s = M/(F/N)
where F would, most conveniently, be a power-of-two multiple...
Why would a power of 2 be \"most\" convenient? I fail to see the significance. Dividing by any integer is trivial. Most effective is a loadable counter with a carry chain. When the counter reaches all 1s, the counter is loaded with the appropriate preset.
A multiplicity of powers of 2 are already present, in the (presumably) ripple-counter
that divides down a 32.nnnn kHz rock to one minute, each stage being a divide-by-two flipflop.

So? If you are designing a watch, and you find designing a divide by N counter to be remotely anything other than \"convenient\", you are in the wrong business.

of the one-per-minute pulse that drives
the minute hand. That implies that one can conveniently set M * N =2^(any small integer) to achieve
the speed one wants.

Only if you are willing to work with an arbitrary F.
The crystal frequency most convenient is what one would use, of course.
It isn\'t arbitrary, it\'s specified.

You seem to be bantering wordplay now. I thought we were discussing engineering. Assume you have a boss, who asks, \"Why are you using an oddball crystal frequency, when we already buy tons of 32,768 Hz crystals? Do you not know how to design a counter to produce a 1 minute clock from that frequency? If not, why would we let you design a chip for this watch?\" How do you respond?

It\'s not clear that a clock division by 60 results in a useful divisor for minutes or seconds or hours.

It\'s clear to me. What part do you not understand? Dividing 1 by 60 gives 1/60 which is the period of a minute. Is that not clear?
Dividing a second clock by 60 does that, but there IS NO SECOND CLOCK.

Only because you are using word play rather than designing anything.

The powers-of-two clock division goes down to MINUTES to form a minute clock,
and 1/64 minute is the closest item in the division chain to a second.

Only if you are constrained on dividing by binary powers, which you are not..

We want uniform revolution of the second hand to complete one cycle
in one minute; the seconds aren\'t electrical events, they\'re markings under the second
hand on the watch face.

Who cares??? You are saying the second hand does not use a second clock. That\'s irrelevant to the minute hand.

I have no idea what you are really thinking or what you are trying to do. Divide 32786 Hz by 32768 * 4 * 15 giving... wait for it... a minute clock. No rocket science involved and no need to specify unique, specific frequency crystals. Now, you only need to specify all the other details that impact the oscillator, start up, initial accuracy, drift, temperature compensation, etc. That\'s where you should focus your attention, to making the oscillator work as well as possible given the harsh environment it will be operating in. What\'s the point of an expensive, fancy watch that doesn\'t work any better than a cheap dollar store watch?

Of course, this is your idea, your product, except that it\'s not a product at all. It\'s play and you are free to do anything you wish with it.

--

Rick C.

-++- Get 1,000 miles of free Supercharging
-++- Tesla referral code - https://ts.la/richard11209

 

[whit3rd]

On Monday, June 19, 2023 at 7:19:09 PM UTC-7, Ricky wrote:

On Monday, June 19, 2023 at 9:55:11 PM UTC-4, whit3rd wrote:
On Monday, June 19, 2023 at 11:34:04 AM UTC-7, Ricky wrote:
On Monday, June 19, 2023 at 1:52:11 PM UTC-4, whit3rd wrote:
[about a wristwatch with crystal clock and divider...]

...The cascade of flipflops that eventually
creates the one minute pulse for minute-hand incremental movement includes
lots of faster-than-a-second values, which can drive multiple phases of subsecond
period. An N-phase clock delivers N pulses to move an M-pole rotor by 1/M of
a revolution, in a synchronous AC motor such as the electrostatic drive of the second hand.

Thus, to achieve 1.0 revolution of the second hand, in one minute, smoothly,
you want a frequency F that satisfies
1 minute= 60 s = M/(F/N)
where F would, most conveniently, be a power-of-two multiple...
Why would a power of 2 be \"most\" convenient? I fail to see the significance. Dividing by any integer is trivial. Most effective is a loadable counter with a carry chain. When the counter reaches all 1s, the counter is loaded with the appropriate preset.
A multiplicity of powers of 2 are already present, in the (presumably) ripple-counter
that divides down a 32.nnnn kHz rock to one minute, each stage being a divide-by-two flipflop.
So? If you are designing a watch, and you find designing a divide by N counter to be remotely anything other than \"convenient\", you are in the wrong business.
of the one-per-minute pulse that drives
the minute hand. That implies that one can conveniently set M * N =2^(any small integer) to achieve
the speed one wants.

Only if you are willing to work with an arbitrary F.
The crystal frequency most convenient is what one would use, of course.
It isn\'t arbitrary, it\'s specified.

You seem to be bantering wordplay now. I thought we were discussing engineering. Assume you have a boss, who asks, \"Why are you using an oddball crystal frequency, ...

