[PIC] 16F887 > CAN > External voltage référence of 1,1 V or 1,2

For the 2'ond time ignore the earlier Newark link, posted in error. I do not support its
conclusion. They are a distributor of parts, nothing more.

Accuracy and precision are two measures of observational error. Accuracy is how close a given set of measurements (observations or readings) are to their true value, while precision is how close the measurements are to each other.

Click to expand...

If I eliminate noise, which causes deviations from true value, then I have improved accuracy.
And yes noise causes impact on accuracy AND precision. Per above wiki definitions.

Like the last time: I asked you to focus your example on accuracy errors like, offset, gain, linearity (Read the document of your own NEWARK link)
* But you provided an example for a precision error (noise. But not offset, gain, linearity).

Click to expand...

You of course read the ap note on HW averaging reducing offsets, per your inquiry above ?

This discusses INL and DNL reduction thru a cycling and averaging technique

If you are not willing learn ... how can we expect forum members to learn?

Click to expand...

Excellent comment for both of us.

And then we have this :



Note I had an error above, by not noting that noise also impacts precision, now corrected.


Knight

Hi,

danadakk said:

For the 2'ond time ignore the earlier Newark link,

Click to expand...

This is how you do it: You post a link. Then if you don't like the information you close your eyes and ignore what you have posted before.

You ignore everything you don´t like.

Then you also have to ignore:
* digilent: https://digilent.com/reference/daq-...accuracy-precision-resolution-and-sensitivity
* the Wikipedia picture (post#15) either (also previously posted by you). Since averaging just makes the distribution graph (marked as precision) more narrow and the peak higher. But averaging does not move the peak to the left to improve the accuracy (distance to the reference value).
You are free to show how averaging moves the peak more to the left. You are free to show your mathematical skills.If you can you gain a lot of respect.

What of the below do you ignore?
* Accuaracy and precision are statistical values
* accuracy is the error between the reference value and the measured value
* precision is the sdev between multiple measurements (repeatability)

Klaus

This is how you do it: You post a link. Then if you don't like the information you close your eyes and ignore what you have posted before.

You ignore everything you don´t like.

Click to expand...

I am glad you think Newark Component Distributor an expert in applications. I would take their
original work any day over IEEE papers, University Professors,Books......NOT !

For n'th time that link post by me was and is in error.

I like this summary, where adding noise added to in-precision, added to in-accuracy,
so being a linear system the inverse holds, removing noise increases accuracy and precision.
One gets more accurate readings and the sdev of the readings (precision) shrinks as noise
is removed, noise corrupts both.

What is it you do not understand about this ?



An excellent treatment of the terms precision and accuracy, their independence and their
inter-dependency depending on.........



Knight

danadakk said:

I am glad you think Newark Component Distributor an expert in applications. I would take their
original work any day over IEEE papers, University Professors,Books......NOT !

For n'th time that link post by me was and is in error.

Click to expand...

Honestly ... are you sure you are OK? I have my doubts.

I ... never brought it up
I ... never said "they are experts"
it was your work....

Wikipedia .. also was your link (not mine) .... you give links but you are not able to discuss about the information in these (your) links.
Where is your proof that "averaging" improves "accuracy" referring to the Wikipeda chart of post#15 (Again: this wikipedia link was from you ... I just refer to your sources)

Did you read your Wikipedia article?
Some excerpts:
In other words, precision is a description of random errors, a measure of statistical variability. Accuracy has two definitions:

1. More commonly, it is a description of only systematic errors, a measure of statistical bias of a given measure of central tendency; low accuracy causes a difference between a result and a true value;

and Wikipedia says:
A measurement system can be accurate but not precise,
--> according your definition: Give an example how a measurement system can be accurate, but not precise. It´s so easy, just focus on the Wikipedia definition and give us your explanation. Make yourself trustworthy.

(My answer: An noisy signal, but with low offset, low gain error, high linearity...)

and again from your Wikipedia link. :
For example, if an experiment contains a systematic error, then increasing the sample size generally increases precision but does not improve accuracy.
--> this is exactly what "averaging does". It not only uses a single sample, but it uses a bunch of samples.. it increases the sample size..and takes many values into account.
... but averaging does not improve accuracy.

