168 Startup 'Lockup' problem

[ian23]

Posted on June 20, 2005 at 11:09

168 Startup 'Lockup' problem

[ian23]

Posted on June 09, 2005 at 08:07

Hello,

I have been experienceing problems with some units which are using the ST10F168-Q6 microproceesor. When i startup the processor, it appears to lockup -never actually initialising. This has only happened on a few units and can be hard to replicate. The problem is intermittent. When the processor 'locks up' once, usually it will continue to lockup for the next several startups. There is an external watchdog being used, which gets triggered after 150ms of not recieving a strobe signal from the processor. When the processor lockups up, even pressing the reset button will not correct the fault - a complete power down is required to remove the fault. This I have been told points to a lockup of the flash memory -which a colleague has seen previously on a lower version of the same chip family. I have checked the RSTin and clock signals to see if the clock had not stabilised in time (as mentioned in another thread in this forum). This was not the problem. I am using a straight foward 20MHz clock signal( not using a 5MHz with PLL implicated.) Initially i thought the problem was caused by a slight power dip on startup, but have now elimainated this from the list of suspects (i believe). Has anyone else experienced any simialr problems with locking up of the processor during startup? Is there a cure? As i said this does not always occur on every unit as is intermittent on units that this problem happens upon. Thank you for your help in advance Ian

[najoua]

Posted on June 10, 2005 at 10:48

Hi,

Please verify that RPD pin (pin84) is connected to VDD via a 220kohm resistor (minimum) and to ground via a capacitor of 1uF typical.(see attached document)

[ian23]

Posted on June 13, 2005 at 06:21

Hi,

Pin 84 (VPP) is connected to a 5V rail, with a 'programming key' placement available. The programming key uses the 5V supply in a 5V to 12V DC-DC convertor - setting the VPP line to 12V to enable the processor to be written to. The programming key has a decoupling capacitor on it and a 4k7 resistor. The 168 uses 12V for programming, would leaving a 5v supply on this pin cause any problems? Thank you

[najoua]

Posted on June 13, 2005 at 08:34

Hi,

In order to program successfully the ST10F168 Flash, VPP MUST be in the range 12V+-5%. Please, use the hardware in the attached document ''RPD_pin_hardware.doc'' and apply directly 12V to pin 84 (without using the 'programming key'). Regards,

[ian23]

Posted on June 13, 2005 at 08:38

Hi,

The problem is not with programming the processor, the problem is only intermittent lockup during power-up of the processor. When programming the processor I have not experienced a problem. Thanks Ian

[najoua]

Posted on June 13, 2005 at 10:51

Hi,

I suspect a reset problem due to the missing hardware on RPD pin. Please refer to the following link for more details about RPD and reset relation: http://mcu.st.com/mcu/modules.php?mop=modload&name=Splatt_Forums&file=viewtopic&topic=860&forum=5 Regards,

[ian23]

Posted on June 13, 2005 at 10:56

Hi,

I have looked into the timings on the /RStin etc, these appear to be fine. You will find attached some scope traces with regards to the rstin signal thank you [ This message was edited by: IanMilly on 13-06-2005 17:11 ]

[ian23]

Posted on June 20, 2005 at 07:32

I have not yet put the RC network on the RPD permanently. However in provisional testing this does help cure the problem. I have found that there is a voltage of 5V being applied to the RPD then the processor lockups, when this voltage is less than 5V (4.992V) then the processor works perfectly fine.

When the processor has started the level on the RPD can rise to 5V, there is no effect on the processor. When the proceesor lockup and i remove the 5v from the RPD line, the processor recovers without a problem. A RC network will help cure this problem, however it will mean a mod to my PCB board. I can however tie the RPD to ground far easier or leave it floating -no connection on the pin. Would leaving the pin floating be a viable option or tying to ground a better method?

[najoua]

Posted on June 20, 2005 at 11:09

Hello,

If interruptible power down is not used, you can connect RPD pin to ground directly or via a resistor. But in this case, the length of the reset signal must be long enough to operate correctly (port 0 needs time to be sampled correctly). If the pulse is short, the clock may not have time to start and the microcontroller may lock in an undefined state(erroneous sampling of port0). I recommend something in the ms range(1ms and more). Regards, Najoua.