capacitor choices

[Fritz Oppliger]

I am upgrading a circuit I inherited... a controller with an ADC to
measure various DC voltages (0 - 2.5VDC).
The analog nets are liberally sprinkled with 1uF Tantalum capacitors to
ground , obviously trying to dampen noise. My questions:

why polarized?
Why tantalum?
Why 1uF?

I am trying to migrate to SMD technology and find to my surprise that SMD
TA caps are rather expensive (unless I am reading the Mouser catalog
wrong). So I am looking for possible (cheaper) alternatives or convincing
justifications.
The original board had serious digital to analog noise problems (no single
ground point, noisy ADC Vref) so I am guessing that all these caps may
not be needed at all or could be changed to something tame and cheap.
Plus, I'm smoothing in SOFTware ;-)

 

[John Popelish]

Fritz Oppliger wrote:

I am upgrading a circuit I inherited... a controller with an ADC to
measure various DC voltages (0 - 2.5VDC).
The analog nets are liberally sprinkled with 1uF Tantalum capacitors to
ground , obviously trying to dampen noise. My questions:

why polarized?
Large values are smaller than nonpolarized types.

Why tantalum?

Low leakage, high reliability if properly applied. Designer may have
thought that it increased his prestige if his circuit used high priced
parts.

Why 1uF?

I am quite sure the designer pulled the value out of his uh.. oh never
mind.

I am trying to migrate to SMD technology and find to my surprise that SMD
TA caps are rather expensive (unless I am reading the Mouser catalog
wrong).

They are spendy.

So I am looking for possible (cheaper) alternatives or convincing
justifications.
The original board had serious digital to analog noise problems (no single
ground point, noisy ADC Vref) so I am guessing that all these caps may
not be needed at all or could be changed to something tame and cheap.
Plus, I'm smoothing in SOFTware ;-)

You should first investigate what can be done with good layout and an
extra board layer or two, if the board is just two sided. Then, after
layout related problems have been brought under control, you can think
about circulating currents and how to supply them locally with
capacitors and what values would be appropriate. Ceramic caps are
available in sizes and with competitive costs for all but the larger
tantalums. And a combination of an aluminum electrolytic and ceramic
may exceed the performance of single tantalums at lower total cost.

--
John Popelish

 

[John Jardine]

John Popelish <jpopelish@rica.net> wrote in message
news:4112B8F1.2CFBF06@rica.net...

Fritz Oppliger wrote:

Why 1uF?

I am quite sure the designer pulled the value out of his uh.. oh never
mind.

--
John Popelish

ROFL!. John, you've just made my day
regards
john

 

[colin]

"John Jardine" <john@jjdesigns.fsnet.co.uk> wrote in message
news:ceuj5b$m5k$1@news7.svr.pol.co.uk...

John Popelish <jpopelish@rica.net> wrote in message
news:4112B8F1.2CFBF06@rica.net...
Fritz Oppliger wrote:

Why 1uF?

I am quite sure the designer pulled the value out of his uh.. oh never
mind.

--
John Popelish

ROFL!. John, you've just made my day
regards
john

out of his hat ?

having one large area ground plane on a pcb isnt always the optimum
solution, if you have noisy part of the circuit close by, noisy ground
currents can propgate noise voltage gradients through the ground plane
(albeit rather small ). but this can cuase problems wich are hard to get rid
of with any amnount of decoupling capacitors.

in order to get round this ive found it better to have an individual area of
gnd plane for the smal siugnal stuff that conected to the rest at one point,
of course it becomes important to select the right place to make the
conection, but it stops the voltage gradient wich u can actualy see on a
scope with a coax conected to a .1 "header" used as a probe.

Colin =^.^=

 

[Fritz Oppliger]

John Popelish wrote:

You should first investigate what can be done with good layout

Amen to that!

and an
extra board layer or two, if the board is just two sided. Then, after
layout related problems have been brought under control, you can think
about circulating currents and how to supply them locally with
capacitors and what values would be appropriate. Ceramic caps are
available in sizes and with competitive costs for all but the larger
tantalums. And a combination of an aluminum electrolytic and ceramic
may exceed the performance of single tantalums at lower total cost.

Thank you.
I think I will place generic largish footprints (1206) where some of these
caps used to go, but not populate them until I find there is a problem.
I am not looking forward to re-working SMD stuff... my eyesight will be
GONE after more of this staring at electronic parts catalogs!

-- On Fri, 06 Aug 2004 12:19:15 GMT, colin <no.spam.for.me@ntlworld.com>
wrote:

having one large area ground plane on a pcb isnt always the optimum
solution, if you have noisy part of the circuit close by, noisy ground
currents can propgate noise voltage gradients through the ground plane
(albeit rather small ). but this can cuase problems wich are hard to get
rid
of with any amnount of decoupling capacitors.

in order to get round this ive found it better to have an individual
area of
gnd plane for the smal siugnal stuff that conected to the rest at one
point,
of course it becomes important to select the right place to make the
conection, but it stops the voltage gradient wich u can actualy see on a
scope with a coax conected to a .1 "header" used as a probe.

Right, the evaluation board I am looking at has two distinct ground planes
and they are connected underneath the very center of the CPU (which has
the ADC built in)