It\'s my choice; I\'m the customer.

It\'s not clear that a clock division by 60 results in a useful divisor for minutes or seconds or hours.

It\'s clear to me. What part do you not understand? Dividing 1 by 60 gives 1/60 which is the period of a minute. Is that not clear?
Dividing a second clock by 60 does that, but there IS NO SECOND CLOCK.
Only because you are using word play rather than designing anything.

The powers-of-two clock division goes down to MINUTES to form a minute clock,
and 1/64 minute is the closest item in the division chain to a second.
Only if you are constrained on dividing by binary powers, which you are not.

You can choose to be constrained to one particular crystal. I do not so choose.

We want uniform revolution of the second hand to complete one cycle
in one minute; the seconds aren\'t electrical events, they\'re markings under the second
hand on the watch face.

Who cares??? You are saying the second hand does not use a second clock. That\'s irrelevant to the minute hand.

Huh? The minute hand wasn\'t going to need a divide-by-60, and the second hand doesn\'t need
a divide-by-60, and the hour hand uses gears, not electronics. Why
are you thinking there\'s a divide-by-60 issue? What got your head stuck in that particular box?

One makes seconds from a 32.768 kHz crystal, with a few flipflops.
One makes minutes from a 30.720 kHz crystal, the same way.

 

[Ricky]

On Monday, June 19, 2023 at 11:23:08 PM UTC-4, whit3rd wrote:

On Monday, June 19, 2023 at 7:19:09 PM UTC-7, Ricky wrote:
On Monday, June 19, 2023 at 9:55:11 PM UTC-4, whit3rd wrote:
On Monday, June 19, 2023 at 11:34:04 AM UTC-7, Ricky wrote:
On Monday, June 19, 2023 at 1:52:11 PM UTC-4, whit3rd wrote:
[about a wristwatch with crystal clock and divider...]

...The cascade of flipflops that eventually
creates the one minute pulse for minute-hand incremental movement includes
lots of faster-than-a-second values, which can drive multiple phases of subsecond
period. An N-phase clock delivers N pulses to move an M-pole rotor by 1/M of
a revolution, in a synchronous AC motor such as the electrostatic drive of the second hand.

Thus, to achieve 1.0 revolution of the second hand, in one minute, smoothly,
you want a frequency F that satisfies
1 minute= 60 s = M/(F/N)
where F would, most conveniently, be a power-of-two multiple...
Why would a power of 2 be \"most\" convenient? I fail to see the significance. Dividing by any integer is trivial. Most effective is a loadable counter with a carry chain. When the counter reaches all 1s, the counter is loaded with the appropriate preset.
A multiplicity of powers of 2 are already present, in the (presumably) ripple-counter
that divides down a 32.nnnn kHz rock to one minute, each stage being a divide-by-two flipflop.
So? If you are designing a watch, and you find designing a divide by N counter to be remotely anything other than \"convenient\", you are in the wrong business.
of the one-per-minute pulse that drives
the minute hand. That implies that one can conveniently set M * N =2^(any small integer) to achieve
the speed one wants.

Only if you are willing to work with an arbitrary F.
The crystal frequency most convenient is what one would use, of course.
It isn\'t arbitrary, it\'s specified.
You seem to be bantering wordplay now. I thought we were discussing engineering. Assume you have a boss, who asks, \"Why are you using an oddball crystal frequency, ...

It\'s my choice; I\'m the customer.

Who\'s customer???

It\'s not clear that a clock division by 60 results in a useful divisor for minutes or seconds or hours.

It\'s clear to me. What part do you not understand? Dividing 1 by 60 gives 1/60 which is the period of a minute. Is that not clear?
Dividing a second clock by 60 does that, but there IS NO SECOND CLOCK..
Only because you are using word play rather than designing anything.

The powers-of-two clock division goes down to MINUTES to form a minute clock,
and 1/64 minute is the closest item in the division chain to a second..
Only if you are constrained on dividing by binary powers, which you are not.
You can choose to be constrained to one particular crystal. I do not so choose.
We want uniform revolution of the second hand to complete one cycle
in one minute; the seconds aren\'t electrical events, they\'re markings under the second
hand on the watch face.

Who cares??? You are saying the second hand does not use a second clock.. That\'s irrelevant to the minute hand.
Huh? The minute hand wasn\'t going to need a divide-by-60, and the second hand doesn\'t need
a divide-by-60, and the hour hand uses gears, not electronics.

You seem to be deliberately thick about this. I\'ve explained it clearly, even giving you the numbers. 32,768 Hz / (32,678 * 4 * 15) 4 * 15 is 60. It makes sense to split the 60 into a 4, which can be combined with the 32,768 leaving the 15. So, technically, you are right. There\'s no divide by 60.

