PLEASE HELP WITH ADC PIC16F877A

[Ian.M]

That comment on line 3 of lcd.c is a lie.  HiTech never updated it when they changed the code to use the compiler's builtin delays.   I didn't edit lcd.c at all, but if you want to blank line 3 (apart from the * to keep the same line numbering), go ahead.

The zip I attached to reply #42 has four source files:
* picdem2.c - The main program, also ADC and fixed point number functions.
* lcd.c  - LCD routines.
* lcd.h  - Header for LCD routines.
* cfg16F877A.c - PIC CONFIG only, as its bad style to put #pragma config directives in a header.

Its also got the MPLAB 8 project file picdem2.mcp, which you can import into MPLAB X.  However, as MPLAB X import from 8 is rather flakey, its probably easier just to create a new empty PIC16F877A XC8 project, uzip the lot in its .X folder and add all the source files.

[neko efecktz]

IAN,
Ok will do that.
I Had a look through the code and there is a lot I don't understand but give me a few days and I will try to figure it out
I will, as you said need to configure the LCD pins for my board.
I will probably Be back about Friday.
I have 3 day week ends and work 10 hours per day.

Mean while I ordered 2 new PIC16F877A from RS components at $8.09 each
Farnells / Element 14 wanted $13.28 ea.

I'll be back an 2 or 3 days with a progress report.

BILL.

[DTJ]

Mean while I ordered 2 new PIC16F877A from RS components at $8.09 each
Farnells / Element 14 wanted $13.28 ea.

BILL.

Plus Farnell want you to pay delivery!

[Ian.M]

You might want to drop $19.22 on a five-pack of PIC16F887-I/P.  It would be easy to port your code to them, they are more versatile and less than half the price per chip.

[neko efecktz]

Good morning IAN and DTJ and any one else listening in,
DTJ, I thought I had lost you.
The new PIC16F877A chips arrived this morning at about 9:30. Only placed the order yesterday at 10:20am.
I was still asleep when they arrived.
I inserted one and it made no difference.


Ian,
The other day you asked to confirm the crystal to be 4MHz
It is labeled and checks out with frequency counter and scope to be 4MEG.  CONFIRMED

You commented on my pcb layout.
You suggested a top side ground plane.
I will be adding that to the updated design once I get this project working correctly.

When I set up the files you sent, into a new project there were red dots on every __delay_us and __delay_ms
I inserted the DELAY.H supplied with Hi-Tech samples.
The same  one is used in several projects and in each project they look the same.
Unfortunately DELAY.H has no delay ms so I had to also install DELAY.C which contains what is required
However, Why was it able to work / simulate on your system but not on mine without a delay header file.
I had to add #include "delay.h" into the code.

I will go through all of the samples and see if there is one that contains both micro seconds and milli seconds

This is the ADC initialization I have used.
from what I have read the Acquisition time should be about 20 - 25us not ms.
I changed it to 25us and the ADC is still giving the wrong results.

Can you see anything else that may be the problem.
What about  clk division factor  I have read that it is required but I don't seem to have it in my code.
I'm not even sure where it would go but I figure somewhere within the ADCON  part of void ADC_Init()
//*******************************************
void ADC_Init()
{
  ADCON0 = 0x41; //ADC Module Turned ON and Clock is selected
  ADCON1 = 0xC0; //All pins as Analog Input With reference voltages VDD and VSS
}
unsigned int ADC_Read(unsigned char channel)
{
  if(channel > 7) return 0;  //If Invalid channel selected Return 0

  ADCON0 &= 0xC5; //Clearing the Channel Selection Bits
  ADCON0 |= channel<<3; //Setting the required Bits
  DelayUs(25); //Acquisition time to charge hold capacitor      // was DelayMs(25);
  GO_nDONE = 1; //Initializes A/D Conversion
  while(GO_nDONE); //Wait for A/D Conversion to complete
  return ((ADRESH<<8)+ADRESL); //Returns Result
}
//********************************************

I am starting to work out what I need to change in your code to match up to my board with the LCD.
The way it is done is a little different to what I have worked on, only 2 projects including this one.

anyway thanks for your help
BILL.

