Average

[ivconic]

These are two approaches trying to do the same task.
This is part of greater loop, performing several readings from analog pin.
Goal is to "stop the loop" and read analog pin until final condition is met (until i < 10) and than store the average value and continue with next instructions in loop.
I want to be SURE that all the readings (in this case 10) are done before loop continues.
Which one is better?

First approach:

loop
...

int average = 0;
int getButton()
{
  int i, z, button;
  average = analogRead(AnalogPin);
  for (int i = 0; i < 10; i++) 
  {
  average = average + (analogRead(AnalogPin) - average) / 128;
  }
   z = average;
...
loop

Second approach:

loop
...

int average = 0;
int getButton()
{
  int i, z, button;
  i = 0;
  do
   {
   average = analogRead(AnalogPin);
   average = average + (analogRead(AnalogPin) - average) / 128;
   i++;
   }
   while (i < 10);
   z = average;
...
loop

[waltr]

There isn't a difference between the For loop and the Do...while loop. They both do the same thing.
If there is a difference it is how the compiler creates the assembly code to implement these loops.

Does your AVR development app have a way to Step Through the code?
This would answer your question and ensure the code runs through 10 loop getting 10 values.

However, the average calc is wrong.
This code
average = average + (analogRead(AnalogPin) - average) / 128;"

can be written as:
average = (analogRead(AnalogPin)+ average - average) / 128;

which simplifes as

average = analogRead(AnalogPin)/ 128;

which is not the average. It is just the last analog read divided by 128.

Something to think about, average is being declared as an 'int' which is a signed integer.
The analog read value divided by 128 will be zero is the analog read value is less than 128.
Example: 100/128 = 0
I don't think you want this.

I think you want the average of 10 ADC reads. It this correct?
What is the number of bits of the ADC? 8, 10, 12?
This matters in the math if you add them and not over flow the variable.
How many bits is the "int"? Typically an "int" is 16 bit which have vales of -32768 to +32767.
This can then hold 10 12 bit sums without overflow.

If you simply want the average of 10 reads then sum 10 reads and then divide by 10.

[ivconic]

Yeah well... i made mistake by not explaining the conditions in exact primer.
Analog read value will not be zero at any point in exact case.
Because it is wired in such manner to produce 1023 when there is no changes in reading the analog pin.
There are only 6 cases (changes) to observe. 5 "ranges". 5 buttons wired on one analog pin through the resistor network.
6 possible cases; 1023 if no button pressed and the rest divided on 5 ranges.
Since analog inputs are prone to drift (without special filtering and (in button case) de-bouncing); in this particular case the certain filtering is advisable.
So the code actually is a sort of IIR filter in this particular case. Simplified though.
Average variable holds the value and builds upon the previous one.
So it is pretty correct math there, providing very good filtering with adjustable weight (128 in this case).
But from the plain mathematics perspective; formula looks a bit nonsense. And that's what tricks you. My mistake, i didn't explained it well.
It is pretty effective sort of IIR filter there, yet simplified to the bone. It works alright. I already monitored it on serial monitor to prove that.
That part of code is not under the question at all.
What i wanted to hear from others is only comparisons between "for... " and "do... while" approaches, when used in much larger loop.
Of course this is Atmega328P, compiler is Arduino IDE. Kinda tired to repeat this at all topics here.
I don't do anything else but 328P model by now.

[ivconic]

"...Kinda tired to repeat this at all topics here...."
Would be really good to divide "Programming" section of forum to several sub-sections like "Microchip", "Atmel", "Arduino" etc...
Just to separate the apples from the pears...

[waltr]

Ok, I will that another look at the IIR part of the code just to satisfy my curiosity.
And yes, some sort of filtering and/or threshold detection is needed.
So this is a 10 bit ADC.

I think I answered your main question in the first paragraph of my post.

If I remember I can refer to the 'C' language bible by Richie & Kernighan tonight on the difference of 'for' verse ' do...while'.
If you program in C this book is a must have:
https://en.wikipedia.org/wiki/The_C_...mming_Language

I do remember your are using Atmega's so assume you are using some AVR development suite (Arduino IDE).
The common definition of AVR:
"The use of "AVR" generally refers to the 8-bit RISC line of Atmel AVR Microcontrollers" which includes the Atmega328P.
Atmel is the maker
Atmega is a line of Atmel processors.

It would still be a good idea to list processor and IDE at the beginning of a thread just to ensure we know.

[ivconic]

All my topics from the thread "Programming" belongs to "Atmel-Arduino" sub-section... which does not exist!
I don't blame administrators, sometimes is too tough to sort out and lineup topics and subjects properly to where they do belong.
Too many mixed "genres".
...
I do combine C and assembly where ever i can and most importantly; where ever it makes sense.
Art of programming is actually an art of skinning a cat.
Sometimes less is more. But sometimes more is less.
100 years from now; we still won't be sure that we picked most proper way to solve the problem in code...
That's why this game is so fun!

[waltr]

Yes, I do like coding u-processors.

Thought about the 'For' verse 'do...while' and looked it up in the K&R book.

For loops that you know the number of the 'for' loop is better only because the entire loop control statement is at the beginning.
Whereas, if the condition that exits the loop is calculated inside the loop then you must use the 'do..while' loop since the test to exit the loop is at the end.

In the code you posted both do exactly the same thing and is the programmer's preference as to which one to use.
I would use the 'for' loop since we know the loop should run 10 times.

Hope that helps.

[ivconic]

Yes, I do like coding u-processors.

Thought about the 'For' verse 'do...while' and looked it up in the K&R book.

For loops that you know the number of the 'for' loop is better only because the entire loop control statement is at the beginning.
Whereas, if the condition that exits the loop is calculated inside the loop then you must use the 'do..while' loop since the test to exit the loop is at the end.

In the code you posted both do exactly the same thing and is the programmer's preference as to which one to use.
I would use the 'for' loop since we know the loop should run 10 times.

Hope that helps.

Yup!
In essence that's it.
Keep in mind than i am still learning C and C++ (actually derivative of both).
Old donkey, old school, last time i did something similar... was in Cobol and Locomotive Basic... rrrrrrrrrrrraaaaaaaaaaaaaaa!