Why
are you thinking there\'s a divide-by-60 issue? What got your head stuck in that particular box?

No, I\'m just being realistic about designing a watch. Also, dividing by 60 creates no issues. It\'s easy to do. It\'s often a small part of a junior level lab project.

One makes seconds from a 32.768 kHz crystal, with a few flipflops.
One makes minutes from a 30.720 kHz crystal, the same way.

Your minutes approach uses the identical number of FFs, not that it matters.. You are designing a watch. What is the full feature set? Once you tell me that, I will give your more info about the divider chain.

No, I won\'t. You are not designing a watch. You are designing a divider chain for an imaginary timepiece with no other information. You are clearly being argumentative at this point. You have zero requirements. You\'ve defined nothing about a product to be designed. You clearly don\'t understand digital electronics in detail, having made many mistakes. I can\'t see any reason to continue this discussion.

--

Rick C.

-+++ Get 1,000 miles of free Supercharging
-+++ Tesla referral code - https://ts.la/richard11209

 

[Martin Brown]

On 19/06/2023 18:52, whit3rd wrote:

On Monday, June 19, 2023 at 2:08:12 AM UTC-7, Martin Brown wrote:
On 18/06/2023 22:01, whit3rd wrote:
On Sunday, June 18, 2023 at 1:25:15 PM UTC-7, Martin Brown wrote:
On 18/06/2023 03:18, whit3rd wrote:
On Saturday, June 17, 2023 at 11:26:09 AM UTC-7, Ricky wrote:
On Saturday, June 17, 2023 at 12:16:38 PM UTC-4, whit3rd wrote:

[about Accutron DNA watch]
The minute-hand \'tick\' is the slowest clock required for that watch (with minutes/hours
being gear-connected). A crystal with power-of-two division to minutes(rather than seconds)
would be expected; nearest to the common 32768 Hz would be 30720 Hz.
Digikey doesn\'t stock that particular value, though: it\'d be a custom rock.

Why are you so afraid of divide by 60? It is trivial as /6 /10 with
classical digital hardware chips. Or /3 /4 /5 depending on your outlook.

That is why the Babylonians so liked base 60!

The second hand is NOT stepping at 1 Hz, though; it\'s a synchronous electrostatic motor,
so there\'s no benefit to \'dividing down\' to one-per-second, rather one can just factor
any 2^N precursor of the one minute clock, and use a number of poles and
phases for the drive to accomplish the smooth sweep hand motion.

But there is an advantage to dividing down to once per minute.

Divide by 3 and 5 are not that hard - eg form first principles:

https://pages.mtu.edu/~suits/electronics/Divide_by_3&5_circuit.html

But they can divide down to 8Hz or 16Hz just as easily and that is more
than good enough to fool the human eye into seeing a continuous sweep
motion on the second hand. ISTR Rolex uses 8Hz on most of its watches.

The point here, though, is that lower frequencies than 32 kHz are used to step the
minute hand, while the 60:1 ratio that drives the slower hour hand is done by gearing,

Have you not noticed that there are also 60 seconds in a minute?

rather than electronically. The cascade of flipflops that eventually
creates the one minute pulse for minute-hand incremental movement includes
lots of faster-than-a-second values, which can drive multiple phases of subsecond
period. An N-phase clock delivers N pulses to move an M-pole rotor by 1/M of
a revolution, in a synchronous AC motor such as the electrostatic drive of the second hand.

Actually you can microstep a M pole motor in half steps by driving
adjacent segments and so get a smoother less jerky drive. Just about all
stepper motor implementations on telescopes did it that way.

https://www.trinamic.com/technology/motor-control-technology/microstepping/

Even the simplest form of this was quite a big improvement.

Increasingly now they use DC servos.

Not sure if it is applicable to the sort of electrostatic motor that
they might use in watches but it avoids having clunky impulses as each
successive node turns on. Smoother motion means less vibration.

Thus, to achieve 1.0 revolution of the second hand, in one minute, smoothly,
you want a frequency F that satisfies
1 minute= 60 s = M/(F/N)
where F would, most conveniently, be a power-of-two multiple of the one-per-minute pulse that drives
the minute hand. That implies that one can conveniently set M * N =2^(any small integer) to achieve
the speed one wants.

It\'s not clear that a clock division by 60 results in a useful divisor for minutes or seconds or hours.

If you are starting from a common as muck standard crystal that is
trimmed for 2^N Hz then it really does make sense to go through 1Hz and
use 8Hz (eg Rolex) for the \"continuous\" sweep second hand drive.

