Computer Doesn't Detect Custom STM32H7 PCB (USB DFU & ST-link)

[sugar0ats]

hey all,

i'm working on a team where most people have little experience with stm32s, and we've designed and manufactured our first board with the STM32H7S3I8T6 chip. we followed the data sheet extensively and did a decent amount of research on peripheral values, and after soldering the chip onto the board we checked that all soldered connections were good (especially D+ and D- on the USB bus). some observations:

1. we tried to first use the USB DFU method using the STM32Cube Programmer, but when plugging the board into the computer, no new usb connection is detected (when holding down boot button to set BOOT0 high).

2. we tried using an ST-link module from a Nucleo board to program the board with SWD, but similarly, no device was detected when programming.

3. when checking the D+, D-, and LSE/HSE oscillator data lines, probing them using an oscilloscope results in no signal / very weak noise.

i'll attach our schematic for the chip, any and all feedback is greatly appreciated. currently, our hunch is that the chip itself is fried and we're looking to replace it. thank you!

[AScha.3]

Hi,

> we've designed and manufactured our first board with the STM32H7S3I8T6 

...But did you look at rm and any "reference" board with H7S3 ? Doesn't seem so.

- vcap ...wrong.

- vdd xspi not connected, no xip ...this cpu needs it; only 64KB flash on chip, for boot...

- try again, respect minimum circuit requirement for this H7S3; maybe show circuit here, before producing it.

see nucleo circuit:   https://www.st.com/resource/en/schematic_pack/mb1737-h7s3l8-b02-schematic.pdf

[FBL]

Hi @sugar0ats 

I suspect VCAP capacitors from your screenshot VCAP1, VCAP2, VCAP3, VCAP4 → each to 0.1 µF → GND ?

For STM32H7S3I8T6, each VCAP pin must be connected to a much larger capacitor check the reference design of the boards provided as mentioned @AScha.3.

[sugar0ats]

thanks for the quick feedback, @AScha.3  @FBL ! i agree that we should've connected vddxspi to 3v3 as some forum posts have mentioned, but the reason we used 0.1uF caps on VCAP was because of a section of the datasheet that says that if you're in bypass mode (which we are) that we should use 100nF capacitors instead. however, because we're in bypass mode, we think we've been feeding 3.3 volts to VCAP when it should really be around 1.3-1.4V, the value of VCORE. 

additionally, are there any resources online that explicitly say that keeping vddxspi floating will result in this behavior? is it just implied from the fact that it's a supply pin?

thank you all so much again!

[Alexandre__Promont]

3. When checking the D+, D- data lines and the LSE/HSE oscillator, analysis with an oscilloscope results in no signal or very weak noise. 

I'm an electronics technician learning the STM32 line, and I'm also having trouble measuring the LSE/HSE oscillators with an oscilloscope. I don't know if there's a technique, but even though the microcontroller is running correctly, I can't get it to work.

[Alexandre__Promont]

I had a problem with the VCAP and had to put a capacitor larger than 1μ in an STM8.