Searching \ for '[OT?] External A/D jitter' 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/io/atod.htm?key=a%2Fd
Search entire site for: 'External A/D jitter'.

Exact match. Not showing close matches.
PICList Thread
'[OT?] External A/D jitter'
1998\09\30@095732 by Dave Johnson

flavicon
face
I've got a PIC 16F84 controlling an A/D converter (Burr-Brown ADS7822P,
12 bit, micro-power) and I'm getting lots of jitter in the lower bits. I
know these things are very susceptible to line noise, and so I have
plenty of bypass capacitors in appropriate spots, but still get lots of
variation in the converted value. There's also a MAX232A in the circuit,
and I know those charge pumps are noisy, but even with that chip removed
it's still happening. Does anybody have tips on calming the poor thing
down? Or do I just need to live with it and find the median or average of
several consecutive samples?

Dave Johnson

1998\09\30@113411 by Matt Bonner

flavicon
face
Dave Johnson wrote:
>
> I've got a PIC 16F84 controlling an A/D converter (Burr-Brown ADS7822P,
> 12 bit, micro-power) and I'm getting lots of jitter in the lower bits. I
> know these things are very susceptible to line noise, and so I have
> plenty of bypass capacitors in appropriate spots, but still get lots of
> variation in the converted value. There's also a MAX232A in the circuit,
> and I know those charge pumps are noisy, but even with that chip removed
> it's still happening. Does anybody have tips on calming the poor thing
> down? Or do I just need to live with it and find the median or average of
> several consecutive samples?
>
Capacitors won't always fix circuit noise.  In fact, excess capacitance
can sometimes cause noise.  What comprises the rest of your analog
circuit?  You can hardly ever expect N noise-free bits out of an N-bit
A/D converter (even without any other analog interface circuitry).  The
better ADCs spec their resolution with consideration put towards output
noise, input notch filtering, and a handful of other things.

What conversion method does the ADS7822P use?  How many stable bit are
you getting?  Don't forget that quantisation error will give you a
"toggling" LSbit even in the most stable system.

--Matt

1998\09\30@124837 by Bob Blick

face
flavicon
face
Hi Dave,

What is the source of the analog signal? The most troubles I've had with
A/D converters nowadays has been related to high impedance analog signals.

By high impedance I mean 1000 ohms or more. seriously. Any noise that
occurs during the conversion process is deadly to accuracy.

-Bob

1998\09\30@165341 by Chris Eddy

flavicon
face
Dave;

I have been very fortunate to learn a few great tricks from old pros.  The
answer to your jitter may have alot to do with your ground system.  I found
that if any currents flow through the ground traces that service the A/D or
analog circuitry, you introduce offsets of DC and AC magnitude.  The answer
is to layout the board with painstaking attention to ground paths.  Route all
analog circuitry back to the central ground node completely independent from
all other sources.  I had one product that when a single LED turned on and
off, it added three degrees of error on the display.  I have had other
designs with poor grounding that had an old pro 'give it the touch' with the
grounding scheme modifications, and where I had noise before, I could
actually pass from one code to another in 12 bit resolution.  Bob has a good
point, try a simple battery with pot for a voltage source.

Chris Eddy
Pioneer Microsystems

Dave Johnson wrote:

{Quote hidden}


'[OT?] External A/D jitter'
1998\10\01@102916 by Dave Johnson
flavicon
face
Chris Eddy wrote:

>I have been very fortunate to learn a few great tricks from old pros.  The
>answer to your jitter may have alot to do with your ground system.
<snip>
Yes, after a little more prowling through my trusty "Art of Electronics"
book, I'm convinced that I've made lots of classic layout and grounding
blunders :-) But I'm learning fast. Some careful reworking of the layout
should fix it, or at least improve the situation markedly. I'll find out
this morning...

