The Cyber-Spy.Com Usenet Archive Feeds Directly
From The Open And Publicly Available Newsgroup
This Group And Thousands Of Others Are Available
On Most IS NNTP News Servers On Port 119.
Cyber-Spy.Com Is NOT Responsible For Any Topic,
Opinions Or Content Posted To This Or Any Other
Newsgroup. This Web Archive Of The Newsgroup And
Posts Are For Informational Purposes Only.
From: email@example.com (Tom Bruhns)
Subject: Re: How to estimate the accuracy of ADC IC?
Date: 10 Dec 2002 10:08:01 -0800
NNTP-Posting-Date: 10 Dec 2002 18:08:01 GMT
Accuracy does not equal precision, and neither equals resolution.
Resolution is determined simply by the number of bits in the output.
Precision is a measure of repeatablity. Accuracy is a measure of how
well the output represents the true input value. So, for example, to
say that an ADC is linear to within +/- 1 LSB and that its RMS noise
is 1/2 LSB tells you something about the precision. But if you
thought that 1000 counts represented 1.000V, and it actually
represented 1.250V, then it would not be very accurate. It's common
to make ADCs accurate by adjusting something (such as the reference
and the offset).
How do you design and manufacture an ADC that meets certain
performance standards? (In part) by understanding how the design
works and what the important parameters for each component are, and by
insuring that the components all provide the performance they must for
the design to work, and by testing the design over a wide range of
operating conditions, and by understanding what the critical "corner
case" operating conditions are, and by refining the design based on
the testing you do, and by testing at "an appropriate level" the ADCs
you manufacture. The appropriate level may involve testing certain
parameters in every part you make, and it may also involve sampling
the production to test a broader range of parameters, perhaps over a
wider range of operating parameters. (Operating parameters could
include power supply and reference voltages, clock speeds,
temperature, humidity, ... perhaps even external magnetic and electric
fields.) In parallel with all that, it's useful to monitor the
processes used to manufacture the part so that you can insure it's
being built the way you intended when you designed it. That is,
integral to the design of the part is the design of the process to
make the part (or at least understanding the process which will be
used to produce the part). The same sort of thing applies to
practically anything you want to mass produce.
As for how you achieve a higher level of precision in your particular
design, I think you need to do some hard work to understand _why_
you're not getting the performance you want. Then you'll have a clue
about what to do about it. I'd _expect_ a dual-slope converter to be
much more stable than what you describe. Is the reference stable?
Is the capacitor voltage at the end of the charging cycle as constant
as you'd expect it to be? What kind of capacitor are you using, and
what did you learn about that type of capacitor when you researched
it? (You _did_ research what the important parameters of the
capacitor are, didn't you? And you _do_ know how well yours matches
those requirements, don't you?) And the capacitor is only one of the
components in the circuit.
"Reala" <-> wrote in message news:...
> I have a task to evalute the accuracy of ADC IC.
> This a simple dual slope 10bits resolution Analog to Digital convertor
> (about 50 FETs in the ADC).
> I find that the performance of this IC is not very good, because the
> output value
> is varied if i used different reference resister and capacitor.
> Even though i use a fix resistor and capacitor, the delta (max - min) of
> 1000 readings
> is about 6 (at the output value ~= 650) in same testing condition.
> However, I need to find out the accuracy of this IC.
> What is the definition of "accuracy for ADC IC?
> resolution = accuracy?
> It is interesting that the delta increase as the output value increase.
> eg. delta = 2 for output = 100 ....so the error = 2/100 *100% = 2%?
> eg. delta = 6 for output = 650 ....so the error = 6/650 *100% = 1%?
> so..what is the precentage error?
> I check the datasheet of other ADC...it say the accurancy is +/- 1 bit..
> How can I acheive this? I think that the overall accurancy will be
> decrase by other
> components...For example, if i want to develop a multimeter by ADC
> IC...how can
> I estimate the accurancy of the multimeter?
> In the IC design level, what is the consideration to acheive a specific
> In the fabricated level, how to ensure the performance between each chip
> is gurantee?
> Thank you.
Go Back To The Cyber-Spy.Com
Usenet Web Archive Index Of
The sci.electronics.design Newsgroup