Compute Module 3 to LAN9512 error reading from device

[nenstrom]

Currently in the process of connecting a LAN9512 to a compute module 3 with the eventual goal of cascading these devices to give me 3 ethernet ports. I have been able to confirm that the chip is running by connecting my laptop directly to the upstream USB port. In the attached images you can see that the device is recognized by windows no problem but that when I try to use the eeprom utility provided by microchip I end up with an error that it cannot read from the device, this is mirrored with the read error I am receiving on the compute module. Worth noting that I can connect a usb keyboard to the downstream port in the PC connected configuration and I am able to use the keyboard, though with some error which is why i am suspecting the USB lines. I've since added some 47ohm resistors to the compute module side but in my pc connected configuration, these resistors are bypassed so I am inclined to believe there is an issue on the traces heading directly towards the LAN9512.

I have been troubleshooting my custom carrier board here and have made some adjustments to attempt for better length and impedance in the USB lines. Is there anything else I should be taking a look at? I have determined it is not a power supply issue, the only thing I haven't added is a ferrite bead for isolation of the 3.3v line, though I have tried powering the LAN chip from a separate power source. I have these beads here and may end up trying this addition but I figure that if I can see on my laptop the device then I should not be having any issues from this.

[nenstrom]

Looks like I missed adding this additional screenshot of the error on the Pi.

Just wanted to update that I've added the 3.3V ferrite bead and I still have the same issues.

EDIT:

Added a live look at dmesg. Gives a better look at the errors.

I disconnect my direct line to my keyboard as seen at USB disconnect, device number 2
I then boot up the LAN9512 and the following sequence occurs.

Also note, I do have a 1Mohm resistor in parallel over my clock.

I've realized my mistake with the microchip software. Bit of a slow moment there, I haven't used an external EEPROM in this case. Still not sure why the compute won't recognize the device.

[nenstrom]

For anyone else who encounters this error, it seems to be thrown when there is a general USB bus error. I tried many of the fixes that Ubuntu users had used in software to no avail. I fixed the problem by simply starting my edits on a different board. My edits were cleaner and must have allowed for a better connection over the USB bus. This is why we always order 5 or more prototypes at a time...

Currently working at cascading these LAN9512's, we'll see what my luck is there. I had it working on my PC but the compute seems to be more particular about the trace lengths/impedances.

[ag2000ir]

Can you please explain more how you solved this issue? I am experiencing the same error messages on my board? Was it a soldering issue? or a PCB layout/impedance issue? thank you!