Re: Correct crystal oscillator choice.

On Mon, 29 Jan 2007 14:07:41 -0600, John Fields
<jfields@xxxxxxxxxxxxxxxxxxxxx> wrote:

On 29 Jan 2007 10:34:59 -0800, pacemkr@xxxxxxxxx wrote:

I am building one of these http://plazma.kapsi.fi/diy/lanc_controller/
Its a LANC Controller (ie. remote control) for a Sony camcoder.

The project uses a PIC16F84A chip, which is a programmable
microcontroller.
I am very comfortable working with assembler, but I am not so
profficient with electronics. I'm having a problem with this project.

I've assembled everything on a breadboard (the chip is programmed and
verified), but it doesn't seem to work. I have a striking suspicion
that I bought a wrong crystal oscillator, and I can't find any
information on how to choose the correct device.

---
First, what you bought isn't an oscillator, it's a quartz crystal.

The oscillator is the circuitry around it that makes it oscillate at
its resonant frequency.

All crystals have two fundamental resonant modes, parallel and
series, and they can be made to oscillate in either mode. The
difference between the two modes is that the crystal will look like
a high impedance when it's run in parallel resonance and a low
impedance when it's run in series resonance. Also, its parallel
resonant frequency will be slightly higher than its series resonant
frequency. The way you choose which mode you need is to look at
how the crystal will be driven. If it'll be driven with an
inverting amplifier, like the one in the PIC16F845A, then you go for
parallel resonance. If it'll be driven by a non-inverting
amplifier, then you specify series resonance.
---

Do crystal oscilators come in different flavors when it comes to
waveform shape? Since this oscillator is used as a clock in the
circuit, does that mean that I need one that generates a square wave?

---
No. There's a class of crystal oscillators called "crystal clock
oscillators" which have all of the necessary circuitry in them to
provide a square-wave clock output which can drive logic directly,
but that's not what you have. In your case the logic-level square
wave clock will be generated internal to the chip.
---

The oscillator I bought and am currently using is this one: http://
www.newark.com/jsp/Passives/Crystals/VISHAY+DALE/XT49S-204M/
displayProduct.jsp?sku=18C1488

Also I bought 22pF capacitors instead of the 18pF suggested in the
project. Would that affect anything?

---
It would lower the oscillator frequency.
---

As a related question, I've noticed that the clock signal circuit for
many chips is the same as the one in this project (crystal oscilator
and two caps), how do you know what capacitor values to use? I've read
in several places that it depends on the characteristics of the
specific oscillator, but I could not find any more detail.

---
That oscillator is classic, is called a "Pierce" oscillator, and
Google will give you lots of hits on how it works.

Briefly, though, you look at the data *** for the crystal and find
what its load capacitance is supposed to be. Then you subtract the
stray circuit capacitance from that and get two caps each twice the
capacitance of what's left over.

---
P.S.

I forgot to mention that since the PIC oscillator is using an
inverter to drive the crystal, the crystal must be operated in
parallel resonance, and since parallel resonant crystals need those
two "loading" caps and the Newark data shows the loading required,
you got the right crystal.



--
JF
.