ISTR some early models used 6Hz which is visibly more jerky and copy
watches use 1Hz (or are there any that do look more convincing now?).

For 2^N you divide down to there with a chain of flip-flops each one
clocked by the output of the previous stage. Classic ripple counters

https://www.ti.com/lit/ds/symlink/cd4020b.pdf

Or you could in extremis go the whole hog and divide by a custom N eg.

https://www.ti.com/lit/ds/symlink/cd74hc4059.pdf

Snag is that I don\'t think there are any other dirt cheap low frequency
crystals (you might find 200kHz and 60kHz but they will require more
power than the humble 32k watch Xtal).

There is no need for a custom crystal for this. Unlike the situation for
a sidereal clock where making it run ~4 minutes a day faster than mean
solar time is way beyond the pull range of a crystal.

The tuning-fork crystals are tuned by electroplating the tines and
then trimming off excess weight with a laser... it\'s trivial to change
the process and make a custom crystal. Old AT-cut disks were tuned by

Only if you are prepared to buy a batch of about 100k all the same and
they would be nothing like as cheap as the mass produced watch crystals
at 32768Hz. I recall someone doing that way back when digital clock
chips were new and it allowed stock parts to make sidereal rate clocks.

Up to then they had to be bespoke made observatory clocks or a computer
program that converted UTC and longitude into local sidereal time.

> So, I expect that a custom crystal is an economic thing for this purpose, today.

Compared to an insanely over priced watch it might be economic but it
still wouldn\'t be sensible engineering. You seem to be labouring under
the misapprehension that division by 60 is difficult. It isn\'t it has
been a solved problem since at least the advent of digital watches and
probably long before that. I can recall digital clocks using nixie tubes
- even built one from TTL parts salvaged from scrapped ICL 1900 boards.

--
Martin Brown

 

[server]

On Tue, 20 Jun 2023 11:24:46 +0100, Martin Brown
<\'\'\'newspam\'\'\'@nonad.co.uk> wrote:

Snag is that I don\'t think there are any other dirt cheap low frequency
crystals (you might find 200kHz and 60kHz but they will require more
power than the humble 32k watch Xtal).

With 60 or 240 kHz crystal it is easier to implement 1/10 s (or 1/100
s) resolution stop watch functionality.

How do you implement such resolution with a 32768 Hz crystal ?

 

[Anthony William Sloman]

On Tuesday, June 20, 2023 at 11:01:05 PM UTC+10, upsid...@downunder..com wrote:

On Tue, 20 Jun 2023 11:24:46 +0100, Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote:


Snag is that I don\'t think there are any other dirt cheap low frequency
crystals (you might find 200kHz and 60kHz but they will require more
power than the humble 32k watch Xtal).
With 60 or 240 kHz crystal it is easier to implement 1/10 s (or 1/100
s) resolution stop watch functionality.

How do you implement such resolution with a 32768 Hz crystal ?

For equally spaced 1 second steps you divide by 546 52 times and 547 8 times.

For 0.1 sec you can\'t do it exactly, but dividing by 3,277 nine times and 3,275 once comes pretty close.

For 0.01 sec you divide by 328 68 times and 327 32 times which is again fairly close.

If you want to get cranky you can set up a phase locked loop to lock a 100Hz oscillator driving a 10:1 digital divider to the roughly 0.1Hz output - the low pass filter in the phase locked loop will take out most of the jitter , such as it is.

--
Bill Sloman, Sydney

 

[Ricky]

On Tuesday, June 20, 2023 at 9:01:05 AM UTC-4, upsid...@downunder.com wrote:

On Tue, 20 Jun 2023 11:24:46 +0100, Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote:


Snag is that I don\'t think there are any other dirt cheap low frequency
crystals (you might find 200kHz and 60kHz but they will require more
power than the humble 32k watch Xtal).
With 60 or 240 kHz crystal it is easier to implement 1/10 s (or 1/100
s) resolution stop watch functionality.

How do you implement such resolution with a 32768 Hz crystal ?

Do you really not know how?

--

Rick C.

+--- Get 1,000 miles of free Supercharging
+--- Tesla referral code - https://ts.la/richard11209

 

[Ricky]

On Tuesday, June 20, 2023 at 10:10:38 AM UTC-4, Anthony William Sloman wrote:

On Tuesday, June 20, 2023 at 11:01:05 PM UTC+10, upsid...@downunder.com wrote:
On Tue, 20 Jun 2023 11:24:46 +0100, Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote:


Snag is that I don\'t think there are any other dirt cheap low frequency
crystals (you might find 200kHz and 60kHz but they will require more
power than the humble 32k watch Xtal).
With 60 or 240 kHz crystal it is easier to implement 1/10 s (or 1/100
s) resolution stop watch functionality.

