DS1307 doesn't keep day of the month

[AnthonyRFC]

Hi all:

I'm having a rare problem with the DS1307 RTC chip, doesn't keep the day of the month when I retire the 5V.

I checked the basic: battery, pull-up resistor, tried with different libraries and nothing, the problem persist. So I think that problem could be in the code. Basically, when you start the microcontroller, the DS1307 init, then enters an infinite loop to display data (7 segments) and see if the user wants to change the time (TIMESET flag). To read the DS1307, I do it in the function mostrar() to avoid flashes on the display while the micro performs the reading.

CODE

[Ttelmah]

First, get rid of the interrupt disable/enable inside the timer interrupt. This is pointless (interrupts are already disabled in the interrupt handler). Just a waste of space.
Then you have two I2C setups. Should only be one. The driver already contains an I2C setup....

Then a worrying one. The driver uses 'reference' parameters to allow it to change the value in an external variable. You have global variables, with the same name as the reference parameters, which are then being passed as the variables for the reference parameters. This is dangerous coding, and I would not be at all surprised if this did cause problems.... The driver expects to be handed _local_ variable names, not global variables, and certainly not with the same name.

Then if you want us to understand your code, it needs to be laid out much better, and have some documentation. As it stands, it is a mess. Tell us where in the code you are actually getting to, and what is missing at this point, as the minimum starting point.

[AnthonyRFC]

Hi

First, thanks for the quick response. One of my reasons for get in this forum is to learn to make more orderly and efficient codes.

About your request, I put a synthesis of software:

- escribir (): I used it to write the 74LS164
-bin_bcd() : take the bin value to bcd
-mostrar() : multiplex the 7 segments displays and read the DS1307
-TIMER 1 : Is called every 250mS, I used to control the colon and the reading rate of the DS1307.

The rest of the program is to set the time through the buttons.

In the other hand, your recomendation is something like this?
-mth,day,hr,min (local variables into the DS1307 driver library)
-mes,dia, hora, minutos (global variables into the main program

**In the read process
mes=mth;
day=dia;
hora=hr;
minutos=min;

**In the write process (like setting the time)
mth=mes;
day=dia;
hr=hora;
min=minutos;

By the way, how I upload the Proteus simulation to the forum?

Best regards.
_________________
From Cumaná, Venezuela.

[Ttelmah]

You don't.

If this is a problem in Proteus, they we don't want to hear about it. Read the sticky post at the top of the forum. The Proteus simulator is the cause of more wasted time than any other piece of software involved with the PIC. It'll tell you 'good' software can't work, and also tell you that designs can work, that have not got a hope of functioning on a real chip. Don't waste your time with it....

[AnthonyRFC]

Ohhhh No, no,no! It's a real problem, real chip, real hardware, i was asking to upload the schematic of the circuit due to you requested more information.

So, what do you think about my proposal ?

Thanks in advance
_________________
From Cumaná, Venezuela.

[ckielstra]

At initialisation you set dow=7
Then at start of main this is overwritten by a call to ds1307_get_date(day,mth,year,dow). Now dow contains a random value that was in the DS1307
You never change dow.
When the user changes the time, you write the random DOW value back to the chip.

How do you know the chip doesn't remember the day of the month?
Nowhere in your code do you use the value.

[AnthonyRFC]

I used the "day" variable inside of the bin_bcd() and mostrar() sub functions.

So, do you think that the random value in the dow variable is the origin of the problem?

Ttelmah make me a warning advice when using the local variables of the driver in the main program.

void bin_bcd(){ //

if (TIMESET==1){
grup2=20;
grup1=year;
}
if ((TIMESET==2)||(estado==2)){
grup2=day;
grup1=mth;

}
if ((TIMESET==3)||(estado==1)){
grup2=prehr;
grup1=min;
}
copia=grup1;
dec= copia/10;
copia= copia%10;
uni=copia;
copia=grup2;
bdec= copia/10;
copia= copia%10;
buni=copia;
}

void mostrar(){
bin_bcd();
output_low(HAB4);
escribir(num[uni]);
output_high(HAB1);
if (leeds==0){
delay_ms(2);
}
if (leeds==1){
ds1307_get_date(day,mth,year,dow);
}
}
_________________
From Cumaná, Venezuela.

[ckielstra]

My mistake. First time I read your question as having problems with the Day Of Week instead of Day of month.

I tried following your code but it took too much of my time. Your code is difficult to follow.
Here are a few hints:
- Use English comments
- When you have parts consisting of more than 20 lines of code, then break them into multiple functions. Repeat until each function is maximum 20 lines.
- At the top of each function add little comment (in English) about what the function is doing. For example, for the timer interrupt it is great help when you write how often the interrupt is called each second.
- Try to use as little global variables as possible. Try to move them into the local function that uses them. This makes your code easier to read and saves memory.
- Convention is to only write constants and defines in all capital letters. Your variables like TIMESET and TS do not follow this rule.
- Instead of hard coded values like 'TIMESET==3' try to use defines or enums for better readability:

[AnthonyRFC]

Hi:

First of all, thanks for your time and help!

I couldn't understand the structure of the "enums". In the other hand, for a quick test, I erased the part of the program to set the time in the DS1307 and ran the code. The problem continues, so I keep confused about it.

Summarizing:

1.- I put global variables for the main program and local variables for the DS1307 chip.

2.- Corrected writing of random values of DOW variable

3-re-re-re-re-re-re-re checked the hardware (battery, pull-ups resistors).

Finally, a short video of the program in the real life.

https://www.youtube.com/watch?v=D6mpoEcyygQ&feature=youtu.be

Best regards from Venezuela

PS: Program with english comments and without the messy part of setting the time.

[temtronic]

I suggest a test using 'dummy data' instead of reading the RTC.
With this known data, does the display correctly show all variables?
I'd try several values especially for the 'day' variable as that's the only one not functioning properly.

your mostrar() function should be split into two functions

1) the actual display function
and
2) the test to see if you should read the RTC.
Actually I'm not sure why you need this test. Simply get the RTC data every time you update the display.

Jay

[ezflyr]

Hi,

Please post a clear, in focus picture of your hardware. I suspect you have a problem with your circuit!

John

[Ttelmah]

If he is keeping all the other clock values OK, then I doubt if there is a hardware problem.

However there is a big problem in overall 'messiness', and abusing the way that parameters are actually referenced.
The key is he needs to step back, simplify, and use one part at a time. Temtronic has suggested a good way to start 'back', and doing this would allow things to be 'narrowed down'.

[ezflyr]

Hi,

My suspicion is that the "hardware problem" is that there is no hardware, and this is a Proteus project. On everything else I agree whole-heartedly

John

[AnthonyRFC]

[AnthonyRFC]

PS: The battery and buttons are located in a different PCB.


I think, there is possibility of a bug in compiler? I have Version 4.104
_________________
From Cumaná, Venezuela.