MCU clock divider vs. VHDL divider

[Matt Clement]

Hello

I have built both a PIC microchip controlled clock divider as well as a CPLD
clock divider in the past for various projects but was told today that a
VHDL or discrete logic will always be "cleaner" than one run with a PIC. Is
this accurate? We are looking to create a clock on the order of 10-20Khz
from something faster. We are looking to get a very low jitter output.
Anyone offer any data backing either design?

thanks

 

[Pascal Peyremorte]

Matt Clement a écrit :

Hello

I have built both a PIC microchip controlled clock divider as well as a CPLD
clock divider in the past for various projects but was told today that a
VHDL or discrete logic will always be "cleaner" than one run with a PIC. Is
this accurate? We are looking to create a clock on the order of 10-20Khz
from something faster. We are looking to get a very low jitter output.
Anyone offer any data backing either design?

thanks

Good evening,

I do not agree fully with the one "cleaner" than the other. "Differents"
would be better...

If you wants an asynchronous divider (driven by original clock), PIC
cannot handle it. PLD can, but jitter will no be handled very well.

If you wants a synchronous divider (synchronized to an external clock),
PIC has its output latched from its main clock. PLD can have a same
output configuration.
10..20 kHz can be handled by both them in the same manner if the main
clock is large enought to use an integer count value for compute wave
timings.
But ... you cannot use "C" to write the PLD description, so do not use
it for pic if you wants to have comparable results Use asm.


Pascal

 

[Mike Treseler]

Matt Clement wrote:

I have built both a PIC microchip controlled clock divider as well as a CPLD
clock divider in the past for various projects but was told today that a
VHDL or discrete logic will always be "cleaner" than one run with a PIC. Is
this accurate? We are looking to create a clock on the order of 10-20Khz
from something faster. We are looking to get a very low jitter output.

The jitter from a hardware divider
would be almost as good as the reference clock.
If that were the only function I needed,
I would just buy a clock/divider chip.

The microcontroller output jitter would depend
on port output execution timing and variable delays
from caches, irq etc. Do the math.

-- Mike Treseler

 

[Marcus Harnisch]

Mike Treseler <mike_treseler@comcast.net> writes:

The microcontroller output jitter would depend
on port output execution timing and variable delays
from caches, irq etc. Do the math.

Unless, of course, the MCU has a programmable PWM timer. Many do these
days.

-- Marcus
The Germans have inherited a filthy Saxon culture and no more need be
said about them.
-- Mark Ballard, The Register