FPGA Central

Hello,


I'm wondering if there is any advantage on using a separate regulator for
the Vccaux.

I already have 2.5V generated on the board by a switching regulator for
the multiple DDR chips and for some Vccio of the spartan 3. Now, with the
new TI regulator that has 3 regulators in a chip, I can have a separate LDO
to power just the Vccaux portion. Is it useful to have it separate ?


Thanks,

Sylvain

Hi,

see below
Sylvain Munaut wrote:

> Hello,
>
>
> I'm wondering if there is any advantage on using a separate regulator for
> the Vccaux.
>
> I already have 2.5V generated on the board by a switching regulator for
> the multiple DDR chips and for some Vccio of the spartan 3.

Do not use a switching reg for Vccaux, it's noisy and not recomended to
power your DCM (DCM is powered by VCCaux)

> Now, with the
> new TI regulator that has 3 regulators in a chip, I can have a
> separate LDO
> to power just the Vccaux portion. Is it useful to have it separate ?

Yes it is.
Aurash

>
>
>
> Thanks,
>
> Sylvain



--
__
/ /\/\ Aurelian Lazarut
\ \ / System Verification Engineer
/ / \ Xilinx Ireland
\_\/\/

phone: 353 01 4032639
fax: 353 01 4640324

Hi,

>> I'm wondering if there is any advantage on using a separate regulator for
>> the Vccaux.
>>
>> I already have 2.5V generated on the board by a switching regulator for
>> the multiple DDR chips and for some Vccio of the spartan 3.
>
>
> Do not use a switching reg for Vccaux, it's noisy and not recomended to
> power your DCM (DCM is powered by VCCaux)

Thanks for the information, just what I needed


Sylvain

On Thu, 18 Nov 2004 11:32:50 +0100, Sylvain Munaut wrote:

> Hello,
>
>
> I'm wondering if there is any advantage on using a separate regulator for
> the Vccaux.
>
> I already have 2.5V generated on the board by a switching regulator for
> the multiple DDR chips and for some Vccio of the spartan 3. Now, with the
> new TI regulator that has 3 regulators in a chip, I can have a separate LDO
> to power just the Vccaux portion. Is it useful to have it separate ?
>
>
> Thanks,
>
> Sylvain

Vccaux powers the DCMs so having a separate power supply should help to
reduce clock jitter. If you have a separate supply available I'd use it.

All,

A separate power supply for Vccaux which powers the DCM delay lines, IO
predrivers, and some bias circuits is a good idea to reduce system jitter.

Switching regulators are just fine to use in this application. As long
as there is no more than a 10 mV droop in a 1 ms period (that would be a
really slow switcher -- 1 KHz????). Variations of hundreds of mV at a
fast rate are smoothed out by the DCM's filtering (typical of a
switcher). It is a sudden droop followed by a slow recovery that
affects the DCM, as it can not track fast enough to correct for the push
out in delay of the delay lines. This primarily affects the falling
edge of the clock coming out, and the Clock 2X output, and Clock DV
output (which all use falling edges to synthesize their signals).

Austin
(who was on the DCM design team for V2, V2P, S3, and V4, and did the
subsequent verification and characterization)

Sylvain Munaut wrote:
> Hello,
>
>
> I'm wondering if there is any advantage on using a separate regulator for
> the Vccaux.
>
> I already have 2.5V generated on the board by a switching regulator for
> the multiple DDR chips and for some Vccio of the spartan 3. Now, with the
> new TI regulator that has 3 regulators in a chip, I can have a separate LDO
> to power just the Vccaux portion. Is it useful to have it separate ?
>
>
> Thanks,
>
> Sylvain

thanks for the information, just planning on making a pcb with a spartan3 on
it (my first own pcb) ...

kind regards,

Y

"Austin Lesea" <[email protected]> wrote in message
news:cnig6f$[email protected]..
> All,
>
> A separate power supply for Vccaux which powers the DCM delay lines, IO
> predrivers, and some bias circuits is a good idea to reduce system jitter.
>
> Switching regulators are just fine to use in this application. As long
> as there is no more than a 10 mV droop in a 1 ms period (that would be a
> really slow switcher -- 1 KHz????). Variations of hundreds of mV at a
> fast rate are smoothed out by the DCM's filtering (typical of a
> switcher). It is a sudden droop followed by a slow recovery that
> affects the DCM, as it can not track fast enough to correct for the push
> out in delay of the delay lines. This primarily affects the falling
> edge of the clock coming out, and the Clock 2X output, and Clock DV
> output (which all use falling edges to synthesize their signals).
>
> Austin
> (who was on the DCM design team for V2, V2P, S3, and V4, and did the
> subsequent verification and characterization)
>
> Sylvain Munaut wrote:
> > Hello,
> >
> >
> > I'm wondering if there is any advantage on using a separate regulator
for
> > the Vccaux.
> >
> > I already have 2.5V generated on the board by a switching regulator for
> > the multiple DDR chips and for some Vccio of the spartan 3. Now, with
the
> > new TI regulator that has 3 regulators in a chip, I can have a separate
LDO
> > to power just the Vccaux portion. Is it useful to have it separate ?
> >
> >
> > Thanks,
> >
> > Sylvain