It is currently 18 Oct 2025, 23:25


Crystal frequency in LED stopwatch watches?

For electronic related stuff like module repair, silver epoxy fixes etc.
  • Author
  • Message
Offline
User avatar

Old Tom

Wizard

Wizard

  • Posts: 274
  • Joined: 15 Dec 2006, 10:48
  • Location: London, England

Crystal frequency in LED stopwatch watches?

Post11 Jan 2010, 11:45

Here's a thing that has bothered me for years- has anyone ever measured or knows for sure the crystal frequency in an LED stopwatch (say a Frontier stopwatch module or one of the others)? The reason is that I cannot see how a 1/100 of a second count can be derived from a 32768Hz crystal ( or a Hughes 786K crystal) by conventional (or forced) divider chains. So, did these modules use an odd (for a watch) crystal to give the 1/100s (or even a 1/10s count) count or is there a "trick" in the divider?
Offline
User avatar

charger105

Guru

Guru

  • Posts: 867
  • Joined: 02 Mar 2007, 02:11
  • Location: Australia

: Crystal frequency in LED stopwatch watches?

Post11 Jan 2010, 14:05

Hi Tom.
I can comment on the MkIV Synchronar which has a stopwatch. It definately uses a 32k crystal. No idea how it's done though.

Rgds,
Andrew.
Offline
User avatar

rewolf

Guru

Guru

  • Posts: 1863
  • Joined: 11 Jul 2004, 15:32
  • Location: Ravensburg, Southern Germany

: Crystal frequency in LED stopwatch watches?

Post11 Jan 2010, 20:06

They use a 32768Hz crystal like other watches. At least all modern watches do so.
The trick is in the divider, resp. the stopwatch counter: the 1/100 seconds are NOT exact, some are a bit too short, the rest too long, so that the total over 100 counts is exactly 1 second.

E.g. the EPSON S1C microcontroller has a special stopwatch timer peripheral exactly for this purpose.
This timer takes a 256Hz (=32768/128) input clock and divides it either by 2 or 3 to generate approximately 10ms intervals, so the 10ms intervals are exactly either 7.8125ms or 11.71875ms.
The divider pattern for 10 successive intervals (to get 1/10sec) is either:
a) 3-2-3-2-3-2-3-2-3-2 (=25/256sec=97.65625ms), or
b) 3-3-3-2-3-2-3-2-3-2 (=26/256sec=101.5625ms),
that is, the 1/10sec (100ms) also aren't exact.
The counter mixes these two patterns using a "b-b-a-a-b-b-a-a-b-b" scheme, resulting in a total of 256/256sec.

I'm sure all stopwatches use a similar scheme to generate the 1/100s intervals, maybe with a higher frequency basis clock, resulting in smaller deviations, but the principle is always the same.

Below is a copy of the respective page in the EPSON S1C Technical Manual that makes it a bit clearer (maybe...):
Image

Here's a patent that describes in more detail how a 1/10s counter can be implemented: http://www.freepatentsonline.com/3975898.html
Offline
User avatar

clockace

Techno Mage

Techno Mage

  • Posts: 544
  • Joined: 25 Feb 2008, 03:28
  • Location: Arizona

: Crystal frequency in LED stopwatch watches?

Post12 Jan 2010, 02:43

awesome rewolf. i knew it was a matter of division, usually by 2, but couldn't get my head around the 1/100 and 1/10. even on a mechanical stopwatch it is a matter of division of the beat frequency of the balance wheel but starting and stopping the stopwatch itself introduces an error that cannot be accurately read on the dial. peter
Offline
User avatar

retroleds

Guru

Guru

  • Posts: 3634
  • Joined: 04 Feb 2006, 10:34
  • Location: Surrounded by hicks and sticks (farms and woods) - Michigan,USA

: Crystal frequency in LED stopwatch watches?

Post12 Jan 2010, 17:01

I'd agree that the 32768 is the correct crystal.

Yeah, it would take a really complicated circuit to get it very close to a true 1/100 of a second, considering the operation of the divider is binary. Whereas keeping count of 60 seconds making a minute, once the signal is dived down to 1, is about the simplest programming function there is. For those unfamiliar, binary is "base 2", counting along as: 0,1,2,4,8,16,32,64,128....32,768 and beyond.
http://www.retroleds.com - Sales of vintage LED, LCD, analog watches, parts and gadgets - repair tutorials & tips
Nov. 2022 - back in business!! BItter divorce is in home stretch, come grabs some great deals, I had to open the safe . . . damn attorneys. piss.
Offline
User avatar

Old Tom

Wizard

Wizard

  • Posts: 274
  • Joined: 15 Dec 2006, 10:48
  • Location: London, England

: Crystal frequency in LED stopwatch watches?

Post15 Jan 2010, 11:22

Well, thank you for all that- I have about four stopwatch modules, none of which work and all of which have been "repaired" by folks unknown.

Anyway, I went to work on one, a Hudson Harris Quantum (Elonex module), and after changing the crystal, cleaning the ceramic board etc it runs again although a little unreliably (the small SM resistor looks very high in value, approx 16 meg) so that might get changed.

A little more reading and poking around in other old watches with 1/10 & 1/00 displays seems to indicate they are using the 1024hz output from the main divider chain to another chain which counts to 10 or 100 and resets, normal 1hz output used to reset this part of the chain every second so max error 24/1024 second which is pretty much un-noticeable for all intents and purposes.

I think all LED and quite a few LCD stopwatches are pre-microprocessor electronics (at watch size/power levels) so this sort of scheme is not what we are looking at. I do have NS timekeeping databook from 1976 which lists all their LED ic's but doesn't give any real indication of how they actually function internally apart from rather general block diagrams of function.
Offline
User avatar

rewolf

Guru

Guru

  • Posts: 1863
  • Joined: 11 Jul 2004, 15:32
  • Location: Ravensburg, Southern Germany

Re: : Crystal frequency in LED stopwatch watches?

Post15 Jan 2010, 12:44

I'm sure none of the early LCD and LED watches is microprocessor-based. That would have been absolute overkill at that time.
The electronics are made of nothing more complex than logic gates and flipflops.
But even with these basic components it is VERY easy to make an 100Hz clock out of 32768Hz in the way described above for the EPSON S1C. The S1C's timer circuit is nothing more than a few flipflops and gates.
Many 32768Hz-based real time clock (RTC) chips also have a 1/100sec counter. They use different schemes, some a 128/100 divider, some a 512/100, maybe others use 1024/100.

Return to Electronical

Who is online

Users browsing this forum: No registered users and 39 guests

cron