[DTJ]

Good morning IAN and DTJ and any one else listening in,
DTJ, I thought I had lost you.

BILL.

Nah I'm still here.  Ian's the expert, I don't have much to contribute so I'm following along to learn what I can. 

[StillTrying]

ADCON0 = 0x41; //ADC Module Turned ON and Clock is selected
ADCON1 = 0xC0

I make them  Fosc/64, = the value for a 20M osc.

[Ian.M]

MPLAB X red squiggles *LIE*!  Try building the code. If the compiler gives any errors you have a real problem, otherwise its just MPLAB X being stupid.   There was a specific problem with older versions of MPLAB X and the XC8 compiler, where the toolsuite was passing different options to the Netbeans syntax checker/hilter than those it put in the makefile so the syntax checker never saw the #defines for the delay macros in pic.h so red squiggled every occurrence even though the compiler had no problem using them.

The clock division factor is set by various ADCS bits in ADCON0 and ADCON1.  Most newer devices have then grouped in a single register, but the old ones with the original 10 bit ADC are a P.I.T.A that way.   Actually, arguably they are named wrong in the datasheet as ADCS2 is the LSB!

WTF were Microchip smoking at the time ?!?

The actual clock divisor used is 2^{(ADCS1 ADCS0 ADCS2)+1}, apart from (11 x) which are the Frc clock.  I'll be expressing them with ADCS2 as the LSB henceforth.

 You've got them defined in two magic numbers:

Code: [Select]

  ADCON0 = 0x41; //ADC Module Turned ON and Clock is selected
  ADCON1 = 0xC0; //All pins as Analog Input With reference voltages VDD and VSS
If you expressed them in binary and provided a key, you'd know what was going on:

Code: [Select]

  // ADCON0: ADCS1 ADCS0 CHS2 CHS1 CHS0 GO/DONE --x-- ADON
  ADCON0 = 0b01000001; //ADC Module Turned ON and Clock is selected
  // ADCON1: ADFM ADCS2 --x-- --x-- PCFG3 PCFG2 PCFG1 PCFG0
  ADCON1 = 0x11000000; //All pins as Analog Input With reference voltages VDD and VSS
So its obvious that the ADC ADCS bits are set to (00 1) (01 1), which TABLE 11-1 in the datasheet tells us is 16 Tosc.4 Tosc and is invalid[ for Fosc>2.5MHz! You would need 8 Tosc for a Tad of 2us (must be >1.6us) so will need to could change the ADCS bits accordingly (to 0 01) as faster conversions are generally preferable if you have a lot of other processing and are busy-waiting for the ADC result. N.B. Grossly excessive Tad can cause loss of accuracy due to charge leakage. however 4ms is within the acceptable range.

Here's how I'd handle it, bearing in mind my comment about their order:

Code: [Select]

PIE1bits.ADIE=0; // not interrupt driven

ADCON1bits.PCFG=0b0000; // All analog, internal refs. See DS39582B-page 128.

ADCON0bits.ADCS1=0;  // ADCS=(01 0), Tad=8*Tosc (2us @Fosc=4MHz)   
ADCON0bits.ADCS0=1;
ADCON1bits.ADCS2=0; //  Datasheet WTF: ADCS2 is LSB !!!

ADCON1bits.ADFM=1;  // Right Justified
ADCON0bits.ADON=1; // *FINALLY* enable A2D converter

Magic numbers in your code are *BAD*.   Used once and properly documented they are forgiveable but without that documentation they are *EVIL*.  I wouldn't be surprised to find four hex digits you didn't document have wasted nearly two weeks of your time and $16 of your money.

@StillTrying,

I make them  Fosc/64, = the value for a 20M osc.

