# Crystal Oscillator Question

#### Kevin2341

Joined Nov 1, 2009
19
Hey everyone, I'm doing a small on-my-own project to build myself a binary clock. Since I haven't personally learned anything about crystal oscillators in any of my EE classes yet, I had to use someone else's design for this part of my project. The circuit I used was this:
http://hackersbench.com/Projects/1Hz/

I understand all of his circuitry here except the crystal oscillator, which has me a bit confused as to how these things work, and why it's set up in this configuration. From my understanding about crystals, is that it's function is very similar to a RLC circuit, without being a RLC circuit (obviously). I noted when I bought the "tuning fork" as many websites refer to it is, that the component itself was rated for a very small capacitance, which I presume has an effect on the stability of the oscillation pulse.

The other thing to note, in my physical build of this circuit (minus the ICs), is that I used values that are definitely going to be off what he suggests; for instance, I bought 22pF capacitors as I couldn't find any 33pF. I didn't have any 6.8M resistors, so I put a series of resistors to make a near-equivalence of 13.6 (I think it's actually 13.2M now that I think about it). Anyway, my point in sharing this is that it doesn't appear to have an effect on the stability of the clock pulse. I have the signal being divided down to 1Hz, and comparing it against a timer I know is accurate (for all intents and purposes), it seems to be pretty darn accurate. If I had to give a theory on why these resistors and capacitors are these sizes, I'd imagine it has something to do with a fairly specific voltage and a very small current (in the hundreds of nano amps from a quick calculation).

So anyway, long story short; how does resistance and capacitance ratios affect crystals, and why must it be in this kind of configuration for my application?

Thanks!

#### GopherT

Joined Nov 23, 2012
8,009
Some things are more important than others. I am not surprised that it works. You'll be fine.

#### MikeML

Joined Oct 2, 2009
5,444

#### MrChips

Joined Oct 2, 2009
28,155
Kevin, you're a bit off track. Crystal oscillators are nothing like RLC circuits.
A crystal oscillator circuit is a basic feedback oscillator. The feedback resistor is typically 10-22M-ohm and is not critical.
The frequency of oscillation is determined by the crystal and is usually very accurate to better than 10 ppm.
What is 10 ppm? That is like gaining or losing 1 second in a day.
You will not be able to judge this accuracy against any watch or stop-watch unless you run the test over about 5 days or longer.

The series resistor R1 is to reduce the loading on the crystal to prevent damage to the crystal.
The capacitances C1 and C2 in parallel provide a load on the crystal and assist in proper oscillator start-up. These are typically 15-33pF and are not critical except for the following.

If you want to create a real-time-clock or accurate frequency counter then you want the oscillator to be as accurate as possible. Better than 1ppm is achievable by trimming C2. For this you replace C2 with a trim capacitor. For this you will need to calibrate the oscillator against a accurate frequency standard or frequency counter.

1ppm is equivalent to gaining or losing 1 second in 12 days.
1/10 ppm is equivalent to gaining or losing 1 second in 4 months.

I would tune a SW radio to WWV at 5.000MHz or 10.000MHz and then compare my watch against their time signal.

#### Kevin2341

Joined Nov 1, 2009
19
Kevin, you're a bit off track. Crystal oscillators are nothing like RLC circuits.
A crystal oscillator circuit is a basic feedback oscillator. The feedback resistor is typically 10-22M-ohm and is not critical.
The frequency of oscillation is determined by the crystal and is usually very accurate to better than 10 ppm.
What is 10 ppm? That is like gaining or losing 1 second in a day.
You will not be able to judge this accuracy against any watch or stop-watch unless you run the test over about 5 days or longer.

The series resistor R1 is to reduce the loading on the crystal to prevent damage to the crystal.
The capacitances C1 and C2 in parallel provide a load on the crystal and assist in proper oscillator start-up. These are typically 15-33pF and are not critical except for the following.

If you want to create a real-time-clock or accurate frequency counter then you want the oscillator to be as accurate as possible. Better than 1ppm is achievable by trimming C2. For this you replace C2 with a trim capacitor. For this you will need to calibrate the oscillator against a accurate frequency standard or frequency counter.

1ppm is equivalent to gaining or losing 1 second in 12 days.
1/10 ppm is equivalent to gaining or losing 1 second in 4 months.

I would tune a SW radio to WWV at 5.000MHz or 10.000MHz and then compare my watch against their time signal.
Thanks for the reply, that was very helpful. The reason I claimed it to be somewhat like an RLC circuit is because I saw several explanations of these crystals compared to those kinds of circuits. (http://en.wikipedia.org/wiki/File:Crystal-oscillator-IEC-Symbol.svg).

As far as accuracy goes, I'll probably look into getting a new tuning fork. Just looked into what I bought and it has a rating of 50ppm lol. I imagine with that I'd have to readjust the time on my clock once every couple days to keep it somewhat accurate. Also, I'm not exactly sure how to go about calibrating a signal like this to a radio like you described. I have access to some fairly decent oscilloscopes at my school's labs, but as for matching a frequency against another I don't have any experience or know-how on this topic (I would imagine that's how you're calibrating your signals against the radio? Forgive me on that one, I'm within my 5th semester as an EE and I'm doing more of a computer engineering approach as compared to RF\Power\etc).

#### MrChips

Joined Oct 2, 2009
28,155
An oscilloscope is not going to get you anywhere near the kind of accuracy you need.
What you need is an accurate frequency counter or universal counter with at least 7-8 digits, something like an HP/Agilent 53131A with a temperature controlled crystal oven.

I have an HP 5325A that is still running strong.