PIC32MZ / EC startup problems
with a new batch of our boards, which use a PIC32MZ2048ECG100-I/PF (Rev. A5), we experience start up problems. The uP does not start sporadically when powered up. The strange thing is, that, if it the uP is in this "stalled" state, it cannot be waked up / reset even by pulling the MCLR (external reset) pin low. Also the ICD3 debug tool does not recognize the uP, also it is not able to reset it.
The only thing which can wake up the uP is temporarily removing the power. Obviously the power seems to be the problem / key to the problem. When switching off the power quite a short time (100 to 200 ms) the uP can be brought into the stalled state as well as it can be started up normally (stalled - working - stalled - working ...).
We checked also the following things:
- VDD = 2.5 V is stable in stalled state
- VDD USB = 3.3V is stable in stalled state
- VDD is ramping up at power on from 0 V to 2.5 within about 1 ms nearly linear, no drops in between.
- Clock is provided by an external oscillator, signal is ok ins stalled state
- USB is used (with VUSB = 3.3 V).
- The MCLR pin is connected as shown in fig. 2-2 of the PIC32MZ data sheet.
- Some months ago we produced the same boards with a prior batch of the PIC32MZ (but also Rev. A5), and we did not see the behaviour with these boards (which we did not really check until now, but at least it seems that it occurred much less frequently, thus we have not realized it).
When reading the data sheet I noticed, that there is a requirement for ramping up the VDD voltage:In table 37-4 (DS60001191D-page 567):DC17 SVDD = VDD rise rate = 0.00004 V/us to 0.0004 V/us
This means that VDD should rise within a range of 6.2 ms to 62 ms from 0 to 2.5 V. This seems to be a quite restrictive requirement concerning a wide varity of power supplies.
Interestingly the new PIC32MZ EF Family has a much wider range for this specification:In table 37-4 (DS60001320B-page 611):DC17 SVDD = VDD rise rate = 0.000011 V/us to 1.1 V/us
(This would result in a power up time from about 227 ms to 2.27 us at VDD = 2.5 V)
This looks like that the power up behaviour of the PIC32MZ EC family is at least unfavourable, thus it has been fixed within the EF family.
Is there a known issue concerning the startup behaviour?
Are there other requirements for the startup e.g. concerning the USB power?
Any hints, what we can do?