thereza wrote:
William -
Thanks for the excellent information. My only confusion is that my adviser
(I'm a grad student) mentioned that the load capacitance of a crystal has to
be equal to what the MCU expects. I had thought that it didn't matter as
long as you got the values equal to what the crystal expects. Who was
right? Or is it MCU specific?
Also, I've noticed that there is a new 16Mhz version of the chip -- have you
had any experience with it or have any opinions about it?
According to the specs, the DCO on the new core is accurate enough to
run a UART directly (it's about 1%, IIRC, and UART normally needs 2.5%).
mvh.,
David
Thanks!
reza
wh_fcab wrote:
Reza,
Isn't T.I. documentation great? There's tonnes of it, yet you can't find
certain basic information... :)
Okay, with the XT2 oscillator, you can use either a crystal or resonator
of up to 4 MHz if your supply voltage is 2.2V or higher. Go up to at
least
2.8V and you can use 8 MHz. The resonator should be hooked up between the
XT2IN and XT2OUT pins. An AT-cut crystal can be hooked up the same way,
but design your circuit board to provide for connecting two loading
capacitors to ground, one from each side of the crystal. The crystal
manufacturer usually indicates the load capacitance to use; in general
they
are 12 to 22 pF for an 8 MHz crystal. (Note that the oscillator will run
at 8 MHz at 2.8V or higher, but the CPU itself needs 3.6V to reliably
operate
at 8 MHz; see the graph in the Recommended Operating Conditions chapter
near
the end of the 'F13x,14x,14x1 datasheet.
The built-in capacitors you saw reference to are only for use with a 32
kHz
watch crystal, if I remember correctly.
Regarding the DCO/multiplier, you can get a DCO output of around 7 MHz
that is
stable enough for 115200 baud communication with the USART. I use a
multiplier
of 218 for 7.134324 MHz (well, it's as close to that frequency as the
crystal is
good) because it's the highest frequency the 'F149 can run at 3.3V over a
reasonable temperature range. 8 MHz can be used only at 3.6V. The key
with the
'F1xx chips, which lack a hardware FLL, is ensuring you have a good
*software*
frequency-locked loop. My colleague and I tweaked our software FLL
algorithm
many times (over many years) before getting DCO output that was
jitter-free
enough to run 115200 baud without spurious framing errors.
Cheers,
William
thereza wrote:
I hope that this isn't too off topic, if so, is there another mailling
list
that I can be directed to?
I'm trying to switch over to the XT2 on a MSP430F149 but I'm having
problems
finding documentation on how the crystal should be hooked up. Can I
attach
a ceramic resonator with built in caps or will it only work with a
crystal
-- it also seems to imply that it has built-in caps for the xtal -- is
this
true?
Also, with the built in frequency multiplier can I get a stable 8Mhz (or
16)
with just a watch crystal -- enough for some high speed serial
communication
via the uart/spi?
Thanks,
Reza
