coldfire start up

[martinschaefer]

Hi all,

I have a electronic with a MFC5270. The electronic runs fine except, when tested in the assembly line.

The BDM/Debug port is contacted by needles to achieve connection to a programmer Unit.

If the needles are connected to the electronic, the µC sometimes does not start-up. This happens

with and without the programmer. It seems as if only the wires can prevent the µC from starting up.

All observations I have, is, that the µC activates the DDATA and PST lines for a short sequence.

If start-up does not work, PSTCLK goes to a 100MHz oscillation. In case of successful start-up, all lines

are quiet after this sequence.

The electronics schematic is according to Freescale's recommendations  (copy from evaluation board).

Is there any detailed information about safe start-ups and disturbances on the BDM/Debug-port?

thanks,

Martin Schaefer

[TomE · Accepted solution]

Make sure you have the correct pullup and pulldown resistors on all the necessary debug port pins.

If "just the wires" are preventing the startup then either their capacitance or (more likely) RF Noise voltages being picked up by the wires (acting like an aerial) are causing the problems.

You may need to improve the ground connections between the board and the debug pod. You may need to add shielding to the debug pod cables.

But it looks like you're either missing essential resistors on the pins or the values are too high. You can reduce their values down to the port and debugger pin drive limits. 4k7 or 10k should be appropriate.

You should be able to MEASURE the voltages on the pins with an oscilloscope, but you might find the oscilloscope loading makes it worse or better on the "sensitive" pin.

Watch out for JTAG versus DEBUG startup if that chip has a pin selecting between those modes. That pin needs to be stable before, during and after Reset.

Does your power-on reset work properly? Do you have circuitry so that the debugger can force a proper CPU Reset through its connection?

Tom