Searching \ for '[PIC] Acceptable PS ripple?' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/microchip/devices.htm?key=pic
Search entire site for: 'Acceptable PS ripple?'.

Exact match. Not showing close matches.
PICList Thread
'[PIC] Acceptable PS ripple?'
2005\08\26@045347 by PicDude

flavicon
face
When using  DC-DC power supply for a PIC-based project, what is considered
acceptable (or good, or excellent) ripple voltage?  I am using National's
Webench to create a DC-DC converter and can get the ripple to a simulated
13mV with components that I can get at DK/Mouser.  I'm curious how good this
is.  I assume that the answer will vary/depend on whether I'm using the A/D
module, which I am.

Cheers,
-Neil.


2005\08\26@051048 by Wouter van Ooijen

face picon face
> I assume that the answer will vary/depend on whether I'm
> using the A/D module, which I am.

I probably can't answer you, but I can guess what others will need to
answer: Do you use the Vdd as reference, and if so, do you use
radiometric mode or an absolute input? And what are your A/D accuracy
requirements?

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu


2005\08\26@051718 by Electron

flavicon
face

At 03.58 2005.08.26 -0500, you wrote:
>When using  DC-DC power supply for a PIC-based project, what is considered
>acceptable (or good, or excellent) ripple voltage?  I am using National's
>Webench to create a DC-DC converter and can get the ripple to a simulated
>13mV with components that I can get at DK/Mouser.  I'm curious how good this
>is.  I assume that the answer will vary/depend on whether I'm using the A/D
>module, which I am.

I guess, just guess, that (given a 4.096V reference for the ADC) / 0.013, i.e.
about 1/315, will be ok up to 8 bit of resolution but not more. The advice
I would give you is to use an even worse PS if you want, but to filter very
well the ADC voltage reference, to make sure there's no injected noise.

Surely 13mV ain't going to be a problem for the digital part.. probably neither
if it was ten+ times bigger.

2005\08\26@073248 by olin piclist

face picon face
PicDude wrote:
> When using  DC-DC power supply for a PIC-based project, what is
> considered acceptable (or good, or excellent) ripple voltage?  I am
> using National's Webench to create a DC-DC converter and can get the
> ripple to a simulated 13mV with components that I can get at DK/Mouser.
> I'm curious how good this is.  I assume that the answer will
> vary/depend on whether I'm using the A/D module, which I am.

13mV ripple is well below what is needed to run a PIC reliably.  However it
could cause problems as you mention if using the A/D with Vdd as the
reference.  The solution depends on whether you need good A/D accuracy, or
whether you just need a smooth short time repeatable linear range where the
scale can vary a bit, or whether +-3 LSB of noise is a problem in the first
place.


*****************************************************************
Embed Inc, embedded system specialists in Littleton Massachusetts
(978) 742-9014, http://www.embedinc.com

2005\08\26@093430 by Mike Hord

picon face
> I guess, just guess, that (given a 4.096V reference for the ADC) / 0.013, i.e.
> about 1/315, will be ok up to 8 bit of resolution but not more. The advice
> I would give you is to use an even worse PS if you want, but to filter very
> well the ADC voltage reference, to make sure there's no injected noise.

? If using a 4.096V reference, what difference does PS ripple make?  A
good reference ought to be producing a cleaner signal than the power
supply provides.

Mike H.

2005\08\26@094632 by Robert Young

picon face


> > I guess, just guess, that (given a 4.096V reference for the ADC) /
0.013, i.e.
> > about 1/315, will be ok up to 8 bit of resolution but not more. The
advice
> > I would give you is to use an even worse PS if you want, but to filter
very
> > well the ADC voltage reference, to make sure there's no injected noise.
>
> ? If using a 4.096V reference, what difference does PS ripple make?  A
> good reference ought to be producing a cleaner signal than the power
> supply provides.
>
> Mike H.

Google for PSR (Power Supply Rejection Ratio).  National Semi and Maxim have
pretty good discussions of the topic (among the first few hits on Google).

Rob

2005\08\26@095407 by Wouter van Ooijen

face picon face
> ? If using a 4.096V reference, what difference does PS ripple make?  A
> good reference ought to be producing a cleaner signal than the power
> supply provides.

AFAIK the OP did not yet state whether he uses an external reference +
absolute input (OK), or Vdd reference + radiometric input (also OK), or
one of the other combinations (maybe not OK).

Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu


2005\08\26@100946 by Electron

flavicon
face
At 08.34 2005.08.26 -0500, you wrote:
>> I guess, just guess, that (given a 4.096V reference for the ADC) / 0.013,
>i.e.
>> about 1/315, will be ok up to 8 bit of resolution but not more. The advice
>> I would give you is to use an even worse PS if you want, but to filter very
>> well the ADC voltage reference, to make sure there's no injected noise.
>
>? If using a 4.096V reference, what difference does PS ripple make?  A
>good reference ought to be producing a cleaner signal than the power
>supply provides.

I've seen people getting 4.096V (which is a nice voltage as a reference)
using a trimmer and a multimeter. (no, I swear, *I* never did such a thing).