How do you implement such resolution with a 32768 Hz crystal ?
For equally spaced 1 second steps you divide by 546 52 times and 547 8 times.

For 0.1 sec you can\'t do it exactly, but dividing by 3,277 nine times and 3,275 once comes pretty close.

For 0.01 sec you divide by 328 68 times and 327 32 times which is again fairly close.

If you want to get cranky you can set up a phase locked loop to lock a 100Hz oscillator driving a 10:1 digital divider to the roughly 0.1Hz output - the low pass filter in the phase locked loop will take out most of the jitter , such as it is.

NCO

--

Rick C.

+--+ Get 1,000 miles of free Supercharging
+--+ Tesla referral code - https://ts.la/richard11209

 

[whit3rd]

On Tuesday, June 20, 2023 at 3:24:57 AM UTC-7, Martin Brown wrote:

On 19/06/2023 18:52, whit3rd wrote:
On Monday, June 19, 2023 at 2:08:12 AM UTC-7, Martin Brown wrote:
On 18/06/2023 22:01, whit3rd wrote:
On Sunday, June 18, 2023 at 1:25:15 PM UTC-7, Martin Brown wrote:
On 18/06/2023 03:18, whit3rd wrote:
On Saturday, June 17, 2023 at 11:26:09 AM UTC-7, Ricky wrote:
On Saturday, June 17, 2023 at 12:16:38 PM UTC-4, whit3rd wrote:

[about Accutron DNA watch]
The minute-hand \'tick\' is the slowest clock required for that watch (with minutes/hours
being gear-connected). A crystal with power-of-two division to minutes(rather than seconds)
would be expected; nearest to the common 32768 Hz would be 30720 Hz.
Digikey doesn\'t stock that particular value, though: it\'d be a custom rock.

The second hand is NOT stepping at 1 Hz, though; it\'s a synchronous electrostatic motor,
so there\'s no benefit to \'dividing down\' to one-per-second, rather one can just factor
any 2^N precursor of the one minute clock, and use a number of poles and
phases for the drive to accomplish the smooth sweep hand motion.
But there is an advantage to dividing down to once per minute.

Divide by 3 and 5 are not that hard - eg form first principles:

https://pages.mtu.edu/~suits/electronics/Divide_by_3&5_circuit.html

But they can divide down to 8Hz or 16Hz just as easily and that is more
than good enough to fool the human eye into seeing a continuous sweep
motion on the second hand. ISTR Rolex uses 8Hz on most of its watches.

The point here, though, is that lower frequencies than 32 kHz are used to step the
minute hand, while the 60:1 ratio that drives the slower hour hand is done by gearing,

Have you not noticed that there are also 60 seconds in a minute?

But, the seconds is represented by a circular sweep hand; as long as it moves
at a constant rate, the seconds are just marks on the dial, not electrical events.
It does one revolution in a minute, and a minute IS associated with an electrical event.

Snag is that I don\'t think there are any other dirt cheap low frequency
crystals (you might find 200kHz and 60kHz but they will require more
power than the humble 32k watch Xtal).

There is no need for a custom crystal for this. Unlike the situation for
a sidereal clock where making it run ~4 minutes a day faster than mean
solar time is way beyond the pull range of a crystal.

It\'s pretty easy to tune a tuning-fork crystal; you don\'t need submicron thickness control,
just adjust the length of the tines, and trim by electroplating mass on the ends...
Statek will accept orders for CX4V with the desired 30.720 kHz, no problem

<https://statek.com/wp-content/uploads/2019/08/CX4_10103_Rev-F.pdf?x37547>

.... but of course the price and minimum quantity are subject to negotiation..

 

[Martin Brown]

On 20/06/2023 23:00, whit3rd wrote:

On Tuesday, June 20, 2023 at 3:24:57 AM UTC-7, Martin Brown wrote:
On 19/06/2023 18:52, whit3rd wrote:
On Monday, June 19, 2023 at 2:08:12 AM UTC-7, Martin Brown wrote:
On 18/06/2023 22:01, whit3rd wrote:
On Sunday, June 18, 2023 at 1:25:15 PM UTC-7, Martin Brown wrote:
On 18/06/2023 03:18, whit3rd wrote:
On Saturday, June 17, 2023 at 11:26:09 AM UTC-7, Ricky wrote:
On Saturday, June 17, 2023 at 12:16:38 PM UTC-4, whit3rd wrote:

[about Accutron DNA watch]
The minute-hand \'tick\' is the slowest clock required for that watch (with minutes/hours
being gear-connected). A crystal with power-of-two division to minutes(rather than seconds)
would be expected; nearest to the common 32768 Hz would be 30720 Hz.
Digikey doesn\'t stock that particular value, though: it\'d be a custom rock.

