Due to the impacts of the coronavirus outbreak, we are experiencing longer than normal lead times on certain products. We encourage back-ordering out-of-stock items to receive them as soon as possible.

According to Pete: All About Crystals

To honor our 15th anniversary, Pete digs into the electrical properties of crystals.

Favorited Favorite 1

Did you know… that you give crystal for a fifteenth anniversary? Yeah, I didn’t either. But did you know that we are entirely dependent on the little electronic variety of crystals, and yet we mostly just strap them to a microcontroller and forget about them? The microcontroller wouldn’t function without a clock source, would it? Oh sure, some of our fave uC’s have an internal RC clock that would serve, and you could arguably go to the moon on an ATMega328 running that way. But most our lives run at much higher speeds these days, and crystals are still at the heart of it.

But what exactly are these things? How do they work? Is it magic? Well, maybe a little. In this video, I’ll give you a quick-ish rundown on what they’re all about. And while it’s not primarily about oscillator design, I’ll show you a couple of circuits to play with at the end, comprised of parts you’ve probably got lying around.

alt text

Comments 7 comments

  • Another great ATP!

  • What is the use of the power supply capacitor and is it necessary to make the oscillator work?

    Normally transistors are used as either switches or amplifiers. When entering the (high) frequency range you can use it as a oscillator. The Electrical reactance of the power supply capacitor will lower when the frequency goes up, it will get close to zero. If you now take a look at the circuitry, the oscillator makes more sense with a feedback (the crystal) from collector to base to oscillate. The higher you make the power supply capacitor the easier it will oscillate. Hopes this helps.

  • You have excellent timing Pete, thanks! I'm working on a project that will screen AT cut crystals for durability, this episode clarified and answered many questions.

    P.S. Enjoyed meeting you and our chat in the Sparkfun lobby the other day.

  • Just a comment: Although crystals ARE available up to about 30 MHz or so, in practice they aren't used much above about 16 MHz. You did mention in passing PLLs, and today it's been my experienace that nearly all "modern" applications that need something other than the handfull of "standard" frequencies use a PLL to generate it. (PLLs can be used to generate frequencies from way below the "reference clock" to way above it.)

    Before the days when PLLs were common, to get frequencies above the crystal's fundamental frequency, you could use "multiplier circuits" to get harmonics. If you needed an even multiple (such as 8 MHz out of your 4 MHz crystal), you'd do a "full wave rectifier" and then an 8 MHz filter to get it back to (resembling) a sine wave. If you needed an odd multiple (such as 12 MHz from the 4 MHz crystal, though this would also work for 20 MHz and even 28 MHz, i.e., 3x, 5x, and 7x the base frequency) you'd ampllify the signal into clipping so that it would resemble a square wave, then apply the appropriate LC filter to "extract" the desired component (and then another amplifier to get it back up to a useful level). Chains of these things could be done, for example, to get a frequency in the range of 144 MHz out of a crystal in the range of 8 MHz or 12 MHz. This sort of thing was even used back in the days of vacuum tubes, and survived into the all-transistor days.

    Serious users of crystals should be aware that the temperature of the crystal DOES effect the frequency. The "old fashioned" way of dealing with this is to put the crystal into a temperature controlled "oven" (usually the temperature is noticably above what the circuit could normally experience, e.g., say around 180F for a "gadget" to be in a vehicle). Some modern circuits that are in essence PLLs measure the temperature "near" the crystal, and know what the temperature "coefficients" are for that crystal (from the crystal's datasheet), and adjust the PLL appropriately. (This can get "drift" for an RTCC [Real Time Clock Calendar] down around one second per month out of a 32.768 kHz crystal that costs well under $1, though the RTCC may be around $10.)

    Speaking of the 32.768 kHz crystal, just as a sidelight these are usually "tuning fork" configurations. Also, 32,768 Hz sounds like a very odd number, but if you realize that it's 2 raised to the 15th power (i.e., 2*2*2*2*... for 15 "2s"), it starts to make a bit more sense. Digital "divide by two" circuits are (relatively) simple, so it is relatively easy (especially if you're designing an integrated circuit) to obtain a crystal-controlled 1 Hz. It's also pretty difficult (or at least expensive) to make a crystal that has a frequency much below 30 kHz.

    • Thanks for chiming in! Your level of experience is obvious (you clearly know what you're talking about) and well represented.

      • One other thought: you can often find (old) crystals for sale at Hamfests, usually in what seem to non-Ham Radio folks to be "odd-ball" frequencies. These are often in cases that are meant to plug into a socket rather than be soldered in (so that they can be easily changed). It used to be that you could order crystals at specific frequencies fairly inexpensively (like around $10 a pop) and get them within a few days, but demand has dropped (with the advent of radios that use "synthesized" frequencies) to the point that most suppliers have disappeared.

Related Posts

Recent Posts

Dumpster Dive is Today

The Teensy Life


All Tags