Of course there are integrated circuits too, but "4.096V" doesn't automagically
imply an IC. ;)


>Mike H.
>
>-

2005\08\26@113158 by PicDude

flavicon
face
Sometimes Vdd is a reference only if the sensor is ratiometric.  But not sure
if the sensor will track it's output (to the PIC) in close enough phase with
the rippling 5V line that powers it.  The official accuracy spec is "as good
as I can get it".

Cheers,
-Neil.


On Friday 26 August 2005 04:12 am, Wouter van Ooijen scribbled:
> I probably can't answer you, but I can guess what others will need to
> answer: Do you use the Vdd as reference, and if so, do you use
> radiometric mode or an absolute input? And what are your A/D accuracy
> requirements?


2005\08\26@113724 by PicDude

flavicon
face
On Friday 26 August 2005 05:17 am, Electron scribbled:
> I guess, just guess, that (given a 4.096V reference for the ADC) / 0.013,
> i.e. about 1/315, will be ok up to 8 bit of resolution but not more. The
> advice I would give you is to use an even worse PS if you want, but to
> filter very well the ADC voltage reference, to make sure there's no
> injected noise.

I use 8 bits at times, and lately have been using 10 bits more.  But I average
a number of samples -- usually around 100 or so (based on the math required
-- if my final result needs to be multiplied by .7 for example, that's
179/256, so I would average 179 samples into 3 8-bit registers, then discard
the lowest bit for an effective divide-by-256.  If that were too slow, I'd
change to 90 samples and divide by 128).


> Surely 13mV ain't going to be a problem for the digital part.. probably
> neither if it was ten+ times bigger.

This is also what I was looking for -- is there a known (ballpark) threshold
for this?

Cheers,
-Neil.


2005\08\26@144028 by PicDude

flavicon
face
On Friday 26 August 2005 06:33 am, Olin Lathrop scribbled:
> PicDude wrote:
> 13mV ripple is well below what is needed to run a PIC reliably.

Is this spec published anywhere?  IE: the threshold of acceptable ripple for
the digital sections to run properly?


> However it
> could cause problems as you mention if using the A/D with Vdd as the
> reference.  The solution depends on whether you need good A/D accuracy, or
> whether you just need a smooth short time repeatable linear range where the
> scale can vary a bit, or whether +-3 LSB of noise is a problem in the first
> place.

For a ratiometric sensor and Vref, I guess I can calculate the effect in
number of bits of accuracy lost.  But...

Assuming a fixed Vref (4.096V or 2.5V) and a non-ratiometric sensor.  Does the
ripple affect the A/D then?  I am sure it would at some level, but not sure
if it would at the same level as the digital circuitry, or otherwise.

Cheers,
-Neil.



2005\08\26@152043 by Gerhard Fiedler

picon face
PicDude wrote:

>> 13mV ripple is well below what is needed to run a PIC reliably.
>
> Is this spec published anywhere?  IE: the threshold of acceptable ripple
> for the digital sections to run properly?

I never went to those extremes, but in principle the PIC should be happy as
long as the supply voltage stays within the specified Vdd limits (and above
other thresholds, which usually are lower) -- which depend on your specific
type and osc mode (and maybe a few other criteria).

Gerhard

2005\08\26@164042 by PicDude

flavicon
face
On Friday 26 August 2005 08:43 am, Robert Young scribbled:
> Google for PSR (Power Supply Rejection Ratio).  National Semi and Maxim
> have pretty good discussions of the topic (among the first few hits on
> Google).


Interesting -- never thought of looking up PSRR for this, as I had only dealt
with it ages ago when dealing with op-amps.  This actually brought up a bunch
of good info.  One hole that remains is where is the ripple limit for the
digital side of the PIC.  This particular article
(http://www.commsdesign.com/design_corner/showArticle.jhtml?articleID=12804251)
is quite interesting, but with my scope on the blink, I can't find out for
myself right now, so still looking for a formal spec.

Cheers,
-Neil.



2005\08\26@165013 by PicDude

flavicon
face
Both.  If I use a reference that is proportional to the power supply voltage
(internal reference or voltage divider to external Vref input), it is usually
for a ratiometric sensor.  Example is a resistive sensor that I make half of
a voltage divider to the A/D input, but also voltage-divide the power supply
to get Vref+.

If the sensor is absolute/fixed output, I use a fixed Vref (such as the
MCP1541).

Cheers,
-Neil.


On Friday 26 August 2005 08:56 am, Wouter van Ooijen scribbled:
{Quote hidden}

2005\08\26@165125 by PicDude

flavicon
face
On Friday 26 August 2005 10:10 am, Electron scribbled:
> I've seen people getting 4.096V (which is a nice voltage as a reference)
> using a trimmer and a multimeter. (no, I swear, *I* never did such a
> thing).

Sure it wasn't -- it's always "A friend..." isn't it? :-)

Cheers,
-Neil.


2005\08\27@031833 by Electron
flavicon
face
At 15.55 2005.08.26 -0500, you wrote:
>On Friday 26 August 2005 10:10 am, Electron scribbled:
>> I've seen people getting 4.096V (which is a nice voltage as a reference)
>> using a trimmer and a multimeter. (no, I swear, *I* never did such a
>> thing).
>
>Sure it wasn't -- it's always "A friend..." isn't it? :-)

Only this time, really. O;-)



More... (looser matching)
- Last day of these posts
- In 2005 , 2006 only
- Today
- New search...