HSE oscillator does not work on Nucleo-U385RG-Q board

[Spiker]

I try to get the HSE working on a Nucleo-U385RG-Q board by the standard settings in the STM32CubeIDE, but can't get it to work. I changed the default soldering connections per the UM3062 User manual to SB19 and SB21 Off, SB20 Off, SB3 and SB4 On. I monitored the HSE by redirecting it to the MCO pin (PA8 / D7). but got no signal. I monitored other clock signals, like the external LSE and the internal high speed oscillator via the MCO pin and had no problem to get a signal out.

Though the Cube's settings should start up the HSE, it didn't, and even when I added a line of code (RCC->CR |= RCC_CR_HSEON;) to enable the HSE oscillator, I did not get anything.

I have no problem to get this working on a Nucleo-C092RC board, so I wonder whether there is something wrong with the HSE oscillator on the Nucleo-U385RG-Q or with the soldering settings recommended in the manual.

I would appreciate any feedback, because I purchased more than 100 of these boards to use in an experiment, and I do need the precision of the HSE oscillator.

[Gyessine]

Hello @Spiker 

>I did not make or break any soldering connections, so the original board is used.

If I understand you correctly, you are still using the default configuration of the board hardware to measure the HSE clock. However, the board user manual states that some changes must be made.

The use of this clock source requires the following configuration:

• Set SB19 to OFF and SB21 to ON. Only PH1 can be connected to CN7 as an I/O.
• Set SB20 to ON. MCO is connected to PH0, and R8 on the STLINK-V3EC side must be connected to provide the MCO from the STLINK-V3EC output. The resistor (R7) and capacitor (C6) can be adapted for the MCO frequency shape. Set SB20 to OFF after configuration.
• Set SB3 and SB4 to OFF. The external crystal is disconnected from PH0 and PH1.

Can you try this configuration and return to us with feedback?
Gyessine

[Spiker]

Hello Gyessine,

Thanks for getting back to me. I very much appreciate your help.

The manual was a bit puzzling to me, so I have some questions / comments.

1. Set SB19 to OFF and SB21 to ON. Only PH1 can be connected to CN7 as an I/O. In the original board both SB19 and SB21 are ON, which connects both PH0 and PH1 to CN7 (pins 29 and 31 resp. on the Nucleo-U385 board). So, I need to disconnect SB19, which I will do.

2. Set SB20 to ON. MCO is connected to PH0, and R8 on the STLINK-V3EC side must be connected to provide the MCO from the STLINK-V3EC output. The resistor (R7) and capacitor (C6) can be adapted for the MCO frequency shape. Set SB20 to OFF after configuration. In the original board SB20 is OFF, so I need to make the connection, which I will do. R8 is connected on the circuit diagram and it is also in place on the PCB. Similarly, R7 and C6 are in place on the PCB, and I do not wish to change them. Now, what puzzles me is the phrase Set SB20 to OFF after configuration. In the manual only SB20 OFF is written, so I assumed that it was a mistake left over in the editing process of the manual. If it is not a mistake, then I first need to test whether I get the HSE clock signal from the STLINK part, and then I need to break the SB20? It does not make sense to me. Isn't this a mistake in the manual? Apparently, SB20 makes the connection from MCU.MCO to PH0-OSC_IN of the U385, and SB19 makes the connection from PH0 to PH0-OSC_IN of the U385, so it kind of makes sense that both SB19 and SB20 are connected. My understanding is that MCU.MCO is PH0 of the STLINK MCU. Is that correct? 

3. Set SB3 and SB4 to OFF. The external crystal is disconnected from PH0 and PH1. These are already OFF in the original board.

Since I have plenty of these boards and I want to increase the turn-around time, I will prepare a board with SB20 set to ON, as well as a board with SB20 set to OFF. (with SB19 set to OFF in both cases)

-----------------------

Edit: I did the above and it worked.

So, summarizing: To use the external oscillator of the STLINK part, I set SB19 to OFF, SB20 to ON, and no other soldering connections were changed. Cube settings in the System Core / RCC section were "BYPASS Clock Source" for HSE, and the box for "Master Clock Output" was set so that the clock can be measured on the PA8 (D7) pin. Input frequency in the clock tree was set to 8 MHz in the box pointing to the HSE box. Other frequencies are apparently not possible, even after updating firmware. Setting a different frequency has no effect, since 8 MHz is always used by default. I selected either the HSE signal or the SYSCLK signal as input to the MCO Source Mux in the clock tree.

The statement to set SB20 to OFF in the manual should be ignored, since it is an error.