12F675 won't reprogram or erase

[modifyit]

These seems to have been discussed on here before, but I can't seem to figure out my problem.

I've got a 12F675 setup and it will program a couple of times and now it doesn't program and won't erase. I'm using MELabs Serial Programmer and trying to set it up so that MCLR is an input and the internal osc is used. From my research this seems to be the combination that starts causing issues for other people. I'm guessing I erased the OSCAL Value and BadGap presets, but I'm not sure how to get them back.

Since MCLR is set as an input I have a 5k pull-down resistor on it which is needed for my circuit. I'm not sure if that will cause an issue or not. My board is SMT so it is a little hard for me to modify, but I can try hooking MCLR to 5V through a diode to see if that helps.

The config fuses seemed to work properly with the following:

OSC = OSCINT
POR = ON
BOR = ON
MCLR = Input Pin
CP = ON
CPD = ON

I can read the dead circuit with the MELabs programmer, and all the fuses come back as above except OSC is set for LP.

From my reading it sounds like I need to forget about using the MCLR pin as an input, but if I can help it I would rather not change my schematic and board design.

I appreciate any input

PS. Does anyone know how the MELabs programmer handles the OSCAL and BandGap presets on the 12F675?

[mister_e]

Sorry if i wouldn't be of help but, i believe that it doesn't worth too much to retrieve those value while those PIC are really cheap, just throw it away and use another one. The next time just read the value in, then write it down on a paper or use liquidpaper underneath your PIC and write value there.

BUT Melanie already post something on that
http://www.picbasic.co.uk/forum/show...84&postcount=2

[modifyit]

I appreciate the suggestion, I can throw away the chip and resolder a new one on, but I want to figure out how to keep this from happening again. I have killed 2 chips already and I know every other one I program will be dead unless I change something.

Do you know if the MELABS programmer will automatically skip the OSCAL and BandGap settings? I need to program 1000s of these once i get this figured out and can't afford the time of writing down the OSCAL values of every chip.

I am not even sure the OSCAL value being deleted is the problem, as I understand it I should still be able to reprogram the chip if that were the case and the internal oscillator would just be off by a little. At this point the programmer won't let me reprogram or erase the chip.

[Melanie]

OSCAL and bandgap are unimportant in making your PIC work.

When you ERASe, all you ERASE is the program codespace and perhaps the EEPROM, the CONFIG settings are untouched.

Just write a simple LED Blink Program for Internal Oscillator and Internal MCLR (ie MCLR as an I/O (Input) pin) and burn that into your PIC. You'll probably find it works.

[modifyit]

That's what I thought Melanie but when I tried that the chip locked up and won't allow for an erase or a reprogram.

Should it matter that I have a 5k pull-down resistor on the MCLR pin since I am using it as an input? I think I also have 5k pull-downs on the ICSPData and ICSPCLK pins also.

I'm starting to think this is more of a schematic issue then a config issue.

Thanks again for any thoughts, they are much appreciated

[Bruce]

Programming in-circuit can be tricky is you have Vpp, data and clock pins tied to the application circuit.

Programming voltage on /MCLR needs to rise fast enough to prevent the program counter from incrementing. If it does not rise fast enough, your code previously programmed into the device can start executing, and it screws up the whole process. In this case, you're trying to program a PIC that's already executing code, and is no longer in program mode.

The internal RC osc makes the start-up process even faster, so it's even more critical for Vpp to rise fast.

Try lifting the pull-down resistor you have on /MCLR, and make sure you have the programming clock & data pins disconnected from any external loads.

External loads on data & clock pins will also screw up ICSP.

That should allow you to re-program & erase in-circuit without any problems assuming the PIC isn't dead.

[modifyit]

I'll give that a shot, but that basically means taking the pic out of the circuit using jumpers or I will loose 3 of the 6 GPIOs. I thought the pic should be able to handle this, I will try pulling the MCLR pull-down off first to see what that does and go from there. If the program programs the first time, should the clock and data pins really matter?

Thanks