[time-nuts] Re: Crystal oscillator for a begginer
Bob kb8tq
kb8tq at n1k.org
Fri Jan 7 13:07:11 UTC 2022
Hi
Setting up a bench to measure clock sources can indeed be expensive.
eBay is one of the ways to find the gear you need at a semi-rational price.
The gotcha there is that the gear may or may not be fully functional. Stuff
from folks like TAPR might cost a bit more. The hassle level often makes
this worth every penny â¦.
The most basic thing you need for your bench is some sort of standard
to compare things to. It needs to be more accurate / stable than the devices
you are testing. A GPSDO is often the answer. Indeed folks do go to somewhat
insane lengths coming up with standards.
How complicated this gets is very much up to you. A $60 counter and an
$80 GPSDO might well do the job. You could spend 10X that and still
not be into the âcrazyâ level of bench equipment.
Bob
> On Jan 6, 2022, at 12:50 PM, Adam Space <time.isanapp at gmail.com> wrote:
>
>> How would you quantify "pretty weak?" You could connect up the components
>> correctly, but you could not design a circuit from scratch? Or you are
>> worried you will let the smoke out the first project you start?
>
> I just mentioned this to indicate that I have only basic knowledge, as
> compared to many on this list, I assume. I'm sure I could understand most
> concepts if explained to me, and can connect things together. But I've seen
> videos where they talk about resistors, circuits, etc., and my knowledge of
> these things is limited to my physics class several years ago (not very
> good). Things like soldering and PCB are things I've heard of, but they're
> still very new to me and it would be a big step up to do those things on my
> own.
>
> Regarding your other explanations, thank you, this cleared a lot of things
> up. To summarize, an oscillator will generate electrical signals at a given
> frequency, but a frequency counter and something to subdivide the frequency
> down to e.g. one second would be needed to get closer to a proper clock.
> (And finally of course something to display the count). It seems like the
> oscillator is by far the cheapest part of this: they can cost as little as
> under a dollar to more for better ones. But it seems like frequency
> counters and subdividers are much more expensive, no? Someone linked a
> subdivider to 1pps, which looks like a good place to start.
>
> Adam
>
> On Wed, Jan 5, 2022 at 10:33 PM Chris Caudle <6807.chris at pop.powweb.com>
> wrote:
>
>> On Wed, January 5, 2022 5:01 pm, Adam Space wrote:
>>> My knowledge of hardware is pretty weak
>>
>> How would you quantify "pretty weak?" You could connect up the components
>> correctly, but you could not design a circuit from scratch? Or you are
>> worried you will let the smoke out the first project you start?
>>
>>> But I am confused on how I could exactly use this [crystal oscillator]
>>
>> A "canned" crystal oscillator is a quartz crystal and circuitry together
>> in a package that provides an oscillating electrical output when connected
>> to power.
>> For the specific device you mentioned, you connect it to a 5V power
>> supply, and the output will vary between 0.5V and 4.5V 10 million times
>> per second.
>>
>>> if I have a crystal oscillator going, how could I compare it or adjust
>> it to
>>> other clocks on my network, or to UTC for example?
>>
>> First, a crystal oscillator has no concept of time, so discussion of
>> adjusting to UTC does not make sense in that context. The output swings
>> low to high to low and that is it.
>>
>> If you want a clock, you need something which oscillates (quartz
>> oscillator, atomic states of cesium atoms, pendulum, etc.) and you need a
>> mechanism to count and display the oscillations.
>> So to make a clock with your 10MHz quartz oscillator you would need an
>> electronic circuit compatible with the electrical levels of that
>> oscillator, which includes a counter that can increment at that rate, and
>> probably circuitry to divide down the 10 million cycles per second to more
>> usable rates (1 cycle per second, 1 cycle per hour, etc.). Unless you are
>> satisfied just knowing that the circuit is busy counting cycles, you
>> probably also want a way to either display the current count, or output
>> the count in some electrical format for use elsewhere.
>> Unless your counter is completely self referenced, you will probably also
>> want a way to set the count to a predetermined count right after it powers
>> on so that it matches your other clocks.
>>
>>> Or to take another example, suppose I have a few crystal oscillators like
>>> the one above. Is there a way I could compare them to each other, or log
>>> the offsets from each other, and so on?
>>
>> Yes, and depending on how quickly you need to know, how precisely you need
>> to know, and over what time periods you need to know (or alternately what
>> frequency offset if viewing in frequency domain rather than time domain)
>> there are quite a few ways.
>> The most basic is to use a frequency counter instrument with a time base
>> much better than the device you are trying to measure and count how many
>> cycles occur over a particular time period. For example, you could could
>> the transitions on the output of your oscillator for 1 second, and display
>> the count. You expect it to be around 10 million, but oscillators vary in
>> accuracy as built, and drift with time, temperature, power supply
>> variations, etc. Unless your counter can generate the start and stop
>> counting signals very accurately 1 second apart, you can't be sure how
>> much of the difference between the actual count and 10 million is due to
>> the device you are testing, and how much is due to inaccuracy in the
>> measuring device.
>> For the oscillator you mentioned the problem is not very difficult, most
>> good quality test equipment has much better accuracy oscillators
>> internally, but as you attempt to measure higher and higher quality
>> oscillators it becomes more challenging making the test equipment good
>> enough to be sure you are actually measuring your device, and not just
>> displaying the limits of the test equipment.
>>
>>> Ultimately, it would be nice to
>>> compare the frequency and time offsets to a reference source that is
>>> accurate long-term, like my GPS hat Raspberry Pi.
>>
>> You have the right idea, although a Rasberry Pi is probably not the ideal
>> tool for that job.
>> There are different tools needed as well for comparing how far apart two
>> devices have drifted over the course of a day, week, or month compared to
>> what you would use to determine how much the frequency varies second to
>> second, or millisecond to millisecond, or how the time period between any
>> two positive going edges of your 10MHz signal compares to the time between
>> any other two consecutive edges.
>>
>>> Any ideas, suggestions, or clarifications are welcome. Additionally, if
>>> anyone knows of any guides (either text, video, or whatever), that would
>>> be great too.
>>
>> You could start with this as a breezy introduction:
>> http://leapsecond.com/ten/
>>
>> This will keep you busy for a long time:
>>
>> http://www.resonal.com/Downloads/John%20R.%20Vig%20-%20tutorial%20on%20Quartz%20Crystals%20and%20Oscillators.pdf
>>
>> (apologies if my mail editor mangles that link, hopefully you can put it
>> back together successfully if that is the case).
>>
>> --
>> Chris Caudle
>>
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at lists.febo.com -- To unsubscribe send
>> an email to time-nuts-leave at lists.febo.com
>> To unsubscribe, go to and follow the instructions there.
>>
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com -- To unsubscribe send an email to time-nuts-leave at lists.febo.com
> To unsubscribe, go to and follow the instructions there.
More information about the Time-nuts_lists.febo.com
mailing list