I'm programming my PIC controller on board...After programming it, the circuit works fine for 1 month but the program misfunctions after that. Can anyone suggest a good on circuit hardware for 18f4520 controller?

 

I very much doubt the flash program is changing... Someting else is wrong.. Your circuit or your software.

I predominately use a pickit 3 to program my chips, but it will be no different with " in circuit" if the program writes and verifies the program won't change.. Maybe your variables are being manipulated wrong.. Software bugs are horrible.. A 0 in one of your divisions can halt operation..

 

It's hard to say what is going on with no details - It's most likely either a hardware or software bug

 

the circuit works fine for 1 month but the program misfunctions after that.

Is it powered continuously during that month?
Does switching off and on 'fix' it?
Does the program use non-volatile memory?

 

Is it powered continuously during that month?
Does switching off and on 'fix' it?
Does the program use non-volatile memory?

No, switching it Off and On doesn't fix it... Its not powered continuously.. May be 12 hours a day. I'm adding the schematic.. Kindly offer your views.

 

Your direct drive to TR1 is most likely the problem ... you need to buffer it with a transistor to drive the transformer.

Even though you have a diode across TR1, the forward voltage across that diode is 1.1 V .... so your injecting a negative voltage into the substrate of the PIC. The PIC has protection diodes built into the I/O's but not designed for the back emf currents produced by TR1. Over time, this will eventually kill the PIC.

A schottky diode would be better across TR1, but ideally you should buffer the I/O with a transistor.

 

Does reprogramming it fix the problem, or do you have to replace the chip?

 

Does reprogramming it fix the problem, or do you have to replace the chip?

reprogramming makes it work again.

 

Your direct drive to TR1 is most likely the problem ... you need to buffer it with a transistor to drive the transformer.

Even though you have a diode across TR1, the forward voltage across that diode is 1.1 V .... so your injecting a negative voltage into the substrate of the PIC. The PIC has protection diodes built into the I/O's but not designed for the back emf currents produced by TR1. Over time, this will eventually kill the PIC.

A schottky diode would be better across TR1, but ideally you should buffer the I/O with a transistor.

Thanks. Will try.

 

Does your program write to program (flash) memory?

Bob

 

Does your program write to program (flash) memory?

Bob

Or to EEPROM memory?

 

Have you tried reading the program memory from the pic after it has failed and comparing it with the data that was programmed into it ? It may be worth doing this more than once to see if it is the same locations in program memory that have been changed.
Does any of the code intentionally write to program memory ? It could be some code that is trying to write to EEPROM writing to the wrong address. (Or the The EEPGD control bit set to the wrong state when attempting to write to EEPROM.)

Les.

 

Never thought of that. I'll try.

Have you tried reading the program memory from the pic after it has failed and comparing it with the data that was programmed into it ? It may be worth doing this more than once to see if it is the same locations in program memory that have been changed.
Does any of the code intentionally write to program memory ? It could be some code that is trying to write to EEPROM writing to the wrong address. (Or the The EEPGD control bit set to the wrong state when attempting to write to EEPROM.)

Les.

 

Or to EEPROM memory?

It's writing to both

 

It's writing to both

So as per post #12, when it stops read flash and EEPROM and look for any errors. It is likely that the problem lies with the code doing that writing.

 

It's writing to both

99.9% it is a software bug then.

Bob

 

Here's one I did..

I used flash as a "kinda" dataloger. This was on a pic32.. I forgot to loop my "log" counter when It maxed out... When the 16 bit counter reached max, you have to re-point to the start... It went to 0 and wiped out quite a few bytes of code... I had to do what you just did.. Pitfalls of programming...

 

Reading and Writing to flash and EEPROM in the midst of back EMF injecting negative voltage into the substrate?!! I still think following my suggestions in post #6 will do wonders.

 

Thanks friends, will revert with feedback once I've tried all the suggestions given by you.