Matt Bonner wrote:
>What comprises the rest of your analog
>circuit?  You can hardly ever expect N noise-free bits out of an N-bit
>A/D converter (even without any other analog interface circuitry).
The analog circuit is simply an unspecified "thing" to which I must
provide 5V and Gnd, and it gives me an analog signal between 0 and 5V to
read. Of course I've been testing with one particular "thing", but I need
to make the circuit work with a whole variety of them, which will vary in
their electrical characteristics (some consume more power, etc.). Because
of that, I'm considering putting a sample and hold or something similar
between the "thing" and the converter, does that sound like a good idea?
I am loathe to increase the component count beyond what's absolutely
necessary (the finished circuit needs to be as tiny and low power as
possible), but this sounds like it might fall into that category.

>What conversion method does the ADS7822P use?  How many stable bit are
>you getting?
It's a classic successive approximation converter. The reason I chose it
is for it's VERY low power consumption (supply current well under 10uA at
1 kHz sample rate, shutdown current something like 50 nA). Right now I'm
only getting something like 7 stable bits, the lower 4 or so are jumping
all over, but as I said my initial, un-thinking circuit layout is really
bad, so I'll work on that...

Bob Blick wrote:
>The most troubles I've had with
>A/D converters nowadays has been related to high impedance analog signals.
>By high impedance I mean 1000 ohms or more. seriously. Any noise that
>occurs during the conversion process is deadly to accuracy.
Youch. I don't even know what the impedance of the signal is, and it will
vary between "things." That sample and hold, or some sort of buffer with
known characteristics, is sounding better all the time...

Thanks very much to all for the responses, I'm fairly new at this but I'm
loving it, and learning fast.

BTW, is it OK to hit the PICList with these sorts of more general
electronics questions? I figure if I'm grappling with these questions
others are too. Or is there some other mailing list that might be better?
(I'm not terribly fond of newsgroups anymore...)

Dave Johnson

1998\10\01@112931 by Matt Bonner

flavicon
face
Dave Johnson wrote:

> I'm considering putting a sample and hold or something similar
> between the "thing" and the converter, does that sound like a good idea?

Sample and holds require careful design and component selection.  Better
to use an ADC with an internal S&H (unless it's too late in the design
cycle).

> >What conversion method does the ADS7822P use?  How many stable bit are
> >you getting?
> It's a classic successive approximation converter.

Ahh...Successive approximation.  Definitely pay attention to what Chris
Eddy said. SAR ADCs are very suceptible to noise - that's why you don't
see many with over 12bits resolution.  I've tried a few high-res SAR
ADCs and they didn't work too well.  We use sigma-delta ADCs with an
adjustable notch filter.

> Right now I'm
> only getting something like 7 stable bits, the lower 4 or so are jumping

Even with careful layout, don't be disappointed if you can't get it past
10 stable bits.  ADC manufacturers seem to play games with specs - real
world circuits can't match up to the test circuits they use in the
labs.  I was really happy to come across a 24bit ADC that spec'd
*effective* resolution from 10bits to 24bits depending upon circuit
configuration (filtering, gain, etc).  My final design matched the
published specs - I calculated that I would get 18 bits with my
configuration and that's what I got.

> all over, but as I said my initial, un-thinking circuit layout is really
> bad, so I'll work on that...

Good place to start.

> BTW, is it OK to hit the PICList with these sorts of more general
> electronics questions?

Definitely (just check out the archives and you'll see a lot of general
questions).  If you get *too* far off topic, embed [OT] somewhere in the
subject line.

--Matt

1998\10\01@185047 by Tony Nixon

flavicon
picon face
Chris Eddy wrote:
> The answer to your jitter may have alot to do with your ground system.

I'd have to agree with this. I designed an accelerometer circuit which
worked quite well. Someone else wanted several of them later on, but I
had to redesign the PCB. They worked terribly, even though it was
exactly the same circuit. The original design worked ok on the
aquisition system, and all the others didn't. The only conclusion I can
come to is the board layout.

--
Best regards

Tony

Multimedia 16F84 Beginners PIC Tools.
**New Improved PicNPost**

http://www.picnpoke.com
Email spam_OUTpicnpokeTakeThisOuTspamcdi.com.au

More... (looser matching)
- Last day of these posts
- In 1998 , 1999 only
- Today
- New search...