I'd like to use an ATS sized crystal (e.g. CTS ATS111) with an 89C664
because it's much smaller than the traditional HC49 case. However the
datasheet for the crystal specified a maximum drive level of 100uW
(yes, microwatts!). I've no idea how much power the 89C66x on-chip
oscillator actually drives the crystal with, but 100uW doesn't sound
like much :-) Any advice?
Unfortunatley it says nothing about Rs. I did find another reference
on the CTS page which specifies the crystal "ESR" as 40..150 ohms,
depending on freqency.
I also found this interesting paper on the topic
Taking Rs around 50 ohms and f=18.423Mhz, it sounds like we're
talking about at least 1mW. Doesn't look good ...
Is there any other option besides one of those giant HC49s??
Yes, that looks like a problem if you can't use a more common
Well, there are SMT oscillators, and they are a bit expensive and not
all that much smaller (7 x 5mm).
"it's the network..." "The Journey is the reward"
firstname.lastname@example.org Info for manufacturers: http://www.trexon.com
Unfortunately, it's not quite that simple - recall that the crystal model
motional inductance (Ls), motional capacitance (Cs), shunt capacitance (Co),
resistance (Rs). It's exceptionally difficult to calculate the real power
disappation in the
100uW is probably higher than your on-chip oscillator is supplying. It's
typically in the 1's
or 10's of uW. Some very low-power IC's are in the 100's of nW.
I'd recommend the following:
(0) use load caps for the crystal that are double the value of the crystal's
specified Cl. For
example, if the crystal is spec'd for 12pF load capacitance, you should use
two 22pF caps.
Always 'round down' to the next standard value - you've got stray C on your
crystal type should be a PARALLEL RESONANT type, and should be a FUNDAMENTAL
type (NOT an OVERTONE type!!).
(1) insert an external fixed resistor in series with the crystal. Place
this resistor between
the crystal terminal and the OUTPUT of the on-chip oscillator. The value
resistor should be 4x the ESR of the crystal you're intending to use.
(2) temporarily include a ~5K potentiometer in series with your resistor.
So, you should
have, in order: IC output pin -> pot -> fixed resistor (3x ESR) -> crystal
(and one load
cap to ground at this node as well). Set the pot for MAX resistance.
(3) at this point, the crystal oscillator shouldn't start when supply
voltage is applied to
the uC. If it does, use a larger pot. You don't want the oscillator to
start with the pot at
(4) decrease the resistance of the potentiometer until the crystal starts up
reliably. If your
application involves working over a temperature range, then make sure you
this test at the temperature extremes. Same goes for voltage range, too, if
you're working off
of an unregulated supply.
(5) remove the pot from the circuit and measure its value.
(6) replace the fixed resistor (4x ESR) with a fixed resistor of the same
value as the pot's
So, you should have IC output pin -> fixed resistor (equal to the value
measured from the
potentiometer) -> crystal (and one load cap to ground).
What you've just done is ensure that you've got sufficient "oscillation
margin" to guarantee
startup - is desireable to have the "negative resistance" in the loop be
about 3 ~ 5 times the
ESR of the crystal. That helps ensure that your crystal isn't overdriven
and runs at a spurious
mode frequency. Unfortunately, the ESR value is really dependent upon the
drive level, but
without ESR data at this particular IC's drive level, this is all you can go