A Crystal Oscillator For A Stable Bench Reference

[Paul] likes a precise oscillator. His recent video shows a crystal oscillator with a “watch crystal” and a CMOS counter, the CD4060. Using such a circuit can produce very stable frequencies and since the 32.768 kHz crystal is a power of 2, you get nice divisions out of the counter.

We’ve seen the same trick done with decade counters (like the 4518B) to divide by 10 instead of powers of two to make frequency standards. A 1 MHz crystal can easily generate 100 kHz, 10 kHz, etc.

[Paul] mentions the clock is a Schmitt trigger input (he said output, but he meant input) and so it can take a noisy input from an oscillator with no problem.

The board is commercial, but the circuit is simple enough. The 4000-series CMOS is nice because you can use it over a wide voltage range and you can bias it in a linear range for things like simple crystal oscillators. Just be aware that if you do want to use a CMOS gate as a linear device you want to look for unbuffered versions (UB suffix). In 1972 the 4000A series ICs appeared and were all unbuffered. However, there were a number of flaws. The “B” series fixed some of these flaws at the expense of increasing propagation delay. by buffering. Since some designs rely on the unbuffered behavior, certain parts remain unbuffered and have a “UB” suffix.

If you want a refresher on Schmitt triggers, we have that. If you are wondering why these are called watch crystals, we can show you that, too.

20 thoughts on “A Crystal Oscillator For A Stable Bench Reference

    1. Completely! A 32.768KHz watch crystal has very poor temperature characteristics, which seems to be the opposite of a stable bench reference An AT cut crystal would have been a better choice and should be possible to hack into the existing circuit with a suitable divider.

  1. Hahaha! Darn right! He aVOIDed hacking… what bigger hack can there be?!? Maybe he should have used transistors instead of ICs? TUBES!!! And he has to MAKE them himself, from scrounged Coke bottles! LOL! It’ll be a good primer for somebody… that’s an important part of this site, too. And FInally, no part of this did he print! ???? Now, can he do it with a Grandfather Clock?

  2. I would not call it ‘super stable’. Stable enough for most things of course, but you can get quite check OCXO on eBay and AliExpress (take into account that they are removed from old electronics. In my experience 1/5 will be broken).

  3. I would have visited the article but it was a video. I don’t waste time on videos.

    HAD, please stop with the videos, as the only ‘article’. its a complete waste of our time, for so many of us. and the author comes out as lazy if there is ONLY a video and no text/photo write-up.

    I know, I know, we’re in the blogger age. but I’m sure if you insisted on at least having TEXT for any valid submission, that the bar would be raised a bit higher for all future HAD articles.

    today, you posted a video of someone who bought a board and used it. probably a low-point in HAD, although this is happening more and more.

    editors: what say you? do you really enjoy posting links to videos? you think that’s good enough?

    1. Obviously I can’t speak for anyone else on the HaD staff, but personally I don’t think ignoring anything without written documentation helps anyone. Rather, I think the posts we write up here on HaD serve as an alternative for those who don’t have the time to watch a 15 minute video, or can’t access it because they’re at work, on mobile, etc.

      1. I never watch the videos, and find them annoying. They contain very little of the technical information I’m interested in, and are a bad medium to convey it.

        I’d just as soon you wrote about other stuff that does have textual information available. The articles here tell you about something but generally don’t go into the details you’d want if you’re interested. The readers here are the type who like to know how things work.

      2. it could be my age, but I find ‘posts’ that are only video-based to be, well, lazy.

        I just can’t get enthusiastic about someone who just does a video and thinks that’s good enough.

        for so many of us, its not.

  4. Tellya what, say what you will about this particular article, but lab timing is an interesting subject. Time-nuts can easily spend silly sums achieving dubiously-relevant levels of precision, but I think there are quite a few of us who have a handful of instruments that would benefit from a modest level of attention to frequency references.

    I’d love to see a primer (since I’m not yet well-enough-versed to write it myself) on how to do instrumentation timing on a budget. Assume we have a lot of instruments with REF IN and a few with REF OUT connections, can they just be daisy chained? What about splitters and tees? How do you decide which one gets to be the upstream source? At what point does a distribution amp become necessary?

    If someone wants to write this, I’d love to read it.

  5. Probably depends where you source those 32.768kHz crystals, but my experience with them are not very good. Clocks with DS1307 and external crystal need frequent adjustments because they can drift a minute in 30 days period. But DS2321 chips that have integrated crystal that is temperature compensated will do at least a year without need for adjustment.

    1. By the look of it, the little round tin with 2 tiny leads, it’s the same one every cheap digital watch uses. So “precision” is measured in minutes per month. It’s not compensated at all. This PCB is no sort of reference, I’ve no idea what it’s supposed to be for, really, is there a need for dirt-cheap inaccurate timing references?

      1. watch crystals have one huge benefit when put into a wrist watch… thanks to the power of the human the temperature is pretty constant. In the winter the X-tal is warmed by the arm and in the summer it is kept out of the sun. The body regulates itself at 37deg C so the wrist will be a few deg. less then that, brilliant. No OCXO required thanks to high tech temp regulation of the human body.

        But when you use such an X-tal everywhere else… perhaps even in a crappy design (X-tal traces to high impedance clock circuitry picking up noise, slightly off load capacitance and large temperature ranges between night and day. Well it’s no surprise that it’s of by a few seconds a month.
        But honestly, even a crappy X-tal setup is much better then a crappy RC-circuit. The circuit shown in this post is intended for people without proper tools, those people most likely don’t care OR lack funds or knowledge to improve. People who do care, do have fund or do have the knowledge… well this post obviously isn’t for them.

        I do agree about the “hack” label not being correct… but sometimes rules are meant to be broken. If for some reason this happens much more often hackaday would be loosing it’s appeal, yet I’m not worried as we see so many beautiful or ingenious projects that you’ve got to realize that it isn’t easy finding them, so sometimes a low-tech article must be posted to keep up with the posting schedule. Anyway, hackaday, please carry on, you’re doing fine.

Leave a Reply to NightCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.