Hi All.

I'm wondering how people debug MSP430-based systems that have external
watchdogs, e.g. one that will hit the MSP430 with an NMI if the
external watchdog (chip) isn't hit by the MSP430 within the watchdog
timeout period.

Is it feasible to single-step, etc?

Do you need to have special "while debugging" code that differs from
production code?

Do people simply disable / disconnect the watchdog when debugging?

I'd like to hear how people deal with this sort of thing ...

Thanks,

--Andrew

Hi Andrew,

aekalman wrote:
> Do people simply disable / disconnect the watchdog when debugging?

Exactly what I do. Reset pin from WDT is connected over three resistors.
Some ASCII art:

.
.         Vcc
.          |
.         47k
.          |
.WDT_OUT---+--4.7k--+--4.7k--MCU_RESET
.                   |
.                WDT_DIS
.

If WDT_DIS is left floating all works OK. However, if WDT_DIS is
connected to Vcc, reset can only be applied from MCU_RESET pin, like via
JTAG. WDT_DIS is connected to Vcc by attaching a debug-connector to the
board. Has worked great so far. This is necessary to disable a OD output
watchdog which can not be disabled in any other way, like time-window
WDT/SVS TPS3813 by TI.

Indrek

-- 
Indrek Rebane
Borthwick-Pignon OÜ, R&D
Riia 185, 51014 Tartu, Estonia
Phone: +372 7 302 641, Fax: +372 7 383 041

Hello Andrew,

> 
> Is it feasible to single-step, etc?
> 

With the external watchdog running, not really.

> 
> Do people simply disable / disconnect the watchdog when debugging?
> 

'tis how I used to do it.


> I'd like to hear how people deal with this
sort of thing ...
> 

If you use a circuit such as the CD4060 where you can hang a large cap
across the oscillator to make it slow you can test at lower speed. But
not really in single step with interruptions. After all, this little
circuit can't know how long that will take or whether the phone will
ring during that time ;-)

Regards, Joerg

http://www.analogconsultants.com/