STM32 USB to I2S multi channel - Hardware part

Jump to Latest
#102
Thanks a lot.

Please correct me if I am wrong: The USB cable shield is always connected to the connector covers on both sides. On the device side the connector enclosure is always grounded for AC via capacitors. Optionally (opt-out) the DC ground connection via the FB on the device side can be broken, if needed.
 
#103
With USB-C cables shield and GND are connected in the cable so receptable shield connection has no relevance as it is anyhow shorted to ground via the cable. IMO if this approach of double-grounding works for USB-C then it can be applied to older USB cables as well regardless of the older "industry convention".
 
#105
JMF11 said:
Spent some time to finalize the PCB:
Click to expand...
Hi JMF11

Looks great!

Did you manage to connect USB to a PCB yet? Is everything working with the USB3300?

I have recently been playing around with a PCB I had made. It’s STM32F4 with USB3300. For debugging I have setup firmware to initialise the as CDC. I am seeing some odd behaviour. After plugging in to the board 20 times, only once did the operating system discover the USB device! Also when I probe the ULPI_CL line I see 60MHz but after 5s or so this reduces to something between 29Mhz and 30Mhz!

Now I am wondering if I made a fundamental error. I have given my USB3300 its own 24MHz crystal and connect the ULPI Clock from the USB3300 to the STM32 ULPI_CK pin. But I am now unsure if this pins is an input or only an output?

Also my STM32 has its own clock at 24.576MHz to be an integer of 48kHz. I am wondering if this might cause a conflict? But I though the USB peripheral was independent in terms of clocks.

Any tips welcomed!
 
#108
Mine also were hand soldered, but I removed the chips using a diy hot-air gun. I have tried first to replace the cheapest chips - USB3300 - but this didn't solve the issue so I presume they were OK.
I have checked the clock lines, there were no issues with 24 MHz or 60MHz, USB just didn't work. The other parts of the chip worked great, without issues: QSPI for W25Q128 to store TouchGFX assets, SPI for ILI9341 display, I2C4 for capacitive touch, even I2C5 for EEPROM
 
  • Like
Reactions: exlabs
#110
exlabs said:
I probe the ULPI_CL line I see 60MHz but after 5s or so this reduces to something between 29Mhz and 30Mhz!

Now I am wondering if I made a fundamental error. I have given my USB3300 its own 24MHz crystal and connect the ULPI Clock from the USB3300 to the STM32 ULPI_CK pin. But I am now unsure if this pins is an input or only an output?
Click to expand...
ULPI_CK pin is normally used as input pin. The frequency drop after 5s may be due to USB3300 entering low power mode when communication with the link (MCU) failed although in low power mode clock is normally shut down. ULPI is quite finicky about the layout. ULPI traces should be kept short and equal length. Check that there are no "cold" solder joints in MCU or USB3300 ULPI pins.

Just to rule out software issues can you show how you setup ULPI pins (usbd_conf.c) and system clock (main.c).
 
#111
Thanks for the helpful tips, yes I’ll post that. To be honest, the issue could easily be with the software side of things. All I did was use CubeIDE for all configuration. The only change I made was to flash a debug LED. This indicates the MCU seems “OK”: well the clock and the LED GPIO!

The layout is fairly good and uses short correct width traces to match the target impedance, and length matching to minimise the time of flight variation across all traces .

Probably if I was doing the layout again I might put the USB3300 directly under the MCU and use double sided assembly.
 
You must log in or register to reply here.
Top Bottom