Why is the 32 kHz crystal oscillator used in RTC, instead of the 16 kHz?

MrChips

Joined Oct 2, 2009
34,629
That's a good question.
32768Hz can be easily divided by a 15-bit counter.
So why not use a 16384Hz crystal and divide by 14-bit counter?

What the heck? Why not go all the way and use a 1Hz crystal so that no division is required?

Ask the crystal manufacturer.
 

AnalogKid

Joined Aug 1, 2013
12,047
A 32 kHz "crystal" actually is a type of tuning fork. This is used because at these low frequencies it is much smaller than a standard AT cut crystal would be. Still, size is everything inside a wrist watch. 32 kHz was a tradeoff between size (higher freq = smaller) and circuit complexity (higher freq = more divider stages).

It has nothing to do with accuracy. A 1% error in a 10 MHz oscillator, a 10 kHz oscillator, and a 10 Hz oscillator will produce the exact same 1% error percentage when they are divided down to 1 Hz.

ak
 

GopherT

Joined Nov 23, 2012
8,009
That's a good question.
32768Hz can be easily divided by a 15-bit counter.
So why not use a 16384Hz crystal and divide by 14-bit counter?

What the heck? Why not go all the way and use a 1Hz crystal so that no division is required?

Ask the crystal manufacturer.

He was not asking, he was telling you - he already had the link and the answer - he was asking you to read his link.
 

MrChips

Joined Oct 2, 2009
34,629
Sorry -- never heard of that one. Was it used in television sets?
Back in the '70s it was a simple way to get an accurate 60Hz time-base signal for digital clocks using a readily available crystal, the color-burst crystal. I built a communications receiver frequency counter using this part.
 

Papabravo

Joined Feb 24, 2006
22,058
Back in the '70s it was a simple way to get an accurate 60Hz time-base signal for digital clocks using a readily available crystal, the color-burst crystal. I built a communications receiver frequency counter using this part.
Oh I see, it was a chip used in conjunction with the color burst crystal. I never knew that.
 

DickCappels

Joined Aug 21, 2008
10,661
Edit: The numbers below are incorrect.

The 3.9545 MHz crystal could be easily divided by a binary divider to obtain nearly 60 Hz (60.6606 Hz).
 
Last edited:

Papabravo

Joined Feb 24, 2006
22,058
The 3.9545 MHz crystal could be easily divided by a binary divider to obtain nearly 60 Hz (60.6606 Hz).
The 3.9545[Correction 3.579545] MHz crystal could be easily divided by a binary divider to obtain nearly 60 Hz (60.6606 Hz).

\(3.579545 \times 10^6 / 59659 \;=\;60.00008381\)
\(3.579545 \times 10^6 / 59660 \;=\;59.99907811\)

Factor(59660) = 2 x 2 x 5 x 19 x 157 and 59659 is prime. What divisor did you have in mind?
the closes divisors to 60.6606 Hz are 59009 and 59010.
Factor(59010) = 2 x 3 x 5 x 7 x 281 and 59009 is prime
 
Last edited:

AnalogKid

Joined Aug 1, 2013
12,047
The output frequency with a perfect crystal is 60.00008381 Hz. This translates into a steadily accumulating error of 7.241 seconds per day (fast), so the circuit is not nearly as accurate as a crystal oscillator based on something that is a decade or binary multiple of 60 Hz.

ak
 
Top