Weird 'touch' problem with atmega16

[agreatstar]

I am experiencing a strange problem with atmega. The code stops working after 4-5 runs. But starts again even if I lightly touch its body(Not even the conductive part, the top black surface of controller). I have tried changing the board, but the problem persists. I can assure you that there is no flaw in the code as the same code works fine on the same board at other supply points. To resolve this I stuck a wire on controller body and connected the other end to a metal rod. This seems to solve the issue but many a time the wire loosens and problem reappears. What could be the possible reason? Could it be static charge? What could be the permanent solution over it?
Please do not suggest to run it at other supply points as it is required to run at that point only.

 

[Golden Electronics]

1.It may related to Capacitance of Track. so connect any cap to Crystal if yes then change it's value.
2.It may possible that ur Socket is not proper solder.
3.Check ground track

 

[agreatstar]

But the same board with no change works perfect at other supply points.

 

[Golden Electronics]

what do u mean by other supply point is separate supply ? It may be ur supply itself have some problem so please check it's grounding path & voltage waveform on Scope

 

[vvarlord]

i had the same problem with atmega 8
i don't know what caused it, but i used a 10uf capacitor between vcc and ground, and now it works fine
i think it's sth about the supply voltage
i used a 10v adaptor to an 7805 ic,to supply the atmega
hope this helps

 

[Golden Electronics]

Can u check adapter output. Because it may contain ripple wit high amplitude. so by connecting 10uf it may reduce. U check supply on scope which give clear idea.