STM32H743 answers via JTAG but cannot connect

[Zw]

Hi everybody,

I'm trying to connect to the STM32H743IIT6 on my custom PCB via JTAG. It seems it responds, I traced it with a logic analyzer (attached). I'm not too knowledgeable with JTAG Protocol, but the 0x5253000 the H7 answers seems not to be the chip ID from the BSD file.

Consequently, I cannot connect via STProgrammer or CubeIDE. 

I checked all VDD pins, they're all ok. BOOT0 pin is Ground, NRST is pulled high, VCAP pins are 1V. Pinning is correct.

I tried it with STLINKV3SET and a SEGGER J-Link base with the same results. 

I also tried pulling NRST down before flashing with no result and I looked at the signals with an oscilloscope, they seem alright (attached). 

Does anyone have an idea? I am out of ideas of what to try. Thanks a lot in advance!

[Tesla DeLorean]

Present Schematic

Would expect VCAP voltage to be closer to 1.25V

LDO or SMPS mode?

[mƎALLEm]

No SMPS on H7 single core.

[tjaekel]

Questions (based on experience to have my ST-LINK debug working on a custom STM32U5xx board):

• what is the error message displayed by the debugger tool?
• do you provide MCU VDD on debugger pin - so that debugger can "see" what the core voltage is?
  (very important to do - otherwise not correct signal level on debug signals)
• do you have this "identification resistor" (for me 100R to GND on a debug header pin) - so that debugger can see it is connected to target board?
• Are you trying as JTAG? Some MCUs, for my understanding, support only SWD (they cannot act and change mode on/from JTAG) - do you force SWD?
• Try to lower the SWD frequency...
• try options in debugger like "connect in low power mode", "connect under reset"
• is the debugger connector (the pins) correct?

Another option to try is:

if you have USB connected - set BOOT to high and see if you can see the bootloader in STM32CubeProgrammer.

If all this fails: potentially the MCU itself is not properly powered.

What is the exact wiring of the DEBUG pins on your board (the debugger header connector and its signals)?

[Zw]

VCAP is at 1.08V

SMPS Supply with 32mVpp ripple. I attached the schematic snippets, because it would not be readable otherwise. The NRST and BOOT0 are NC in the schematic, I pulled up NRST with 10kOhm, and pulled BOOT0 down.

[Zw]

Hi everyone,

thanks for the many answers on short notice. Most things you mentioned I considered. Yet again, it was a small problem, which tjaekel post reminded me about. 

We use a 10-pin jtag plug on our boards, which is different from the STLINK pinout. The first STLINK pinout you come across on the internet is also a 10-wire, so I build an adapter for it.

The problem is, the STLINK has a 14-pin on the debugger, where a 10-wire flat ribbon is put in the middle, meaning pins 1&2 and 13&14 are NC. I needed to change my adapter to that pinout and flashing with STLINKV3SET worked (see attached and UM2448). 

This still doesnt answer why the segger j-link base does not work (pinning there is correct), or why the observed JTAG communication happens - but I can work with the STLINK, so that suffices. 

[tjaekel]

Great, that at least ST-LINK works.

Correct, I had same confusion: NUCLEO boards populate a 2x5 debug header (1.25mm pitch) but ST-LINK adapter has 2x7 pins. I have to "center" the 2x7 ribbon cable as well.

The left and right pin columns are not used. One side is intended to connect the VCP UART to the ST-LINK (coming from a real UART Tx and Rx on MCU, ST-LINK provides these as VCP UART - I do not use).

Here my ST-LINK header.
I went with the same layout as on NUCLEO boards. And I have to "center" the 2x7 ribbon cable.

BTW: on older NUCLEO boards it was possible to "hijack" the ST-LINK on these boards. A NUCLEO board could also act as debugger interface, disconnecting the onboard MCU but connecting another board via the debug header.
On my NUCLEO-U5xx board - this feature seems to be gone.