The second hand is NOT stepping at 1 Hz, though; it\'s a synchronous electrostatic motor,
so there\'s no benefit to \'dividing down\' to one-per-second, rather one can just factor
any 2^N precursor of the one minute clock, and use a number of poles and
phases for the drive to accomplish the smooth sweep hand motion.
But there is an advantage to dividing down to once per minute.

Divide by 3 and 5 are not that hard - eg form first principles:

https://pages.mtu.edu/~suits/electronics/Divide_by_3&5_circuit.html

But they can divide down to 8Hz or 16Hz just as easily and that is more
than good enough to fool the human eye into seeing a continuous sweep
motion on the second hand. ISTR Rolex uses 8Hz on most of its watches.

The point here, though, is that lower frequencies than 32 kHz are used to step the
minute hand, while the 60:1 ratio that drives the slower hour hand is done by gearing,

Have you not noticed that there are also 60 seconds in a minute?

But, the seconds is represented by a circular sweep hand; as long as it moves
at a constant rate, the seconds are just marks on the dial, not electrical events.
It does one revolution in a minute, and a minute IS associated with an electrical event.

1 minute is 60 seconds and divide by 60 is very easy as many people have
been trying and failing to tell you. There is no need for a custom Xtal.

The only optimisation that might be present in a watch specific ripple
counter is that the intermediate outputs remain internal and unbuffered
apart from the 1Hz (and 8Hz ones for Rolex) which are actually used.

Typical CMOS chips have all of the ripple counter stages buffered and
available for external use on a pin.

Snag is that I don\'t think there are any other dirt cheap low frequency
crystals (you might find 200kHz and 60kHz but they will require more
power than the humble 32k watch Xtal).

There is no need for a custom crystal for this. Unlike the situation for
a sidereal clock where making it run ~4 minutes a day faster than mean
solar time is way beyond the pull range of a crystal.

It\'s pretty easy to tune a tuning-fork crystal; you don\'t need submicron thickness control,
just adjust the length of the tines, and trim by electroplating mass on the ends...
Statek will accept orders for CX4V with the desired 30.720 kHz, no problem

https://statek.com/wp-content/uploads/2019/08/CX4_10103_Rev-F.pdf?x37547

... but of course the price and minimum quantity are subject to negotiation.

You can do just about *anything* if cost is no object.
But what you propose adds extra cost for absolutely *no* benefit.

You are beginning to look like two short planks.

--
Martin Brown

 

[Martin Brown]

On 20/06/2023 14:00, upsidedown@downunder.com wrote:

On Tue, 20 Jun 2023 11:24:46 +0100, Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote:


Snag is that I don\'t think there are any other dirt cheap low frequency
crystals (you might find 200kHz and 60kHz but they will require more
power than the humble 32k watch Xtal).

With 60 or 240 kHz crystal it is easier to implement 1/10 s (or 1/100
s) resolution stop watch functionality.

How do you implement such resolution with a 32768 Hz crystal ?

Dither between two divisors and live with the jitter. No-one will *ever*
notice unless they subject the thing to laboratory testing.

Since human reaction times are ~>100ms even for the best trained visual
observers(*) the precision shown on such a toy chronometer is completely
bogus anyway unless a robot hits the button.

(*) 150ms to 250ms is more typical for your average Joe.

0.1s is divide by 3277 4x and 3278 1x = 32768
0.01s is divide by 327 8x and 328 17x = 32768

The 0.01s interval is still good to within 0.3% and 0.1s within 0.03%
worst case. (it is actually better than that on average for 0.01s)

--
Martin Brown

 

[server]

On Tue, 20 Jun 2023 08:08:32 -0700 (PDT), Ricky
<gnuarm.deletethisbit@gmail.com> wrote:

On Tuesday, June 20, 2023 at 10:10:38?AM UTC-4, Anthony William Sloman wrote:
On Tuesday, June 20, 2023 at 11:01:05?PM UTC+10, upsid...@downunder.com wrote:
On Tue, 20 Jun 2023 11:24:46 +0100, Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote:


Snag is that I don\'t think there are any other dirt cheap low frequency
crystals (you might find 200kHz and 60kHz but they will require more
power than the humble 32k watch Xtal).
With 60 or 240 kHz crystal it is easier to implement 1/10 s (or 1/100
s) resolution stop watch functionality.

How do you implement such resolution with a 32768 Hz crystal ?
For equally spaced 1 second steps you divide by 546 52 times and 547 8 times.

For 0.1 sec you can\'t do it exactly, but dividing by 3,277 nine times and 3,275 once comes pretty close.

