scrambled internal eeprom

[MagnumScar]

Hi all,

I have the following somewhat strange problem on a circut, using an 18F4580 and a CCS PCH c-kompiler.

The circut consists of a DC/DC konverter(24-5V) and the mcu + LED's, caps, resistors etc.

It works fine most of the time exept for one condition. If we turn power off and then turn power back on very quickly, before the supply is drained down to 0 completly(residual charge on the supply caps i would gues), there is about a 10% chance that the internal eeprom gets scrambled.

Ive looked around forums and it apears im not the only one with this problem. The things that bugs me is that i dont usualy write to the eeprom, just read. I write to the eeprom only when recieving certain CAN-bus messages, wich i dont during these tests.

I have a prototype of this circut in wich this problem doesnt happen. On the prototype tho, i use a 7805 regulator rather then the dc/dc.

Also i have both power up timer and brown out detection enabled.



Any ideas? This strangenes has me alitle worried. Scrambling an eeprom like this i thought should not happen...

[PCM programmer]

Disconnect all external circuits from the PIC. Also disconnect power
from all the external circuits. Then test if the problem still happens.
This will tell you if the problem is in the PIC (and MCLR circuit, etc),
or if it is caused by the DC-DC converter circuit.

[MagnumScar]

Id try that but the circut is already built, using SMD components no less...

On the prototype im running basically only the processor but with a 7805 regulator instead of the dc/Dc. That seems to work fine

[PCM programmer]

OK. I misunderstood your original post. I thought your PIC was
controlling a DC-DC converter. Actually, the DC-DC converter is
the power supply for the board.

What are the external circuits that the PIC is controlling ?

Also, do you have 100 nF (0.1 uF) capacitors on the PIC's Vdd pins ?

[MagnumScar]

Actualy the mcu is only connected to an adum1250 i2c isolator. on the other side of the isolator there is a MAX5842 DAC, an ADS7828 ASC and an IO extender MAX7323. The DAC outputs are amplidifed thru an LM224D op-amp.

The i2c isolator is obviusly powered from both sides. THe problem seems to be the same regardles if i power the "chip" side of the isolator. the "mcus side" is powered from the same 5 V as the mcu(obviusly).

AS mentioned the problem only happens if we turn power of and then oon again very quickly, wich makes me to belive there is somthing about the reseting that scrambles my eeprom.

Any more idea?

Thanx for the help btw

[PCM programmer]

Look at the Vdd pin on the PIC when you cycle the power. If possible,
catch the waveform with a storage scope. Then do the same thing with
a linear regulator (instead of a DC-DC converter) running the board.
Look for a glitch on the Vdd voltage.

[MagnumScar]

Thnx, ill check that, altho it seems obvious to me that there is something going on with the Vpp here. The processor does reset itself more or less, altho it seems not properly (thus the scrambling of the eeprom)

[MagnumScar]

How exactly do i turn on the brownout detction?

Im worried i didnt turn it on properly.

i have set the fuses to:

#fuses BROWNOUT, BROV46

not done anything else... this should do the trick yes?

[gjs_rsdi]

#FUSES BROWNOUT //Reset when brownout detected
#FUSES BORV46 //Brownout reset at 4.6V

[MagnumScar]

So i got my fuses set correctly then.

Cheers

[gjs_rsdi]

yes
I have built a proto board with DC/DC. have ripples specially at startup, the 7805 don't.
I solve the problem by connecting a 4.8v nicad battery in parallel
I had some problems with the oscilator(didn't check the internal EEPROM)
Try a 10uF capacitor and 10k resistor on the reset pin. It solved my problem in an other project.
powerup timer (#FUSES PUT //Power Up Timer 18F252) helped me also.
joseph

[MagnumScar]

Thx for the tip. Ill check that

I did enable the PUT and it seems to have helped a bit but not 100% ok.
Ill have a look at the dc/dc and see if that is the reason for this trouble.

Thx again

[Ttelmah]

The other question, is what is connected to MCLR?. Also have you got LVP enabled. If the answer to the latter is 'yes', then great care is needed to with the layout round RB6, and RB7. The MCLR pin, especially if used as a digital connection, should have an external diode attached, when it is not used for programming, to clamp any possible transient spikes, since this is not present internally, and signals above the supply rail on this, can trigger programming behaviour...

Best Wishes

[rnielsen]

You might want to consider a CPU supervisor attached to the MCLR pin.

[MagnumScar]

A voltage supervisor external circuit would be good, but i need something that can handle the ICSP interface applying 13,5 Volt to my MCLR pin.

Any idea of a external reset circuit that can do that? Ive tried reading microchips app note TB087 but it doesn't make any sense to me, especially the way in figure 3 the Rup is connected to Vdd/Vpp ... how does that work?

Any insight would be much appreciated