Bricked Sparkfun nRF52832?

RE: Bricked Sparkfun nRF52832?

cswanson420 — Mon, 24 Apr 2017 02:10:54 GMT

The device suffered from some trauma that caused Pin27, the pin used for RX for a UART DFU, to remain high or 3.3V. I was able to use a modified Blinky program to toggle an LED on this pin to witness that the LED always remained lit. On a "good" device, the LED flashed properly.

Thus, although I can flash the device with the NRF52_DK using nrfjprog, anything dependent on P26 such as the nrfutil program using UART the device is impaired.

RE: Bricked Sparkfun nRF52832?

cswanson420 — Sun, 09 Apr 2017 16:00:16 GMT

Thanks for the input. The insistence on the UART is the setup I had just settled on. I use MacOS and I was able to get a script to compile any SDK example and flash it to the device using the nrfutil program. I am new to this type of activity, so I was elated that things were going so well.

What is fantastic about this experience is I have a much deeper understanding than what the UART method would have brought. This is important as when I move onto my own circuit board using the nRF52832, I will need to program it somehow and this is where I shall be blessed to have had this experience.

I anticipate I shall come to understand the system better to be able to explain with certainty why the board is exhibiting the behaviour it is.

Until then, thank you.

RE: Bricked Sparkfun nRF52832?

endnode — Sun, 09 Apr 2017 08:10:50 GMT

This looks promising! I still doubt that you would "damage" some GPIO port and it's now impossible to use UART on it. I would also be very careful to attempt writing whole FICR and UICR regions, there are only few registers which are user (app) specific, rest should be managed by the chip without interference. Is there any small chance that Sparkfun loads some diversified data so FW works only if it matches S/N of the chip (8-byte string in FICR region)? Isn't there somewhere source code to Sparkfun bootloader (or entire pre-loaded FW)?

Another point is that Nordic tools are multiplatform but I've always used Windows tools (nRFgo Studio and NRFJPROG from NRFTOOLS package) to recover/erase and flash nRF5x chips. Alternatives like NRFUTIL and OpenOCD/pyOCD are known to behave slightly differently...

Another one: why insisting on proprietary UART bootloader if you can use SWD programming?

RE: Bricked Sparkfun nRF52832?

cswanson420 — Sun, 09 Apr 2017 04:04:52 GMT

Looks like the device can not be programmed using the UART Sparkfun Beefy 3 FTDI Basic Breakout. However, I can program it using the J-link programmer on the nRF52 DK. This makes me consider I did affect the hardware negatively perhaps just the UART device.

Thanks for all your help. It pointed me in the right direction.

RE: Bricked Sparkfun nRF52832?

cswanson420 — Sun, 09 Apr 2017 03:38:29 GMT

Almost there. Using your keen suggestion pertaining to the nRF52832 Product Specification I made copies of the code, FICR and UICR memory regions and wrote them to the affected device. Presently, the blue light is flashing as I was wanting, but, alas, the nrfutil can not flash through the UART device. I get the same error. I have verified that the nrfutil can flash a good device. So I am going to try making copies of the data ram and peripheral memory regions.

RE: Bricked Sparkfun nRF52832?

cswanson420 — Sun, 09 Apr 2017 02:32:55 GMT

I did some analysis of the nrfutil program and have determined that the utility will wait and timeout just after sending the init packet. Furthermore, the Beefy UART programmer has rx/tx leds and the tx LED (tx from the device) never illuminates. This makes me suspicious that the hardware has gotten affected by my mistake, but I am not aware of a test that I can perform to qualify or disqualify the hardware.

RE: Bricked Sparkfun nRF52832?

cswanson420 — Sun, 09 Apr 2017 02:26:31 GMT

Ah, what I am expecting it to do: The "good" board will enter a bootloader mode when a button sequence is followed. This presents the blue LED to flash in a specific pattern called the timebomb pattern. However, this does not happen on the malfunctioning board. Also, I tried to flash the device with the nrfutil and a known application that works, but I get the 'Failed to upgrade target' error. So, I am trying to get the board to a state that it was with a functioning bootloader.

I am using J-Link Commander to flash and specifically, the savebin and loadbin commands.

RE: Bricked Sparkfun nRF52832?

endnode — Sun, 09 Apr 2017 01:22:29 GMT

But actually you are right, there might be some UICR register values needed as well (if FW running on Sparkfun board uses it for storing some bootloader information or similar). You can dump UICR as well (at address 0x10001000, 0x210 bytes should be enough but maybe it will give you up to 4kB of data) and analyze it by using nRF52832 Product Specification (to filter "default" values from something what Sparkfun FW uses). I would assume that if certain FW needs some values in UICR it would have default values hardcoded and writes it to particular UICR registers if it detects them "empty" during boot sequence but developers of FW you are using might see it differently...

RE: Bricked Sparkfun nRF52832?

endnode — Sun, 09 Apr 2017 01:03:11 GMT

Yes, WFE is wait for event, standard command in main loop of BLE app which runs on top of Nordic stack (aka Soft Device).

If you "dump" full memory of 512kB into HEX and flash it to another nRF52 device (which has the same PCB layout and BOM) then it should work the same after reset. So this is strange... What exactly you run on the boards (FW) and what do you expect from it (aka what do you mean by "works well" vs. "is bricked")?

RE: Bricked Sparkfun nRF52832?

cswanson420 — Sun, 09 Apr 2017 00:45:46 GMT

So, I just tried the step command 's' and it device may be stuck in a loop.

0001B07A: FD E7 B #-0x06

0001B078: 20 BF WFE

repeat...

Looks like a branch command, but not sure what WFE is. (Wait for event?)

RE: Bricked Sparkfun nRF52832?

cswanson420 — Sun, 09 Apr 2017 00:39:14 GMT

Yes, this is the article that lead me to programming the breakout board with the nRF52 DK.

I am not clear on the layout of the MBR or the starting sequence of the nRF52832 device. I have the supporting documentation from the SDK, but that seems to rely on a deeper level of understanding regarding such things as UICR.BOOT_ADR. I must be missing these things.

I can see from J-Link Commander that the system seems to stop when PC = 0x0001B07A. I can reset, go and halt from the J-Link Commander so the system is responsive. I only made a 512KB file from the good device and flashed this to the malfunctioning device. Should I make a larger file? Also, am I wrong in thinking that copying the entire flash memory from a good device will not "restore" a malfunctioning device?

RE: Bricked Sparkfun nRF52832?

endnode — Sat, 08 Apr 2017 23:07:06 GMT

Hi, have you checked similar questions on the forum, e.g. this one? It's almost impossible to completely "brick" nRF5x chips, they should always be able to "recover" through Nordic tools like nRFgo Studio or nrfjprog with JTAG/SWD programmer (e.g. any Nordic nRF5x DK with SEGGER J-Link on board).