Power-cycling reset vs JLink reset

[Nicolas Tzovanis]

We've developed a custom board based on the nRF51822-QFAA and an MPU-6050 IMU. It runs on a li-pol battery and uses S110.

I've encountered a strange problem when power-cycling the device (taking the battery out and putting it back it) that doesn't happen when I reset the device using Jlink for example.

Link to the code.

The problem is the following:

We initialize the MPU (set acceleration range, interrupts, etc) once at the beginning of the execution and when the initialization it's done, we set a global variable to "true" (that variable is initialized as false). The reason for that is to prevent more than one initialization.

When the device is reset using Jlink, the MPU is correctly initialized and the global variable set to 1. But when we power cycle the device, it looks like the global variable is not set back to it's default value (false) so the MPU is not initialized. That problem is solved if we reset the device using Jlink again.

The problem consistently appears every time we power-cycle the device and goes away when we reset it using Jlink.

I've found this other post and it seems to be the same problem.

Is there any difference between the Jlink reset and a power-cycle? I've been trying to figure this out for days.

Thanks!

[Dave_couling]

Often time with accelerometers there are required delays after a Power-cycle before you can safely write to their registers them. This is because there are internal sub-routines that occur. I've interfaced to several difference sensors which all had their own required delay after a POR. On a Reset-pin Reset this delay isn't required. Something to look into.

[RK]

There are differences between a power-down reset and a jLink reset, but not many. Both of them restart the code at the reset_handler and if your variable is correctly a static variable with default zero it will be in the BSS section and it will be initialised to zero in both cases.

So it's unlikely that's your problem.

[awneil]

Yes - the big difference between a power-cycle and a jLink (or other) reset is that the power remains connected during the reset. So there are going to be hardware things that are not affected - in particular, memory contents.

So the first thing to check would be that you're not relying upon any uninitialised variables or memory areas...

[RK]

His problem is the wrong way round for that. If it were the case the memory was not being reset when the power stayed on, then his JLink reset would fail, but the power-cycle one would work. His issue is the opposite to that, it works with JLink and fails when he power-cycles it. I think it's more likely to be what Dave mentions above, the accelerometer IS set up, however it's set up too early.

[awneil]

Fair point. (I guess the accelerometer counts as other "hardware things"...?)