Strange reset behavior. PIC doesn’t start.

kender · Joined: 09 Aug 2004 Posts: 768 Location: Silicon Valley

Colleagues,

I have a PIC18F4620 that refuses to start when the power is turned off and then re-applied. Vcc has a fair amount of capacitance and the voltage stays between 0.4V and 0.6V for some time. The problem is that if Vcc goes back to +5V, the PIC doesn’t start. If MCLR# is toggled, the PIC starts and operates normally. MCLR# has 47k pull-up.

A bleed resistor could be added between Vcc and ground, but I don’t want to add it because my device is battery powered. I have a brownout detection enabled, but something’s wrong.

Any suggestions or insights would be appreciated.

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

Ttelmah · Guest

Two other comments.
Use PUT, rather than NOPUT. If the supply rail rises slowly, there can be problems with the oscillator refusing to start. PUT will help in this case.
Consider adding a diode, resistor and small capacitor to the MCLR pin. If you pull the data sheet for the 18F452, and look at the section for 'reset', this shows the circuit you should use (it is illustrated in several other PIC data sheets as well). Basically, what you do, is have a small capacitor (0.01uF say), connected at one end to 0v, and the other end via a resistor to the MCLR pin (use perhaps 150R). This capacitor then has the pull up (perhaps 22KR), and a diode, to discharge it quickly when the power rail falls. This is the recommended reset circuit, if the supply takes more than a few uSec to rise to 5v. It sounds as if you have a lot of capacitance on your supply, and as such you should consider this.

Best Wishes

nostep · Joined: 04 Mar 2009 Posts: 16

What's connected to your outputs, anything? I've seen external devices hold up the processor from starting properly if they are powered and the power finds its way back to some processor pins. As suggested, the PUT timer should help.

kender · Joined: 09 Aug 2004 Posts: 768 Location: Silicon Valley

Fixed this problem by adding an external reset supervisor MAX809.
_________________
Read the label, before opening a can of worms.

Guest · Posted: Sat Jul 11, 2009 7:37 am

so basically U R saying
it WAS the lack of hardware management
of the MCLR pin - that was your problem?

kender · Joined: 09 Aug 2004 Posts: 768 Location: Silicon Valley

bkamen · Posted: Sun Jul 12, 2009 2:44 pm

Whoa whoa...

If you have ICD=TRUE, I don't think the pic will normally just run (depending on environment settings)

If you don't have the debugger hooked up and then enable the firmware to go.. You shouldn't be programming with ICD=TRUE.

When you want the system to be autonomous, you have to comment out the ICD=TRUE and then recompile and program.

Otherwise, when I use ICD mode, I'm always at the helm anyway.

So maybe this helps?

Cheers,

-Ben

BTW, I'm using a project now with an 18F4620 and it works just fine with ICD=TRUE, PROVIDED I'm ACTUALLY debugging. Otherwise I comment out ICD=TRUE for a more "production" build.
_________________
Dazed and confused? I don't think so. Just "plain lost" will do. :D