Go Down

Topic: Why are real-time clocks always 32.768khz? (Read 8914 times)previous topic - next topic

lightaiyee

Apr 12, 2013, 05:02 pm
I notice that real-time clocks are always 32.768khz. Does anyone know whether there is a particular reason for this? is it just coincidence? Just curious.

Thanks.

DirtBiker

#1
Apr 12, 2013, 05:06 pm
It is easy to divide that frequency using a binary counter to get 1 second pulses.

32768 / 214  = 1
Dirt Biker

Magician

#2
Apr 12, 2013, 06:07 pm
Quote
32768 / 2^14  = 1
Clock "adjustment":  32768 / 2^15 = 1

DirtBiker

#3
Apr 12, 2013, 06:21 pm

Clock "adjustment":  32768 / 2^15 = 1

Right you are!
Dirt Biker

fungus

#4
Apr 12, 2013, 06:39 pm

Also, low frequency crystals require significantly less power to oscilate, which is a handy thing when operating small battery-powered devices such as RTC  chips or wristwatches.

So...why not 16384?
No, I don't answer questions sent in private messages (but I do accept thank-you notes...)

bobthebanana

#5
Apr 12, 2013, 07:04 pm
I once heard somewhere that a specific natural occuring shape of crystal would oscillate at exactly 32.768 khz. Don't know if this holds any truth though

BillO

#6
Apr 12, 2013, 07:08 pm
Or even 8,192?

Well, I think the answer is that even the 32768Hz is a pretty delicate and susceptible crystal as it is.  For 16384, they'd have to make them twice as long or half as thick and this would make them weaker (will tend to age quicker) and more susceptible to interference.  As it is, for maximum stability the 32768Hz crystals need special guard planes and such to keep out EMI.  Although lower frequency crystals are available, their stability/susceptibility characteristics don't make them ideal for clocks.  32.768kHz seems to be the sweet spot.
Facts just don't care if you ignore them.

BillO

#7
Apr 12, 2013, 07:14 pm

I once heard somewhere that a specific natural occuring shape of crystal would oscillate at exactly 32.768 khz. Don't know if this holds any truth though

Clock crystals are usually artificially grown and cut in a tuning fork shape.  Tuning forks give pretty pure signals with less dominant overtones near the fundamental frequency.  I think their 2nd overtone is about 6 times fo.  This makes it easy to filter out without diminishing the fundamental too much.
Facts just don't care if you ignore them.

dc42

#8
Apr 12, 2013, 07:52 pm
The 32768Hz frequency was chosen many years ago, in the early days of digital watches. The need was for a crystal that was small and tough enough to use in a watch, but the IC used to divide it down to 1 second pulses had to draw very little current. So the frequency chosen was a power of 2 to keep the divider IC as simple as possible, and 32768Hz was a good compromise. Lower would have required a larger, more fragile crystal. Higher would have increased the current consumption of the divider chip (current consumption of CMOS chips mostly scales with frequency). These days, chips take less current, and I believe modern watches use smaller, higher frequency crystals. But clocks and RTCs typically still use 32768Hz.
Formal verification of safety-critical software, software development, and electronic design and prototyping. See http://www.eschertech.com. Please do not ask for unpaid help via PM, use the forum.

afremont

#9
Apr 12, 2013, 10:50 pm
First quartz controlled wrist watch at 8192Hz in 1969:
http://en.wikipedia.org/wiki/Astron_%28wristwatch%29
Analog dial and hands.

The Bulova Accutron (ca. 1960) watches used an actual mechanical tuning fork that oscillated at 360Hz.

Experience, it's what you get when you were expecting something else.

BillO

#10
Apr 12, 2013, 11:08 pm

First quartz controlled wrist watch at 8192Hz in 1969:

And cost as much as a Toyota Corolla at the time!   Yikes!
Facts just don't care if you ignore them.

sonnyyu

#11
Apr 13, 2013, 01:55 am

First quartz controlled wrist watch at 8192Hz in 1969:

And cost as much as a Toyota Corolla at the time!   Yikes!

not there yet, but close.

First Generation (1968-1969) Toyota Corolla starting price is \$1,700 USD, we have \$550 USD gaps.
"cost as much as a Toyota Corolla engine plus 4 wheels" might be better tried.

sonnyyu

#12
Apr 13, 2013, 02:41 am
Price table of world's first commercially electronics product

Seiko Wristwatch             35SQ         \$1,250       1969
JVC  VHS VCR                   HR-3300    \$1,400       1976
Sony  Compact Disc player  CDP-101   \$730          1982
Matsushita DVD player               \$700 to \$1,200   2000
Samsung  Blu-Ray player   BD-P1000   \$999.99      2006

There is magic number \$1,000 which come from consumer market research.

BillO

#13
Apr 13, 2013, 03:03 am

... is \$1,700 USD, we have \$550 USD gaps.

Well, \$450, but close enough.

The 1969 watch was more than the 1966 car.  According to Wikipedia: "The price of Toyota Corolla E10 was 432,000 yen (US\$1,200) when it was released in 1966"
Facts just don't care if you ignore them.

Jack Christensen

#14
Apr 13, 2013, 03:19 am

First quartz controlled wrist watch at 8192Hz in 1969:
http://en.wikipedia.org/wiki/Astron_%28wristwatch%29
Analog dial and hands.

Quote from: Wikipedia
The Astron was accurate to ±5 seconds per month, or one minute per year.

2ppm, better than the average 32.768kHz crystal-controlled RTC

Go Up

Please enter a valid email to subscribe

Confirm your email address

We need to confirm your email address.
To complete the subscription, please click the link in the email we just sent you.

Thank you for subscribing!

Arduino
via Egeo 16
Torino, 10131
Italy