I don't think so! You obviously misread the *EVIL* *MAGIC* *NUMBERS* ! 
However you can be forgiven as its not your code, and I misread them myself earlier in this topic when I checked the O.P's code attached to reply #28. 

Edit: Thanks for the correction by Still trying - I misread the Tad setting yet again inspite of showing correctly how to decode it.  Dyslexia and Binary is a bad combo!   

[StillTrying]

I haven't looked at any of the attachments in this thread, I just got the hex values from 3 posts above, converted to binary in my head and tried to lay them over the data sheet table.

I haven't got MPLAB X so can't open that link!

Code: [Select]

Rar!  }ÔHƒ)I3,     PIC16F877A LCD.X\build\default\debug\main.p1 °ÿætÍÍ
}ÔHƒ)I3,     PIC16F877A LCD.X\build\default\debug\main.p1 °ÿætÍ

[Ian.M]

@StillTrying: Its a misnamed RAR archive.  IZARC (freeware) opens it just fine.   Inside it is a folder with the source + a lot of MPLAB X crap.  However for your convenience, I'll attach it as a real ZIP file containing a MPLAB 8 ZIP packged project and no MPLAB X crap.
To double check, I just opened Excalibur 2.0, put it in Comp Sci mode entered:
   [Hex]41C0[Bin]
and got:
   0100 0001 1100 0000b
which is exactly the same as I posted and decoded above.  Humans don't do mental arithmtic in binary very well . . .

[StillTrying]

Yeah, if I write the binary nibbles down I get Fosc/16 = 4 us for the ADC clk.

I'm not very good with embedded C, so only looking at small snippets, from main.c in PIC16F877A LCD.zip this bit doesn't look right.

else if (FAULT_7==0)
{
        char  Volt_Char[5], DisplayVolt, InVolt;
        ADC_Init();           
        InVolt = ADC_Read(0); //Read Analog Channel 0

[neko efecktz]

GOOD MORNING EVERYONE.    (a bad impression  of Robin Williams there.)
Ian, DTJ and StillTrying
good news
Ian,
I incorporated the ADC parts from your code and I am getting a display if 5.00v with an input of 4.94 volts
and 3.30V with an input of 3.2 volts.
There is one small problem. But I know what the problem is.
With the 5 volt supply connected across a 10k pot I am able to take the input down to zero v and display 0.00V
As I wind the pot up, the display seems to be showing near to the correct voltage however the refresh rate of the LCD seems to change and it flickers badly.
Or is this the ADC taking longer to read?
So much so that the display is only readable when I turn up the contrast.

It is only at each end of the pot that the display is clear and stable, Not Flashing.

I duplicated the experiment with 2 x 2k2 resistors and the display was flashing, at mid point, slightly but much easier to read.
I added 3 more resistors to the chain Now 10.1k and the display  is about the same as using the pot.
This is most likely input impedance not giving a good read.
The display is showing approx 100mV high across most of the range.
Now all I have to do is work out how to pre-scale the input to allow for reading > 30 volts. Probably Max 35 volts ie. divide by 7.

In the final project I will need about 4 analog input for V+ V- and two for temperature of each of the heat sinks.
Each ADC will only be displaying anything if a fault occurs. ie. over temp or supply imbalance.
I will however have a switch to allow me to manually monitor voltages and temperatures.

Thank you for all of your help and suggestions.
Come over for a BBQ and some beer one day.

BILL. (Much Happier.)

[Ian.M]

Try 0.1uf from the ADC input pin to Vss.  It should take a lot of the 'flicker' out over the middle half of the pot's range.

[DTJ]

Bill are you updating the LCD with readings "as fast as you can"?

As Ian says the 0.1uF will clean up the signal. In addition to this perhaps try putting a delay in so the LCD only updates say every 200mS or whatever so you can see what's going on. If you still have problems with a noisy sample perhaps sample multiple times and do an average.

It's great when it works isn't it! 