******
Again: this all is not my opinion ... I just refer to your given sources.
So instead of asking "What is it you do not understand about this ?" You either should ask "what your sources don´t understand" .. or "What you don´t understand".

How can you blame me for just using your sources?

Klaus

Input to OA, Output of OA, Output of LPF, below plots

Observations :

1) OA has its own response which impacted SDEV, eg. its pole.
2) # System Measurement samples at perfect accuracy increased. I would say that system accuracy has improved if
sdev drops. Keeping in ,mind accuracy and precision apply to systems where > 1 measurements are being made,
otherwise the concept of precision goes out the window. After all wiki and Analog Devices say this (although AD has
more to say on this) -

Accuracy and precision are two measures of observational error. Accuracy is how close a given set of measurements (observations or readings) are to their true value, while precision is how close the measurements are to each other.

Click to expand...

and Wikipedia says:
A measurement system can be accurate but not precise,
--> according your definition: Give an example how a measurement system can be accurate, but not precise. It´s so easy, just focus on the Wikipedia definition and give us your explanation. Make yourself trustworthy.

Click to expand...

Not sure but these folks talk about that, see for yourself. https://www.3d-scantech.com/understanding-measurement-precision-accuracy-and-trueness/

How can you blame me for just using your sources?

Click to expand...

When you repeatedly use the one I discussed then corrected, and notified you of same.....?

So in short I demonstrated an increased # of accurate samples and increased precision as well, occurs if averaging
is done, the sdev drops as well as a consequence.

So if I raise the number of occurring absolutely accurate samples by averaging in a dataset is that considered more accurate ?
Gee whiz, my vote is yes.

Knight

Hi,

I guees it was intentionally that one can not see any details what the diagrams show.

Now you use a low pass filter (for the first time. Sadly you can´t focus on the previous discussion). Before you talked about SDEV. SDEV uses each sample with equal weighting, a low pass filter uses a different weighting for each sample. Not that it has an effect on the accuracy according Wikipedia definition ... but it makes it impossible to compare. Before we talked about digital values, now you are in the analog world...

This jumping form one topic to the other, not staying focussed, bringing up random new informations instead answering the one question ... is just to spread confusion.

Constantly ignoring that the Wikipedia diagram shows a horizontal shift ... while I asked to refer to this....

danadakk said:

So in short I demonstrated an increased # of accurate samples and increased precision as well, occurs if averaging
is done, the sdev drops as well as a consequence.

So if I raise the number of occurring absolutely accurate samples by averaging in a dataset is that considered more accurate ?
Gee whiz, my vote is yes.

Click to expand...

"I demonstrated" --> I can´t see this
"of accurate samples" --> if you have accurate samples, then there is nothing to improve accuracy
"increased precision" --> I alsways said that averagin increases precision. ..
"the sdev drops" --> I always said so
"absolutely accurate samples" --> then there is no room to improve accuracy. Meaningless.
"is that considered more accurate" --> no. You can´t improve accuracy of absolute accurate input

******

So either you do all this intentionally ....
or you are doning thin unintentionally.
In either case I can not take you serious anymore.

Spread confusion wherever you want.
Post your links that you never read...

But I don´t want to play your games anymore.

Klaus

The LPF is a simple construct, for you, that reduces noise, an operation similar
in many respects to averaging. You can do your own homework and see online
thousands of histogram plots for ADCs and noise related issues. Most modern
DelkSig ADCs have averaging built into them for that purpose. I know, too
many subjects for you to handle per your prior comments. I am not going
to hold your hand anymore.

"I demonstrated" --> I can´t see this Clear to me now that you cant
"of accurate samples" --> if you have accurate samples, then there is nothing to improve accuracy foolish statement of the obvious
"increased precision" --> I always said that averaging increases precision. .. and accuracy as sdev drops
"the sdev drops" --> I always said so hence measurement accuracy of dataset rises, more samples are more accurate
"absolutely accurate samples" --> then there is no room to improve accuracy. Meaningless. foolish statement of the obvious
"is that considered more accurate" --> no. You can´t improve accuracy of absolute accurate input foolish statement of the obvious, never said that

You wanted some math, here I go again hand holding


But I don´t want to play your games anymore.

Knight