Question on crystal selection: parallel vs series cut

[khufumen]

Upon reading the specs of the 18F1320 PIC that I'm using, it says I need a parallel cut crystal. When I go on Digikey, I see lots of crystals such as the HC49/US but I don't see any mention of the crytals being parallel cut. I'm planning on using an 8Mhz crystal. I'd use the internal oscillator but the PIC is going to be outside in fluctuating temperatures. In any event, should I just assume an 8Mhz HC49/US crystal from digikey is a parallel cut crystal?
Hoping you can point me in the right direction.
Eric

[Melanie]

Assume NOTHING!

Download the Datasheet for the crystal.

This applies to ALL components that you intend to use. Yes, even humble Resistors have Datasheets.

But an 8MHz crystal will almost certainly be designed for parallel operation. The difference between parallel and series is only a few kHz anyway (and all crystals will oscillate in BOTH modes), but if you look at the way Microchip is asking you to connect the crystal, those Capacitances that you add, are being added in PARALLEL (as far as the oscillator is concerned) across the crystal.

More important (if you want accuracy) is to make sure that the Capacitance that you add matches what the crystal wants (see Crystal's Datasheet). Add too little and it will oscillate high (or not at all), add too much and it will oscillate low (or not at all). There is only a small range within which you can 'pull' a crystal before it ceases to work.

The Capacitances that Microchip quote are a GUIDE, the actual values you need are determined by the Crystal. Your design (layout etc) all adds Capacitance and a few pF here and there all adds up.

It is a popular MYTH that there are two ways you can CUT a crystal. RUBBISH!!!!

There are DOZENS of ways you can CUT a crystal. Some of my favourites are on display at Tiffany's (5th Avenue, New York). But there are only TWO ways you can make a crystal resonate... in SERIES mode or in a PARALLEL mode oscillator CIRCUIT. And like I said, ALL crystals can resonate in BOTH modes. Crystals (regardless of which CUT they use), are designed to be CLOSEST to their design frequency with the correct Capacitance applied when using the correct MODE. Low frequency Crystals (eg your 8MHz) are almost always designed for a PARALLEL MODE operation.

Finally I'm going to upset you...

DON'T think that a crystal will solve your frequency stability problems with fluctuating temperatures. It won't. It'll be a lot better than an RC circuit, but it will still drift with temperature (most likely by several kHz). This is also because temperature and humidity will affect your board Capacitances and components. If you are using the Crystal for Serial I/O, you may find the drift is actually well within tollerance. If you are using the crystal for ACCURATE timekeeping... no chance! If you are using the Crystal for Radio Communication (for say a PIC controllerd Digital Synthesiser)... *laughs* you're going to need help...

[Ioannis]

... Some of my favourites are on display at Tiffany's (5th Avenue, New York)...

Diamonds are girls best friend! Do you compromise with high grade HC49 Xtals with golden SMD pads to hang from your earlobs?:)


...If you are using the crystal for ACCURATE timekeeping... no chance! If you are using the Crystal for Radio Communication (for say a PIC controllerd Digital Synthesiser)... *laughs* you're going to need help...

An oven will be needed. Even a simple one like Romans Black: http://www.romanblack.com/xoven.htm

Ioannis

[J. Mark Wolf]

Upon reading the specs of the 18F1320 PIC that I'm using, it says I need a parallel cut crystal. When I go on Digikey, I see lots of crystals such as the HC49/US but I don't see any mention of the crytals being parallel cut. I'm planning on using an 8Mhz crystal. I'd use the internal oscillator but the PIC is going to be outside in fluctuating temperatures. In any event, should I just assume an 8Mhz HC49/US crystal from digikey is a parallel cut crystal?
Hoping you can point me in the right direction.
Eric

This is why I gave up on crystals years ago, in favor of oscillators and resonators.

[Ioannis]

This is why I gave up on crystals years ago, in favor of oscillators and resonators.

It not as bad as Melanie showed :)

And relatively to Resonators even with not precisely selected capacitors, the accuracy is much better.

Ioannis

[khufumen]

Thanks for your help! The PIC is going to be used for serial communication at a 9600 baud rate so I should be good. I was not aware of the critical role that these small capacitances played in the crystal's performance. (Now I know) That crystal oven seems like an interesting idea for possible future timing applications. Thanks again.
Eric

[Paul]

Despite what people may think about crystals and their stability issue, as with all things, high tolerance, good stability and frequency stable crystals can be purchased at a price. Most crystals that were designed for standard microprocessor based applications were never intended for high stability / accuracy apps. These crystals tend to have a frequency stability range between 250ppm - 50ppm / C.

I purchased a 10Mhz crystal not so long ago that has a stability factor of 1ppm/C. At 10Mhz, it drifts 10Hz per C, which for accurate timing applications would be more than enough.

Search through electronic suppliers catalogues for crystal specs and you should see what I mean.

Paul.

[Michael Wakileh]

EEVblog #61 - Crystal Oscillator Drift http://www.eevblog.com/episodes/
may be worth watching