My RTCC can't keep its time when VDD shutdown

[BeeElectronic]

Seriously there should not be any detectable ripple on the Vbat connection. 2 or 3mV at most.
-> yes I have ripple on Vbat juste enough for stop my scope with a 2.9V trig level, this is the MCU regulator
switching (between VDD and VBAT). But it's the same ripple for a VBAT BOR and normal power up, NOT more or worth.

How long is the connection?.
-> 35mm

Is the battery ground connected to the same point as the processor ground?.
-> GND Plane

How it the battery connection physically made?.
-> Through a S8411 harwin holder

Do you have any capacitor on the Vbat connection?.
-> No

[Ttelmah]

Generally, every 'supply' pin, should always have something like a 0.1uF
capacitor close to the pin. In their own recommendations for other chips
Microchip show one.
Add one.

[BeeElectronic]

I never use an internal RTCC with a PIC but regularity with STM32.
On STM32 I used ST never put a capacitor on Vbat pin.

I don't find this information on the PIC16F19176 datasheet but I tried with
a 100nF X7R and it's worst. I have Vbat BOR at nearly every power up...

[PCM programmer]

MCP79400 resets occasionally after power loss
https://www.microchip.com/forums/m1084666.aspx

[BeeElectronic]

Thanks for your time PCM programmer and Ttelmah but I will not use the internal RTCC because of the poor reliability.

[Ttelmah]

[BeeElectronic]

The last one I use is the STM32F746BET6. On the "Power supply scheme" they don't use capacitor on Vbat. On my biggest STM32 dev board, STM3240G-EVAL, they don't use capacitor on Vbat.

Do you have this application note? Is it for the new STM32?

I can do 2 power off:
- slow -> When the MCU restart I have a BOR resart cause
- Fast (disconnect the board) -> When the MCU restart I have a normal power up or a Vbat BOR (1 every 20 startup / 16 every 20 startup with a 100nF capacitor)

The last time I post to the CCS forum for a I2C problem and after a few days it appears the problem came from CCS, a bug with there library or routines. I don't know what happen
in the setup_rtc function but I think something weird...

[Ttelmah]

That is interesting. Suggests it is the period spent in the transition
(deciding whether to switch to battery or not), that is causing the issue.
Do you have the brownout enabled?. What voltage it it set to?.
Obviously it is vital that code cannot actually 'run' during the period
when the supply is going off. Do you have PUT enabled?. What period for?.
Again it is vital that things don't happen in code terms until the supply is
fully 'up'.

[temtronic]

Assuming that the 'RTCC module' in that PIC is common to others, it's hard to believe the problem is 'in the die' rather a software bug. If it was a common issue, there'd be a lot of Google hits and MC errata updates.
None of the PICs I use have the RTCC,so I can't test, but I suspect a minor code change(like timing or sequence) may resolve the issue.
It could be your power supply doesn't conform to the standard, or EMI from the on/off switch.

Jay

[PCM programmer]

My suspicion is either a hidden errata or poor board layout, or something
wrong with the power supply circuit. It's something that we can't see,
so far.

[Ttelmah]

We have seen oddities in the past when supply rails rise slower than
allowed. It is common not to realise that there is a minimum specification
for this time.

[BeeElectronic]

[BeeElectronic]

I don't think I have any problem with my power supply because this PSU is common for a lot of our projects and we have certification for it.

I can't have board layout issue because I use a 4 layers board (Signal/VDD/GND/Signal). The main supply come from a +12V battery and my
+5V (VDD) regulation is made by a LDO regulator (LM317T) with varistor/fuse/diode protection and low ESR tantalum+X7R decoupling.

[BeeElectronic]

Same without PUT...

[PCM programmer]

He didn't mean to try it without PUT. He wondered if you had it at all.

I would try it with PUT_64MS.