Hello, i'm currently assisting Jesse with this SD Card Issue.
we have quite made some progress.
The board revision we are testing has issues with the SD communication at 50MHz Clock rates.
No changes on SD connectivity have been made on the latest revision wrt the 'functional' revision.
To check this we have verified the following items on the SD interface:
- Trace length
- board stackup
- trace width,
- ESD diode type (designed vs mounted, to exclude assembly options)
we have found no differences between the functional and non functional revisions.
We suspect timing issues on the bus, this is more or less confirmed by applying cool spray to the TXS0108, a SD card is identified in Linux, but errors are still reported.
Removing the SD Pullups, and replacing the 33E series resistors with 0Es has a positive effect on the performance, the SD interface operates stable.
As a test we have moved the TXS0108E from a functional board of the first revision to a 'non functional' board of the second revision (with 33E series resistors and pullups in place) and behold, the SD interface is fully functional!
Question 1: Has there been a change in chip Die recently? This might be a reason for this behaviour.
The 'working' TSX has the following markings: YF08E, 84KG4, D27K
The 'faulty' TXS has the following markings: YF08E, 2BKG4, C9SY
Question 2: Are we using the device out of spec when using 50MHz and were we 'lucky' in the previous board revision?
as Figure 6 of the datasheet indicates a minimum pulse width of 16.7ns when VCCA = 1V8 and VCCB = 3V3, which equals a max frequency of 30MHz
Regards Arne Maathuis
