Warm Resets

[Ali]

I'm wondering what the concensus out there is on using the "trick" of modifing your initialization to detect a "WARM" reset versus a true POWer ON "COLD" reset.. and what the members feel is the best way to handle these occurances.


Do you always create code to determine between as WARM and COLD reset and allow the system to "recover" from a WARM reset as if nothing had happened. ?

Do you always treat any reset as a COLD reset and re-initialize everything?

Have you ever had problems with "unexplained resets to the system?

Do you use an external Reset Device on your PIC chips to avoid the problem of unexplained resets?

Would your reset approach change any if your unit was to be on 24 hours a day for months at a time and a random reset would make the entire system useless until some one checked the unit. Which this check might not come for months.

Thanks in advance for your answers/comments

[PCM programmer]

Tell us more about your proposed product. Does it run from batteries
or a wall transformer ? Does it have a voltage regulator ? What is the
part number ? What filter capacitors are used with the regulator ?
Is there an input line filter ? Does your product run in an electrically
noisy environment ? Is there powerline noise ? Is there static
electricity ? Is the unit touched by anyone ? Does it have front panel
switches ? What connections does it have to the outside world ?

[Ali]

Thanks PCM Programmer for responding.
Here are more details about the product.

The product runs from batteries.
It does have a voltage regulator, a Holtek HT7533-1 3.3V 100MA LDO REG , 10uF caps on input and output.
The PIC part is a 16F648A.
It has front panel buttons wit the little carbon disks to make the connections. Those will be touched by humans to setup, but then that's
it. The unit also have an lcd glass. And it controls a solenoid valve which
is internal to the product.
Product is used outside so little static build up should occur. The
housing is plastic.

[PCM programmer]

[Ali]

Thanks for the information...

Wish I had thought to look at that errata sheet before. This is my
first PIC and I didn't even think about it before we selected the part.
We were too worried about PRICE on this job that I started to just look at that, and nothing else.

I've used the restart_cause and found that I'm getting MCLR resets. occasionally. (Like once in a week)

I checked with ICD2 and I do have version A5 of the silicon.

I am using a two layer board.. and in the next rev I do have a substansial ground pour on it. Should have it soon to test out.

I have a .1uF cap on the VDD and GND.

I also do have #NOLVP in my fuses, and I'm not using the #separate command. And I've checked the list file and my stack is at 7.

With all this information you've provided I'm guessing that you don't like the WARM COLD reset trick at all. Am I correct?

Once I get the new boards in any suggestions on how to "cause" the problem.. instead of waiting around for weeks on end?

[PCM programmer]

[Ali]

[PCM programmer]

I think the diode (in series with the 10K) will turn out to be the problem.

Read the section on "Vpp (/MCLR)" on this page.
It's about 2/3rds of the way down the page.
http://www.newfoundelectronics.com/isp%20guide.htm
I always remember this article. I only use a 10K resistor
between MCLR and Vdd.

[Guest]

PCM:
Thanks for that great writeup. I have been using many different pics for quite some time and was not even aware of all these things. Perhaps you should label this as faqs and put it in the code library or something. Probably will be of great help to many people.

[Ali]

PCM Programmer,
Thanks for all your assistance. You;ve been most helpful.

I hope that removing the diode is all that it takes to make this sytem more reliable. Though I plan on doing a little extra testing once the new board is complete just to be sure. I plan to do some through HOT /COLD testing and hope that will flush things out. I'm going to leave one with the diode in it as a "CONTROL" I hope. Any suggestions on other possible tests that might help flush it out would be greatly appreciated.



On the original question though, you never really answered nor anyone else for that matter.

[PCM programmer]

I assume your concern is about the difference between a power-on reset
and an MCLR reset. That information is in the 16F648A data sheet.
Look in the section on Special Features of the CPU. Look at Table 14-6
and especially Table 14-7. There is a column for Power-on reset and
for MCLR reset. You can compare them and see the differences.

For example, one difference is the T1CON register. In a power-on reset
it comes up as all zeros. In an MCLR reset, it comes up unchanged.
If you're using Timer1, you will likely be calling the setup_timer_1()
function near the start of your program, so you're initializing it anyway.
It doesn't really matter what it comes up as. If you're not using Timer1,
it will come up as all zeros in both types of resets, because in the case
of an MCLR reset, it's unchanged and it was all zeros in the initial
power-up reset.

I noticed that you have Watchdog timer enabled. WDT is in the same
column as MCLR, so the same arguments hold.

So I think the answer is that you don't have to do anything special.
Any peripherals or variables that you're using will be initialized by your
own code near the start of main(). Anything that you're not using will
stay in the initial power-up reset state.

[Ali]

Actually in this instance I'm more worried about my application code.
I'm running a real-time clock in this code and upon an official powerup reset I ask the user to SET the time. After that I really don't want the
user to have to reset the time.

So the problem is.. Do I assume that any reset is BAD.. and requires my application code to be completely reset, meaning they have to reset the time.. Or do I allow an MCLR reset, to pass by without resetting my application code and requesting the user enter the time again. (Taking into consideration of course any processor registers that get reset by MCLR and initializing only them. )

The other problem is that I occasionaly get a 25 for the Normal PowerUp value of a RESTART CAUSE instead of 24. I'm not sure why that happens yet. I sent an email to CCS but have not gotten a reply just yet.

[PCM programmer]