[neko efecktz]

OK.
That worked.the caps didn't do a lot.
made some difference but not a cure.
The delay is perfect.
I got sooo excited about getting it all working I forgot to put the delay back in that I had in the original attempts.

I had some photos of the PCB
During various stages of construction but my phone died and had to be replaced under warranty.
Tomorrow I will take some more pictures of the completed board and put then into this feed.
I will have to clear a space and set it all up before taking pictures.
My desk is a bit of a mess.

Once again
Thank you for all of your help.

BILL.

[StillTrying]

Well I'm glad you're getting accurate readings from the ADC, - but it's not bleeding obvious to me what the final fix was.

Display flicker, other than use a delay, move the print position (only) and just update the 4 characters 'V.VV', or don't update them at all if their value hasn't changed from last time.

Am I looking at an obsolete version of main.c ?  I still can't see how you get the 10 to 16 bits of the ADC reading into a 8 bit char. Is it one of those C things that works, - until it doesn't.

else if (FAULT_7==0)
{
        char  Volt_Char[5], DisplayVolt, InVolt;
        ADC_Init();           
        InVolt = ADC_Read(0); //Read Analog Channel 0
        DisplayVolt = InVolt *2;

[Ian.M]

Must be an obsolete version of main.c.  Bill hasn't attached any code since I posted the XC8 port of the HiTech C demo code for the ADC on a PICdem 2 Plus with PIC16F877A fitted. back in reply #42.

@Bill (O.P.):  Package your project USING THE MPLAB X PROJECT MENU OPTION TO DO SO and attach it That puts us in a much better position to help you with getting it working for multiple channels and correct scaling.   I suggested much earlier that if you care about the accuracy, you'd use a 4.096V external reference rather than the PIC's 5V rail.  You'll also need to calibrate your input dividers and inverting buffer for the -ve rail.  You can either physically twiddle presets (which should be chosen to have an adjustment range only slightly greater than the wort case error due to fixed resistor and reference tolerences, or you can do the calibration is software.  The more complex way is to have a calibration mode i.e. store the factors and any offsets in EEPROM, but if you are building this as a one-off project and don't mind recompiling for your specific board, one could simply measure the voltages accurately, and calculate and hard code the calibration factors as #defined constants in a header file.

[neko efecktz]

Good morning everyone.
I have done some more mods to the program and have included some pictures and the latest version of main.c

I had to go to work just a few minutes after my last post yesterday so I didn't have time to do the pictures and tidy up the main.c

I was messing about this morning with the   #define MAXVOLTAGE  5.00and made it 500.00 a picture of the result is included.

After I set up the second line on the display and had it working, I found in the code

/* this is the maximum value of an A2D conversion. */
#define MAXVOLTAGE 5.00  // prescale here
#define ADCMAX (1U<<10) // One greater than max. count for 10 bit ADC
void init(void){
//   lcd_init(FOURBIT_MODE);
   PIE1bits.ADIE=0;   // not interrupt driven

   ADCON1bits.PCFG=0b1110; // Only AN0 as analog
   ADCON0bits.ADCS0=1;  // ADCS=0b001, Tad=8*Tosc (2us @Fosc=4MHz)
   ADCON0bits.ADCS1=0;
   ADCON1bits.ADCS2=0;
   ADCON1bits.ADFM=1;  // Right Justified
   ADCON0bits.ADON=1;   // enable A2D converter

}

I hadn't change this value but the AN1 still worked when I set    TRISA =    0b00011;
Also I noticed there are only 4 I/O in that line of code but there are actually 8 AN I/O or am I misreading the way the code is written?

Anyway my next step is to design the hardware to get the power supply connected to the inputs.
I will be using OpAmps to reduce the voltage to the right level and polarity.
Plus a 5.1v zener diode on each input for safety.
I will also be including thermal sensors and display temperature of two heat sinks and maybe the transformers.
The whole power supply will be going into a ABS 2RU rack enclosure mounted on 3mm aluminium plate.
I picked up a couple from Jaycar Electronics for about $8.50 each and a couple of 1RU for $6.50 each. Unsold Stock

Once again I would like to thank everyone that has helped me in this project with comments and suggestions
and a special thank you to Ian, without your help Ian, I was about to give up on the Whole idea of having analog.