For 0.01 sec you divide by 328 68 times and 327 32 times which is again fairly close.

If you want to get cranky you can set up a phase locked loop to lock a 100Hz oscillator driving a 10:1 digital divider to the roughly 0.1Hz output - the low pass filter in the phase locked loop will take out most of the jitter , such as it is.

Watches are often designed with very low power consumption in order to
maximize battery life time and hence avoid frequent battery changes.
A battery change can be expensive and may risk the water resistance of
the watch. A long life non-replaceable battery would be ideal, if it
lasted 10 years or more.

With a 60 kHz crystal, one could use a 4 stage ripple counter to
divide by 16 down to 3750 Hz followed by a 375 divider down to 10 Hz
to drive the 1/10 s display.

The power consumption would be only twice compared to a ripple counter
from 32768 down to 1 Hz due to the oscillator and first divider
running about twice the frequency.

NCO

If it is acceptable to charge the watch as often as a mobile phone,
why not. I have not checked for low power slow NCOs.

If power consumption is not an issue, I would make the NCO with a
small 8 bitter uController, clocked at 32768 Hz and interrupted with a
16 divider chain (62500 us). If a 1 us resolution time accumulator is
used, add 62500 to the time accumulator during each interrupt. There
are more than 2000 clock cycles between interrupts, which is more than
enough for doing four 8 bit additions with carry.

With a processor, you could handle leap seconds standard/daylight
time, time zone, weekday, date and leap year counting. These do not
have to be done in the interrupt context, just calculate them in
advance the previous minute and apply changes during the next full
minute.

Counting the date of Eastern is a bit more complicated and depends of
country .

 

[whit3rd]

On Wednesday, June 21, 2023 at 1:06:22 AM UTC-7, Martin Brown wrote:

On 20/06/2023 23:00, whit3rd wrote:
On Tuesday, June 20, 2023 at 3:24:57 AM UTC-7, Martin Brown wrote:

Snag is that I don\'t think there are any other dirt cheap low frequency
crystals (you might find 200kHz and 60kHz but they will require more
power than the humble 32k watch Xtal).

There is no need for a custom crystal for this. Unlike the situation for
a sidereal clock where making it run ~4 minutes a day faster than mean
solar time is way beyond the pull range of a crystal.

It\'s pretty easy to tune a tuning-fork crystal; you don\'t need submicron thickness control,
just adjust the length of the tines, and trim by electroplating mass on the ends...
Statek will accept orders for CX4V with the desired 30.720 kHz, no problem

https://statek.com/wp-content/uploads/2019/08/CX4_10103_Rev-F.pdf?x37547

... but of course the price and minimum quantity are subject to negotiation.

You can do just about *anything* if cost is no object.

But what you propose adds extra cost for absolutely *no* benefit.

And your proposal to use a crystal that divides by 2^N to make seconds instead of
one that divides by 2^M to make minutes has absolutely no benefit.

I\'m unaware of any reason a crystal manufacturer cannot make tuning forks for
any note in the normal range of their product line, so I do not see an eonomic
barrier. No one has offered a credible cost estimate.

> You are beginning to look like two short planks.

No one has asked Statek to give bids for both proposed crystals, for comparison,
so I question the assumption of cost.

 

[Anthony William Sloman]

On Wednesday, June 21, 2023 at 10:10:39 PM UTC+10, upsid...@downunder.com wrote:

On Tue, 20 Jun 2023 08:08:32 -0700 (PDT), Ricky
gnuarm.del...@gmail.com> wrote:

On Tuesday, June 20, 2023 at 10:10:38?AM UTC-4, Anthony William Sloman wrote:
On Tuesday, June 20, 2023 at 11:01:05?PM UTC+10, upsid...@downunder.com wrote:
On Tue, 20 Jun 2023 11:24:46 +0100, Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote:


Snag is that I don\'t think there are any other dirt cheap low frequency
crystals (you might find 200kHz and 60kHz but they will require more
power than the humble 32k watch Xtal).
With 60 or 240 kHz crystal it is easier to implement 1/10 s (or 1/100
s) resolution stop watch functionality.

How do you implement such resolution with a 32768 Hz crystal ?
For equally spaced 1 second steps you divide by 546 52 times and 547 8 times.

For 0.1 sec you can\'t do it exactly, but dividing by 3,277 nine times and 3,275 once comes pretty close.

For 0.01 sec you divide by 328 68 times and 327 32 times which is again fairly close.

If you want to get cranky you can set up a phase locked loop to lock a 100Hz oscillator driving a 10:1 digital divider to the roughly 0.1Hz output - the low pass filter in the phase locked loop will take out most of the jitter , such as it is.

