How long do quartz crystals for watches usually last?

[Wade]

The death rate seems (by my unscientific analysis) high in old LED watches. Are the crystals dead because of age "just burned out", or because of shock. I can empathize if they are burned out due to age. (':-D')

Do the higher frequency Hughes crystals last longer than the usual 32768hz ones?

[retroleds]

The old crystals were an actual mined, quartz....full of micro-flaws and fairly expensive. Sometimes there were even shortages of them
http://www.ledwatches.net/articles/PopEle/PELCD7.jpg

Modern crystals are lab-grown... formed by chemical deposit on the contacts within. Sort of like cubic zirconium "diamonds" - very few flaws. So the hope is they are more stable and less susceptible to natural flaws making them go bad. The modern crystals like the old ones go through a burning in period where their frequency changes [hopefully] just a little. If you are look ing at them in a catalogue you will see specs. for aging rates. Lower is better.

The Hughes do seem to suffer substantially fewer failures of the crystal. IMO.

[bruce wegmann]

Theoretically, they could last for a very long time, indeed. But, they do have vulnerabilities. They can be destroyed by being subjected to voltage spikes or severe mechanical shock. And, there is one other failure mechanism not often considered; air leakage. All oscillator crystals are hermetically sealed under vacuum. The presence of air in the housing would tend to damp out the crystal's vibrations, so a good vacuum is established, and the can is sealed. Corrosion can, of course, attack the seal at any point, and, once breached, the can slowly fills with air, and the crystal stops oscillating. Early frequency-control crystals were natural quartz, selected for purity and optimum piezoelectric qualities. Now, it is all furnace-grown, semiconductor-grade purity, laser-trimmed material that is many times as stable as the best available 35 years ago. A characteristic of quartz is the tendency for the frequency to drift slowly over long periods of time [months or years]. Initally, the rate of drift is relatively high, but diminishes gradually. Eventually, it "settles down" and maintains a stable frequency for the remainder of its' life [here we are disregarding temporary shifts due to temperature changes]. This is why Time Computer recommended checking [and adjusting if necessary] the quartz every time the batteries were changed. Motorola produced the first 32KHz crystals for Time Computer in 1972. They were good by the standards of the 70s, but still required one, then two, trimmer capacitors to compensate for "aging" frequency drift. Manufacturing technology progressed rapidly, and the crystals in the last Pulsar models are about a sixth the size of the ones in the first, with better stability and less drift [as evidenced by the fact they went back to a single trimmer in the P4s and after; I have seen at least four different crystals in completely original P3 modules...there may be others]. In some cheaper modern quartz watches, the trimmer is eliminated, as the frequency can be held to very close tolerance by laser-trimming methods.

[Chris S.]

On the subject of crystals. Where is the best place to buy them from for the repair of old LED watches. I think I've found the right type at Farnell but would appreciate some confirmation.

Order Codes (at Farnell)

1611825
1611826
1216227

The first two are 12.5pF the third is 6pF and is relatively expensive.

[Led-Time]

The 32768Hz crystals from Maplins work great.

http://www.maplin.co.uk/Module.aspx?ModuleNo=480

[Chris S.]

Thanks Klippie. Just ordered some.

[rewolf]

On the subject of crystals. Where is the best place to buy them from for the repair of old LED watches. I think I've found the right type at Farnell but would appreciate some confirmation.

Order Codes (at Farnell)

1611825
1611826
1216227

The first two are 12.5pF the third is 6pF and is relatively expensive.

12.5pF is the correct load capacitance.
6pF is uncommon and thus more expensive.

[bruce wegmann]

Clearly, these will work, but the thing that bothers me is...the frequency tolerance [stated here as + or - 15ppm...parts per million]. That sounds pretty good, but it's actually awful! Considering there are slightly more than 31.5 million seconds in a year, thats allowing an error of as much as 470 seconds a year, or 39 seconds a month...hardly the 5 seconds a month guaranteed by Time Computer [60 seconds a year is just under 2ppm]. It may be difficult to regulate these down to this level of accuracy...but I would say a watch in any sort of working condition is better than one that is not. You'll definitely need a quartz timing monitor to have any chance of getting good timekeeping with these.

[rewolf]

The standard tolerance for quartz crystals is +/-20 ppm (in scientific notation: +/-20e-6). That is 24 x 3600 x 20e-6 = 1.7seconds per day, or almost one minute per month.
Note that this is INITIAL tolerance at STANDARD temperature (normally 25°C). That is, add to this the aging and temperature drift.

You can get 32kHz crystals rated at 10ppm, but they are difficult to get and expensive. I've never seen a 32kHz crystal rated better than 10ppm. So you definitely need some sort of trimming.

But you can simply be lucky: my Ventura v-tec Kappa has been running for 18 months without adjustment and is only 2 sec off - whereas my Ventura d-sparc px has 2 sec deviation per WEEK.

[Led-Time]

You can get 32kHz crystals rated at 10ppm, but they are difficult to get and expensive.

Here's some... CLICK...

[Chris S.]

Doh!!

Cheap as Chips. Had to get some of these as well at that price!

Thanks again Klippie!!!!