[dannyf]

That's not the best code you could have written: when you write a piece of code, try to isolate them into modules in a logic fashion so that those modules can be used later on in other projects. For every piece of code you write, you should write them in a way that you don't need to write them in the future.

Take the latest example: I would have written them this way:

Code: [Select]

void init(void){
   lcd_init(FOURBIT_MODE); //initialize the lcd
   adc_init(); //initialize the adc
   //other functions here
}

//initialize the adc
void adc_init(void) {

   PIE1bits.ADIE=0;   // not interrupt driven

   ADCON1bits.PCFG=0b1110; // Only AN0 as analog
   ADCON0bits.ADCS0=1;  // ADCS=0b001, Tad=8*Tosc (2us @Fosc=4MHz)
   ADCON0bits.ADCS1=0;
   ADCON1bits.ADCS2=0;
   ADCON1bits.ADFM=1;  // Right Justified
   ADCON0bits.ADON=1;   // enable A2D converter

}


Put those files in the appropriate .h/.c files, like adc.h/adc.c, or lcd.h/lcd.c

So in the future, all you do is to copy those files into your new project and reconfigure them for the appropriate pins. Hit the compile and you are confident that your code will work.

Again, the only reason you are writing a piece of code is so you don't have to write them in the future. Those pieces are "investments" of yours so you can rip the benefits over and over again.

[neko efecktz]

DANNYF,
Thank you for your comments, but I didn't write the code.
It was taken from a Hi-Tech program supplied to me by Ian.
I know my code needS a little better structure but I am still learning.
I have to tidy up my LCD.H file which is working fine but I knew from the start I would need to break it down into a couple of files.

However, as I am only just starting to work with PICS and my first projects use LCD I have everything together in just two files.
I will be splitting it up at a later date.
LCD.H into LCD.H and LCD.C
and MAIN.C into MAIN.C and ADC.C or something like that.   (Modified this post. Fixed a file name mistake)
Hopefully by my next project post I will have things a little better.

Thank you for your comments.

BILL.

[dannyf]

It was taken from a Hi-Tech program supplied to me by Ian.

Doesn't really matter where you got it. Anyone, Hi-Tech included, is perfectly capable of writting shitty code, like this example demonstrates. There are tons of hd44780 code out there and you should have no trouble getting them to work for y ou.

and PIC16F877A.C into PIC16F877A.C and ADC.C or something like that.

You generally want to have a main.c,

[neko efecktz]

Point taken Danny
I dont know Why I put PIC16F877A.C  It actually is MAIN.C
Sorry about the confusion.
BILL

[neko efecktz]

Dear all,
Please find the attached zip file of my LCD program with ADC.
I have left it with max voltage set to 30v   #define MAXVOLTAGE 30.0
This actually displays as 99.99V not 99.9V as I thought it would.
If you make it to be greater than 100.00 volts it  displays 4 or 5 digits  ie.  99.99v      999.99v

So Long And Thanks For All The Fish.
I'll Be Back.
BILL.

[StillTrying]

All of the 'main.c's including the last one posted, are Last Modified 10/9/2016 15:43 which seems a long time ago!

But while there I tried to see (from its files) what the compiler does with the char/int InVolt problem when it needs the 2 bytes, I think it overwrites the first byte of int HADFAULT.

[neko efecktz]

very sorry about that.
I picked up an older version and didn't notice.
here is the latest version.
I have started to pull the files apart into a more logical set up as suggested.
Been shopping today and just found your request.
This version is working properly as far as I can tell, but needs more work to tidy it up and add more ADC inputs, to measure temperature.

once again sorry bout that.

BILL.

Sorry I forgot that I couldn't send .rar files in this forum
here is a zip file of the complete set up as of to date.

BILL.