PLL Stability and Any Gotchas?

Hi,

We'll be using the PLL for the first time in a product, out of necessity 
(the micro only allows a crystal up to 8MHz, but we need a 32MHz clock). 
The PLL will also be clocking a CAN module within the microcontroller.

I'd be grateful for insight into any of the following questions:

a)How do long do PLL's take to start?

b)Do PLL's ever fail to start?

c)If there is more than once choice for a crystal frequency where we can 
multiply it up to 32MHz (for example, 8Mhz x 4 = 32MHz, but 4MHz x 8 = 32MHz 
also), what are the best choices?

d)Is there anything to watch out for in the hardware or software design (any 
gotchas)?

e)Is there anything we can do to make the system as robust and stable as 
possible (i.e. decrease the probability of a PLL unlock)?

f)Can anyone recommend any books or technical papers they've found useful?

Thanks for the help,
Dave Ashley. 

David T. Ashley wrote:
> Hi,
> 
> We'll be using the PLL for the first time in a product, out of necessity 
> (the micro only allows a crystal up to 8MHz, but we need a 32MHz clock). 
> The PLL will also be clocking a CAN module within the microcontroller.
> 
> I'd be grateful for insight into any of the following questions:
> 
> a)How do long do PLL's take to start?
> 
> b)Do PLL's ever fail to start?
> 
> c)If there is more than once choice for a crystal frequency where we can 
> multiply it up to 32MHz (for example, 8Mhz x 4 = 32MHz, but 4MHz x 8 = 32MHz 
> also), what are the best choices?
> 
> d)Is there anything to watch out for in the hardware or software design (any 
> gotchas)?
> 
> e)Is there anything we can do to make the system as robust and stable as 
> possible (i.e. decrease the probability of a PLL unlock)?
> 
> f)Can anyone recommend any books or technical papers they've found useful?

"Phaselock Techniques" by Floyd M. Gardner is
pretty much my bible on the subject.

http://www.amazon.com/gp/product/0471042943/103-2979903-6085438?v=glance&n=283155

David T. Ashley wrote:
> Hi,
> 
> We'll be using the PLL for the first time in a product, out of necessity 
> (the micro only allows a crystal up to 8MHz, but we need a 32MHz clock). 
> The PLL will also be clocking a CAN module within the microcontroller.
> 
> I'd be grateful for insight into any of the following questions:
> 
> a)How do long do PLL's take to start?

That depends on the manufacturer.  Motorola was pretty good about 
letting you know in their data sheet once you found the one page in 1000 
that was pertinent.  I would expect any manufacturer to be pretty 
up-front with this.
> 
> b)Do PLL's ever fail to start?

Egad!  I would expect that if a chip with a PLL in it had that sort of 
problem you'd hear the screaming from wherever you were.
> 
> c)If there is more than once choice for a crystal frequency where we can 
> multiply it up to 32MHz (for example, 8Mhz x 4 = 32MHz, but 4MHz x 8 = 32MHz 
> also), what are the best choices?

In general the higher the reference frequency the better the jitter, so 
your 8MHz choice is probably best in that regard.  But -- do you need 
that sort of jitter performance?

Here again this is highly manufacturer-specific, and I would expect to 
find it in a data sheet.
> 
> d)Is there anything to watch out for in the hardware or software design (any 
> gotchas)?

Hardware:  Pay attention to grounding and bypassing.  Do what the 
manufacturer tells you.

Software:  Don't change the PLL rate unless you _really_ know what 
you're doing.  Set it at start up and leave that thing alone!
> 
> e)Is there anything we can do to make the system as robust and stable as 
> possible (i.e. decrease the probability of a PLL unlock)?

The PLLs in those things are, AFAIK, pretty simple and robust.
> 
> f)Can anyone recommend any books or technical papers they've found useful?
> 
> Thanks for the help,
> Dave Ashley. 
> 
The Garner book mentioned is a good one, AFAIK.  I prefer "Phase-Locked 
Loop Circuit Design" by Wolaver (Prentice-Hall, 1991), but only because 
I took the class from the guy who wrote it.  Mostly what understanding 
PLL theory will do for you is help you understand why the manufacturer 
tells you what they do.  The details of how to best make _their_ circuit 
work depends so heavily on the details of _how_ they made their circuit 
that you can't really influence much.

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google?  See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

David T. Ashley wrote:

> Hi,
> 
> We'll be using the PLL for the first time in a product, out of necessity 
> (the micro only allows a crystal up to 8MHz, but we need a 32MHz clock). 

Unusual Micro Spec - which one ?

> The PLL will also be clocking a CAN module within the microcontroller.
> 
> I'd be grateful for insight into any of the following questions:
> 
> a)How do long do PLL's take to start?
> 
> b)Do PLL's ever fail to start?
> 
> c)If there is more than once choice for a crystal frequency where we can 
> multiply it up to 32MHz (for example, 8Mhz x 4 = 32MHz, but 4MHz x 8 = 32MHz 
> also), what are the best choices?
> 
> d)Is there anything to watch out for in the hardware or software design (any 
> gotchas)?
> 
> e)Is there anything we can do to make the system as robust and stable as 
> possible (i.e. decrease the probability of a PLL unlock)?
> 
> f)Can anyone recommend any books or technical papers they've found useful?
> 
> Thanks for the help,
> Dave Ashley. 

Do you need a PLL ?
A 32MHz xtal module could feed straight into the uC ?

-jg

"David T. Ashley" <dta@e3ft.com> wrote:
>We'll be using the PLL for the first time in a product, out of necessity 
>(the micro only allows a crystal up to 8MHz, but we need a 32MHz clock). 
>The PLL will also be clocking a CAN module within the microcontroller.

<..PLL questions deleted..>

If budget, PCB real state, etc. allow it, wouldn't it be simpler to
use an external 32Mhz oscillator + divider to provide 8Mhz for the
CPU?