2
\$\begingroup\$

There seem to be many ways to take a 32.768kHz signal and turn it into 1Hz.

  • I can use a CD4060, but still have to add a flip-flop... so 2 "large" chips with excess (potentially) unused functions there. Sure the other 4060 outputs could come in handy, and I might find a use for the other flip-flop on board, but still a lot of real estate for a pile of maybe with a built-in oscillator likely as accurate as a 555 - negated by a crystal anyhow.

  • The "Swiss Army Knife of dividers" - I can use the programmable CD4536, which provides a single output from any stage from 1-24. But again, with chip size and "wiring" for programming - a lot of real estate for one function - and again with the on board oscillator.

  • The picture of simplicity, Nexperia makes a small 5-pin 14-stage divider. Oh so close. Still need to add a flip-flop, but at least the '4214 is not a 16 lead divider like the '4060.

  • Just for fun, I could cascade 8 CD4013s.

There are other counter/dividers that fall very short or way beyond 15 stages as an option. It seems that 15 stages of binary division is not nearly as important as I would suspect.

So my question is: With the 32.768 kHz Crystal being so prolific, and the (I would think, common) need to yield "seconds" with clocks and timers, why is there no discrete solution for doing so? Am I missing something?

P.S. I know that microcontrollers, GPS RX, RTCs, and other modern devices can provide all the timekeeping function one might need, but with all of the other "archaic" technology still available, I find it hard to believe such a device doesn't/didn't exist.

Addendum I seem to have created some confusion with my example, where I was personally looking for a 1Hz signal from what I thought was a "jellybean" frequency. But, it's about the stage count availability, not the frequency yielded. My example is my discovery of a lack of devices with a specific intermediate division capability - regardless of the input frequency - where the output of \$2^{15}\$ (and 16th and 17th) seems to be almost avoided, and that with the exception of one example, an additional device is required to achieve that number.

I already had my 1Hz solution before this question, but got curious and sought out what I thought was the "old way" of achieving this frequency in small, cheap timing products. So, my search was for a 15th stage readily available within a single device. Maybe the 32.768k wasn't part of the old way at all. Regardless, it seemed odd that all of the available preset counters/dividers I found that provide select ranges between \$2^0\$ and \$2^{24}\$ seemed to beat around that \$2^{15}\$ bush.

\$\endgroup\$
  • 1
    \$\begingroup\$ because now most users of medium scale ICs don't care about size, neither do their makers. \$\endgroup\$ – user3528438 May 9 '17 at 20:43
  • 1
    \$\begingroup\$ A tiny CPLD will do it for under a quid. Specialist ICs for small logic functions were replaced by programmable logic. Until then, people were quite accepting of 74HCT4060 and the other options. I don't recall people being as exasperated and frustrated as your excessive bolds and italics sound. (Over-emphasis reads rage-y.) \$\endgroup\$ – TonyM May 9 '17 at 20:59
  • 2
    \$\begingroup\$ Also, I'm not sure why you think there's any great demand for 1-Hz square waves anyway. Clocks and wristwatches have the divider built into dedicated chips, as do RTC chips for computers. What sort of application needs a 1-Hz square wave that doesn't also have a microcontroller of some sort in it? \$\endgroup\$ – Dave Tweed May 9 '17 at 21:08
  • 1
    \$\begingroup\$ @MarcusMüller: Not all that tragic. It isn't at all hard to deal with the 1000/1001 factor in the counter chain. \$\endgroup\$ – Dave Tweed May 9 '17 at 21:14
  • 1
    \$\begingroup\$ Can't be sarcastic to history, old boy. History presents you with a record of what the world of electronics felt it needed and could sell. And it didn't 32.768-to-1 Hz dividers in a dedicated package :-) Out of interest, can you give 5 example applications for it? Not just be 5 uses for counting 1 second, though. \$\endgroup\$ – TonyM May 9 '17 at 21:42
2
\$\begingroup\$

The reason is that that's not all a counter needs to do.

Since any timer or counter will need to provide BCD counters for the seconds (and minutes and hours, if appropriate), along with display formatting and driving, it makes far more sense for the custom IC to incorporate both the oscillator and the divider chain into the IC, rather than having this as a second, external IC.

What you seem to be asking is why nobody makes a 6-pin IC which takes power, ground, 2 pins for the crystal, and outputs a 1 Hz signal. Well, that's because there isn't much demand for it.

If you want 1 Hz, a much smaller solution is this ASTMK part, which provides 1 Hz logic-level output in a 2 mm x 1 mm package, which is smaller than a standard 32,768 crystal, which typically will run about 4 mm x 8 mm.

\$\endgroup\$
  • \$\begingroup\$ Thanks for the suggestion. It's perfect. I actually already had that oscillator in my design, but I don't think I can solder it. My edit might clear up the confusion I caused making my point though. \$\endgroup\$ – Jay May 10 '17 at 15:08
  • \$\begingroup\$ You could use a CD4521 with a 4.194304 MHz crystal. The Q22 output is 1Hz. \$\endgroup\$ – henros Dec 19 '18 at 11:36
2
\$\begingroup\$

with the 32.768 kHz Crystal being so prolific, and the (I would think, common) need to yield "seconds" with clocks and timers, why is there no discrete solution for doing so?

Is that need really all that common? aside from consumer analog wall clocks, which tend to use cheaper things than DIP packaged ICs, I see none right now – anything that does more than move a seconds hand will simply have a minimal MCU (microwave oven), usually, or even some ASIC (watches, for example).

The use cases for something that only divides a slow clock by \$2^{16}\$ kind of seem sparse.

\$\endgroup\$
  • \$\begingroup\$ A look at the supplier I tend to use shows 96 different frequencies of crystal available, and 32.768kHz is the 5th most common, accounting for 80 of the 1462 different parts they list. So yes, it probably justifies the term "common". \$\endgroup\$ – Finbarr May 9 '17 at 21:19
  • \$\begingroup\$ Oh, might have put that ambigously, @Finbarr: Not the 32.768kHz oscillator isn't common – the need for a 1 Hz signal is! \$\endgroup\$ – Marcus Müller May 9 '17 at 21:24
  • 1
    \$\begingroup\$ Fair enough! Even so, I was surprised just how many different parts were listed with the same frequency... \$\endgroup\$ – Finbarr May 9 '17 at 21:26
  • \$\begingroup\$ And as if by magic... Someone else has just asked a question about the NCP PCF8563 real time clock chip - which takes a 32.768kHz crystal and contains oscillator, divider, real time clock/calendar, alarms and I2C interface in an 8 pin chip. And you can even program an output pin to generate a 1Hz signal! \$\endgroup\$ – Finbarr May 10 '17 at 8:47
  • \$\begingroup\$ I don't know how common that need would be. It just seemed so, if not currently, at some point. But see my addendum regarding the division factors over the resulting frequency. And @Finbarr. LOL. Learning an RTC will come after I have become comfortable with the logic process and then maybe CPLDs. I know nothing about I2C. \$\endgroup\$ – Jay May 10 '17 at 15:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.