False Resets - MC9S08QG

[DSbob]

My product has three identical PCB modules:
Each module has one MC9S08QG
Each module has a connector per the BDM Tool definition
  BKGD pin 1 connected on module to BKGD on 'QG and to 100k pullup
  n/c pin 3 to PTB7 pin, n/c in module except internal pullup
  even pins connected to 3.6 volt battery and reset circuit from a forth board
  Reset pin 4 to 'QG Reset' pin 1
Each 'QG in Stop1 mode when typical false reset occurs

The reset circuit on the forth board differs a little from basic system connections:
100 nF to ground for each 'QG as recommended
from each 100nF-Reset pin node, 330 ohm series resistor to reset switch to Vss,

3300 Ohm pullup from switch to Vdd
I have tried other values including shorting the series resistors and up to 10nF||2uF for the

capacitors to ground.

The product works well enough when reset out of Stop1 mode, in a range of applications:
Most applications including transient physical shock.
Some applications in addition present a severe transient electromagnetic environment.
In rare cases false resets only after removal from either of those environments, with associations

similar to the false resets from Stop1 mode.

The typical false reset is a wakeup from Stop1 when the reset switch is not actually operated.  We

strongly associate it with a finger touch to silicone rubber; materials usually listed near opposite

ends of the triboelectric series.  We discount static effects becauss metal always connected to Vss

separates the rubber from the reset circuit.  If anything, false resets occur more in our static safe

assembly area than outside it.  False resets occur:

Quite often only one module, usually a particular one of the three modules,

often two particular modules, or

rarely all three reset at the same touch,

sometimes only after multiple touches.

Of course the reset circuit works fine when we operate the reset switch.  All three modules lamp test, run for 3 minutes, and return to Stop1 nearly in sync.  I realize my choice of termination values

might have been better and the data sheet does not anticipate 3 uCs to one reset circuit.  I hesitate

to switch 3x100nF through gold contacts of the reset switch and expected 3300 as the pullup would

improve EMI resistance.  The modules are contract made and I cannot do much with them, thus the reset components on the fourth board in case of operating environment needs.  Each module has 100 nF for power decoupling.  Increasing Cblk decoupling on the fourth board from 22uF tantalum up to 3300 uF aluminum benefits accuracy of operation in the harsh EM application and has no effect on false resets.

Has anyone had false reset problems with the 'S08'?  Should I be trying or considering something else?  Am I doing something really stupid?

Added p/n to subjecdt.

Message Edited by NLFSJ on 2009-05-06 06:59 PM

[Lundin]

In tough environments it is generally custom to not expose the reset pin at all. You may want to consider some buffer circuit, or even galvanic isolation, if you must expose this pin.

Though to me it seems odd to have the user manually doing a hardware reset with a switch connected directly to the pin. I would have implemented it as a gp I/O and then do a reset through software (by writing the wrong value to COP register). What about debouncing the switch?

[DSbob]

Yes Ludlin, as I originally said in tough physical and electromagnetic environments, after reset out of Stop1, it works well.  A reset there would disrupt functionality and render the product useless.  For example, I had to give up on a spring contact battery holder; DC environments yanked batteries out of contact.

The user reset is a wake up for 3 minutes once or more daily operation, timed out to ultra low power sleep.  (I assume for some users an on-off switch is too complicated.)

Between the RC filter and now yesterday adding software debounce of the GPI while not sleeping,  I'd say it is pretty well debounced, short of adding something like a RS-FF that cannot be turned off.

If no one has guessed, the silicone rubber is the switch cap and housing, with a nickel plated spring dome inside always touching system common terminal and strong tactile feel when pushed to contact the reset network terminal.  How touching it lightly can reset only some of the modules, who knows?

There is an air cavity between the switch body, rigid terminal holder and PCB.  I am going to put ferrite beads on one or both of the leads, and also on a test fixture to see how they hold up in the pulsed electromagnetic environment.

[BasePointer]

We lived similar problem with LC60 before. And unfortunately solved the problem by disabling reset pin functionality.

[DSbob]

Thank You for the response, BasePointer.  I have been busy elsewhere for the last month but finally got to act on it.  Of course the QG reset pin reverts to that function in Stop1 mode, so I disabled it and wrote code to software reset on polling it as a general purpose input for 240x consecutive active low samples (32ms).

Unfortunately I do not see any change in sensitivity to false reset from Stop1.

I was unable to provoke a false reset when not in Stop1, but that has always been harder than when in Stop1.