Resistor on crystal pin?

[MatCat]

I am working on a board using an STM32F103 chip and noticed the recommendation of a resistor on the crystal, I was curious about this and noticed a reference to another AN about oscillator circuit design which I took a look at, and basically said the resistor to limit the current through the crystal to lengthen it's life.  Now my thought is on this is that I can't recall a schematic where I have seen a resistor used on a crystal, and since reading this even purposely went looking up schematics using this chip and NO ONE puts the resistor, so I am curious to think what the real EEs out there have to say on this subject?

[kizzap]

Most people will say to follow the datasheet. Reason being that the datasheet will have been checked and tested over and over, and is guaranteed to work by the manufacturer. Only the manufacturer truly knows what is going on inside their device.

As to whether or not it is needed...I have no idea. So my vote goes to following the datasheet.

-kizzap

[ElectroWarrior]

Is it serial or parallel resistor? - I've never seen such a thing, however we can always theorize:)
Series: The only application of a series resistor as I see is inrush current protection when the micro is powered up and the crystal is activated. How much capacitance a crystal has in the initial state is unknown for me. Otherwise it could be to prevent ringing effects, like you would use a small serial resistor between a micro and a mosfet gate.
Parallel: Wikipedia has a interesting theory:
"A DC voltage bias between the electrodes can accelerate the initial aging, probably by induced diffusion of impurities through the crystal. Placing a capacitor in series with the crystal and a several-megohm resistor in parallel can minimize such voltages"

Best Regards
ElectroWarrior

[Stonent]

I've only seen pF range caps on crystals before, so yeah, this is a new one for me. (Not that I have a whole lot of experience in the stuff...)

[Thor-Arne]

Not that uncommon, i.e. the Arduino Uno uses parallel resistance on the crystals.

I don't know why though.

[MatCat]

To see what I am talking about check out Section 2.2 Figure 9 and the associated notes.  The resistor is in series. http://www.st.com/st-web-ui/static/active/en/resource/technical/document/application_note/CD00164185.pdf

[tszaboo]

and since reading this even purposely went looking up schematics using this chip and NO ONE puts the resistor,

I did it for ST, it is working. Cannot share it, NDA, sorry. The datasheet and the AN is correct.

[MatCat]

and since reading this even purposely went looking up schematics using this chip and NO ONE puts the resistor,

I did it for ST, it is working. Cannot share it, NDA, sorry. The datasheet and the AN is correct.

Interesting, I can understand NDA for certain things but it seems pretty plain in the AN that the purpose is to limit current, I understand that, but my original question if you will was really how come in reality very few follow this practice in their layouts? Maybe a better question even to ask is this something unique to ST, or something inherit to all microcontrollers?  I imagine that question should be answerable without violating NDA (assuming not going into details?)

[tszaboo]

I mean NDA for posting Schematic or component type. And I'm over cautious.
If I remember correctly Atmel even has settings for drive strength of the crystal. But also, if you read about it,
http://en.wikipedia.org/wiki/Pierce_oscillator
"n addition to the biasing resistor R1, Ruan Lourens strongly recommends a series resistor Rs between the output of the inverter and the crystal. The series resistor Rs reduces the chance of overtone oscillation and can improve start-up time."
Also this:
http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1824&appnote=en020706
http://www.freescale.com/files/microcontrollers/doc/app_note/AN3208.pdf
Not putting the resistor there is probably just common ignorance for facts by the industry/hobby mixture. I've seen worse. One of the favorite is the USB:
http://nandblog.com/usb-5v/

[Hideki]

I guess you haven't looked at enough schematics. Many of ST's own devboards use a resistor; for example 390 ohms for the STM3210E-EVAL.
Others use 220 ohms and some don't use a resistor at all. It depends on your choice of crystal.

[MatCat]

Yeah I went through and actually read the rest of that AN, quite interesting. 

[SeanB]

The resistor is there for limiting drive to the crystal. If the oscillator is designed correctly then it uses small mosfets internally with a high channel resistance that does this automatically. If it is just a regular inverter then it can drive the crystal with too much power, which can cause problems later with the crystal wearing, which can then mean it will either drift or will not start up correctly. In extreme cases it can crack the crystal as it exceeds the elastic limits of the crystal.

It was a problem on the original Apple II where the clock would stop running, the cure being to unsolder the crystal and turn it around and resolder it, it would then work again. I did the same on an old massmeter where it had an intermittent startup and this got it going again for close to a decade before the unit failed again.

[Whuffo]

The amount of crystal drive you apply to your crystal is important. Too little and it many or may not oscillate, or maybe it'll run at some overtone. Too much and you'll damage the crystal. A little too much shortens its life, way too much leads to short term total failure.

The proper amount of drive depends on the particular crystal; check its datasheet. And yes, series resistors (or some other method of limiting drive) are almost always required. When designing crystal oscillators, never cut corners - use the crystal specifications and design your circuit accordingly. Resistors are commonly used because they're cheap, and it's easy to stuff a different value resistor when a different crystal is used.

Resistors are also often used because the "engineer" has trouble with polar / rectangular calculations and sticks with something that doesn't have a varying phase angle associated with its impedance. The Jedi engineer knows these things, and designs the circuit to provide the appropriate impedance through its reactive devices. Datasheet designs are fine - as long as the crystal type and frequency are close to what you'll be using. Otherwise, you're going to have to do the math.

[dannyf]

It (a serial resistor) is to control the drive levels to a (typically low frequency) crystal. Typically a few 10k - 100ks. Incidentally, many will not oscillate (at all or at the right frequency) without such a resistor, particularly if a ST gate is used as the driver.