Forums

PLL on LPC2000

Started by jdauchot February 2, 2009
Hi

I have a timer interrupt going at 1ms interval driving some hardware
interface. Once the PLL is programmed, can the oscilator frequency
drift?

Regards

Jean-Jacques

An Engineer's Guide to the LPC2100 Series

On Mon, 2 Feb 2009, jdauchot wrote:

> Once the PLL is programmed, can the oscilator frequency drift?

While I couldn't tell you the PPM, I have little doubt it will!

-Kenny

--
Kenneth R. Crudup Sr. SW Engineer, Scott County Consulting, Los Angeles
O: 3630 S. Sepulveda Blvd. #138, L.A., CA 90034-6809 (888) 454-8181

Am Montag, 2. Februar 2009 schrieb jdauchot:
> Hi
>
> I have a timer interrupt going at 1ms interval driving some hardware
> interface. Once the PLL is programmed, can the oscilator frequency
> drift?
>
> Regards
>
> Jean-Jacques

Hello Jean-Jacques,

yes, it can. This is because the crystal varies the frequency as the
temperature changes. Also, aging influences the frequency. But normally these
changes are rather small.

Greetings,

Chris

Hi

Thanks for the reply.

I have this timer interrupt driving a lagacy output subsystem which
uses 1ms pules like a UART protocol with 8 data packets with a start
bits (positive). How could I make sure these pulse are 1ms accurates?

Regards

Jean-Jacques

--- In l..., Kenneth Crudup wrote:
> On Mon, 2 Feb 2009, jdauchot wrote:
>
> > Once the PLL is programmed, can the oscilator frequency drift?
>
> While I couldn't tell you the PPM, I have little doubt it will!
>
> -Kenny
>
> --
> Kenneth R. Crudup Sr. SW Engineer, Scott County Consulting, Los
Angeles
> O: 3630 S. Sepulveda Blvd. #138, L.A., CA 90034-6809 (888) 454-
8181
>

> > > Once the PLL is programmed, can the oscilator frequency drift?

> > While I couldn't tell you the PPM, I have little doubt it will!

On Mon, 2 Feb 2009, jdauchot wrote:

> I have this timer interrupt driving a lagacy output subsystem which
> uses 1ms pules like a UART protocol with 8 data packets with a start
> bits (positive). How could I make sure these pulse are 1ms accurates?

Well, that's why I'd used the drift-value units "PPM", "parts per million";
between the (relatively stable, but still variable, depending on temperature
and the like) crystal oscillator and the PLL action, by the time that works
its way down past the counters and provides your 1ms pulse, I'm sure you'll
have no timing issues.

IOW, I wouldn't worry about it. :)

-Kenny, who has also used an NXP counter output as a clock source

--
Kenneth R. Crudup Sr. SW Engineer, Scott County Consulting, Los Angeles
O: 3630 S. Sepulveda Blvd. #138, L.A., CA 90034-6809 (888) 454-8181

OK how do I grandee this 1ms pulse not drifting

Regards

Jean-Jacques
--- In l..., Kenneth Crudup wrote:
> > > > Once the PLL is programmed, can the oscilator frequency drift?
>
> > > While I couldn't tell you the PPM, I have little doubt it will!
>
> On Mon, 2 Feb 2009, jdauchot wrote:
>
> > I have this timer interrupt driving a lagacy output subsystem
which
> > uses 1ms pules like a UART protocol with 8 data packets with a
start
> > bits (positive). How could I make sure these pulse are 1ms
accurates?
>
> Well, that's why I'd used the drift-value units "PPM", "parts per
million";
> between the (relatively stable, but still variable, depending on
temperature
> and the like) crystal oscillator and the PLL action, by the time
that works
> its way down past the counters and provides your 1ms pulse, I'm
sure you'll
> have no timing issues.
>
> IOW, I wouldn't worry about it. :)
>
> -Kenny, who has also used an NXP counter output as a clock
source
>
> --
> Kenneth R. Crudup Sr. SW Engineer, Scott County Consulting, Los
Angeles
> O: 3630 S. Sepulveda Blvd. #138, L.A., CA 90034-6809 (888) 454-
8181
>

On Mon, 2 Feb 2009, jdauchot wrote:

> OK how do I grandee this 1ms pulse not drifting

You can't, really (temperature control, maybe?), but what I/we're saying
is it shouldn't matter for your application; I'm pretty sure that at 1KHz
measureable drift won't really be a factor, esp. in a UART-style application.

-Kenny

--
Kenneth R. Crudup Sr. SW Engineer, Scott County Consulting, Los Angeles
O: 3630 S. Sepulveda Blvd. #138, L.A., CA 90034-6809 (888) 454-8181

Am Montag, 2. Februar 2009 schrieb jdauchot:
> OK how do I grandee this 1ms pulse not drifting
>
> Regards
>
> Jean-Jacques
>

Like Kenneth already said, normally the drift is so minimal that it makes no
big difference.

However, if you really want better stability, use a TCXO (Temperature
Compensated Crystal Oscillator) or an crystal-oven. TCXO's are pretty nice
already, while an oven gives you the best possible stability. But you have to
pay for it: with increased power consumption and a much higher price.

Greetings,

Chris

jdauchot wrote:
> OK how do I grandee this 1ms pulse not drifting

You can't stop it drifting without external reference. Use ntp, radio
time (en.wikipedia.org/wiki/Time_signal), gps, manual adjust... -- pick one.

--

Timo
> I have this timer interrupt driving a lagacy output subsystem which
> uses 1ms pules like a UART protocol with 8 data packets with a start
> bits (positive). How could I make sure these pulse are 1ms accurates?

Does the other end of the communications link also generate
1ms events?

Lets say your LPC is a slave device, and the other end of
the link is the master.

At startup the LPC slave initializes its 1ms timer to a nominal
count, and establishes communications with the master. The LPC
can time the master communication events and use it to calibrate
its 1ms timebase.

So long as the initial timing is close enough for the link to
be established, and the LPC tweaks its timing to match the master,
the drift of the oscillator should not matter.

You could also consider an auto-baud procedure where the master
sends 1ms pulses until the LPC responds. The LPC can use that
to calibrate its timers, and then respond.

Cheers,
Dave