If I use I2C my PIC sometimes stops work

[darkmagic]

Hi, I'm using PIC 18F67J50 (datasheet) and CCS compiler version 4.057. PIC is on the I2C bus master; two PWM drivers NXP PCA9635 (datasheet) and one serial EEPROM (24FC256) are slave devices. I was using this board for three months and I have no problems with I2C (I'm new in MCU programing), but now I'm having serious problem with I2C communication.

I compile my project (no errors, MPLAB + CCS, PICKIT2 - debugger) and program my PIC. Than I start my program (F8 Step Over) and void main(void) runs.

main.c:

[PCM programmer]

1. Try it with software i2c. Remove the FORCE_HW parameter.

2. What value of pull-up resistors do you have on SCL and SDA ?

3. Run the program in the link below, and see what i2c slave chips
it finds on the i2c bus:
http://www.ccsinfo.com/forum/viewtopic.php?t=42368&start=4
Modify the #include, #fuses, #use delay(), #use i2c(), etc., to fit your
board. Use software i2c.

[asmallri]

[darkmagic]

Both pull-up resistors are 2k2 to 3V3 and to GND there are 100p capacitors (img).

I'll remove the force_hw parameter and test it. I'll try the program in the link too.


Can you explain me why are in my ASM code this two lines?
00BAE A69E BTFSS PIR1, 0x3, ACCESS //stepping (F8) in Disassembly shows that the PIC works
00BB0 D7FE BRA 0xbae //on this two lines (cycle, loop)

In PIR1 register in bits b1 and b0 (0x) are only interrupt flags for TMR2 and TMR1.
In disassembly listing for software I2C is nothing with PIR1 register.

edit1:
1. I compile my project without force_hw and it runs good.
2. I put back force_hw parametr and it doesn't work.

[asmallri]

[Ttelmah]

1) Why are you adding 100pF capacitance to the bus?.
Normally bus capacitance is your _enemy_. It is what limits the bus speed. How long are the tracks to the chips?. The bus specifies a _total_ bus capacitance not in excess of 200pF, for resistive pull-up, at 100KHz. Given the capacitances of the chip inputs, and lines you may well be pushing over this, with your added capacitors.....
2) Is the PIC running off 3.3v?. For 3.3v I2C, 2.2K, is _high_. In fact it is right on the end of the line for legal values if your bus capacitance is up near to 200pF. The Rp (min) value for standard mode, with a bus capacitance approaching 200pF, is only about 900R. For bus loads above 200pF, switches pull up, or active pull up is _required_....
using software I2C, will help, and might get it working, since the high level input voltage is lower, and the bus may well run below 100KHz, but you really need to look again at your circuit.

PIR1.3 is the SSP interrupt flag. The ,3, is the bit number for a BTFSS test, not a mask value (referring to bits 0 and 1).....

Best Wishes

[darkmagic]

asmallri: If I measure resistors between PIC (SDA, SCL) and 3V3 net, I'll get 2170R.

Ttelmah:
1. I use capacitors because I see it in some circuit and it works perfect. I know that capacitance is my enemy but I try it with capacitors and it works.
The tracks are long about 10 cm.
2. Yes, my PIC is running off 3.3V.
I'll change my circuit. I'll get rid capacitors.

Thank you for the explanation ASM code.

[Ttelmah]

The point about the voltage questions, is that on I2C, the _slave_ device, can hold the clock line 'low' while it does things. The PIC, when operating as a master, cannot then proceed to the next I2C operation. This will happen, if the bus doesn't get up to 2.64v (for a PIC on 3.3v), when released by the master. So, you need to look at everything affecting this. Obviously capacitance will slow the rise time down, but should not stop it from (eventually) reaching the required level. Depending on the leakage currents of your devices (slave and master), inadequate pull up could cause this, or if the 3.3v line you are connecting to for the pull-ups, is at a lower voltage than the line running the PIC. You could also add the keyword 'SMBUS' to the I2C declaration, which reduces the voltage needed for a 'high', to the SMBUS specification, rather than I2C. If it then starts working, it'll be telling you 'where' the problem is - you still need to fix it for reliability....

Best Wishes

[darkmagic]

I got rid capacitors.
I changed value of pull-up resistors - now 1320 ohm.
PIC, PCA drivers, EEPROM and pull-up resistors are connect to the same 3V3 net. Before each device is switch and I can turn the supply off. Pull-up resistors are connect to +3V3_PIC net. (image)

I did small test:
I compiled my program with force_hw parameter and ran the program. It stopped. I measured SDA net and there was only 0.04 V. On SCL net was 3.21 V. After that I turned off supply for the EEPROM -> on SDA was 3.20 V, too.

I ran the same program without the EEPROM (supply for EEPROM is off) and the program was running right.

test2:
I compiled my program without force_hw and with SMBUS parameter. It ran well. Without EEPROM and with too. It ran at 100kHz and 1MHz was no problem, too.

test3:
Combination SMBUS and FORCE_HW doesn't work at all (with and without EEPROM). If I remove EEPROM from the circuit, it works well.
In test3 is behaviour quite irregularly. Sometimes it works with EEPROM, sometimes not.


edit:
I changed my EEPROM and put a new one it the circuit. It works (#USE I2C (MASTER, SDA=PIN_C4, SCL=PIN_C3, FAST=100000, FORCE_HW)
It works at 100kHz and 1MHz.


edit2:
some tests (for each test I turned supply for a whole board off and than with delay on):
#USE I2C (MASTER, SCL=PIN_C3, SDA=PIN_C4, FAST=100000,FORCE_HW)

last version of my program: new EEPROM - runs ok; old EEPROM - 2x doesn't work, 1x works

14 days old version of the programm: new EEPROM - ok, old EEPROM - 1 test it stoped, 4 tests ok

I compere old and actual asm code: img. ASM code for actual program is much larger. (old asm code, actual asm code)

[Ttelmah]

SCL is being intermittently held low by the old EEPROM.
Sounds as if the chip is faulty, and cannot always complete an operation....

Best Wishes

[darkmagic]