Watches are often designed with very low power consumption in order to
maximize battery life time and hence avoid frequent battery changes.
A battery change can be expensive and may risk the water resistance of
the watch. A long life non-replaceable battery would be ideal, if it
lasted 10 years or more.

I had a Casio watch that lasted twenty years (until it went through the washing machine) going through three lithium coin cells in the process.

> With a 60 kHz crystal, one could use a 4 stage ripple counter to divide by 16 down to 3750 Hz followed by a 375 divider down to 10 Hz to drive the 1/10 s display.

If you could find a cheap close tolerance 60kHz crystal. The problem is that 32768Hz crystals are mass-produced in huge volumes and offer a much better cost-performance ratio.

> The power consumption would be only twice compared to a ripple counterfrom 32768 down to 1 Hz due to the oscillator and first dividerrunning about twice the frequency.

So half the battery life.

> If it is acceptable to charge the watch as often as a mobile phone, why not. I have not checked for low power slow NCOs.

If a mobile phone battery runs flat, the network resets it\'s clock when it gets recharged. That isn\'t an option with a watch.

> If power consumption is not an issue,

It obviously is an issue.

<snip>

--
Bill Sloman, Sydney

 

[Martin Brown]

On 21/06/2023 23:01, whit3rd wrote:

On Wednesday, June 21, 2023 at 1:06:22 AM UTC-7, Martin Brown wrote:
On 20/06/2023 23:00, whit3rd wrote:
On Tuesday, June 20, 2023 at 3:24:57 AM UTC-7, Martin Brown wrote:

Snag is that I don\'t think there are any other dirt cheap low frequency
crystals (you might find 200kHz and 60kHz but they will require more
power than the humble 32k watch Xtal).

There is no need for a custom crystal for this. Unlike the situation for
a sidereal clock where making it run ~4 minutes a day faster than mean
solar time is way beyond the pull range of a crystal.

It\'s pretty easy to tune a tuning-fork crystal; you don\'t need submicron thickness control,
just adjust the length of the tines, and trim by electroplating mass on the ends...
Statek will accept orders for CX4V with the desired 30.720 kHz, no problem

https://statek.com/wp-content/uploads/2019/08/CX4_10103_Rev-F.pdf?x37547

... but of course the price and minimum quantity are subject to negotiation.

You can do just about *anything* if cost is no object.

But what you propose adds extra cost for absolutely *no* benefit.

And your proposal to use a crystal that divides by 2^N to make seconds instead of
one that divides by 2^M to make minutes has absolutely no benefit.

Yes it does!

32kHz watch crystals are about an order of magnitude cheaper than any
custom low frequency crystal. There is one in every consumer item with a
low power RTC as well as in watches - they are made in billions!

They are about half the price of any other common LF crystals like 60kHz
and 65.536kHz for example. Dividing by 60 is *NOT* that difficult!

I\'m unaware of any reason a crystal manufacturer cannot make tuning forks for
any note in the normal range of their product line, so I do not see an eonomic
barrier. No one has offered a credible cost estimate.

You are beginning to look like two short planks.

No one has asked Statek to give bids for both proposed crystals, for comparison,
so I question the assumption of cost.

It is up to you to find out how much they would charge you and what the
minimum order is. ISTR when it was last done for sidereal rate it was
~$0.5 each minimum order 10k.

Today you can get a 32k watch xtal for ~$0.2 (and less in bulk)

The advent of cheap PICs like the 16877 pretty much did for that ancient
swap the crystal way of making a sidereal clock. That had enough legs to
direct drive an LCD display and very low power operating modes.

--
Martin Brown

 

[Piotr Wyderski]

whit3rd wrote:

> The mystery of a (2^15 - 1 ) Hz oscillator\'s purpose is open to other explanations: what\'s yours?

Add the 24.567 and 24.576MHz oscillators to your list.

Best regards, Piotr

 

[Piotr Wyderski]

Ricky wrote:

> Oh, BTW, you seem to ignore the need for an N bit state detector for whatever is making your sequence generator. That\'s even more logic.

If I understand him right (enough), then actually he doesn\'t.
\"Fractional state\" is the whole point of random counters, among others.
You can have a smaller state than required by the fully accurate
approach to be about right most of the time (statistically).

But I can see no advantages of these techniques in this particular
application: dividing by 2^N is so easy.

Best regards, Piotr

 

[Piotr Wyderski]

whit3rd wrote:

Capacitive energy loss thus favors a Gray code for counting
with minimal energy cost

That\'s not true, as pointed out by Rick, but how minimal is minimal?
RV3028 needs 48nA wile being significantly more complex than a simple
2^15 divider. What lower value would make you happy?

Best regards, Piotr