The microSD card made by Catalex in China has a fault in the circuit. The MISO line never goes into the tri-state mode. This means that the card works correctly, but any attempt to use the SPI bus for any other communication is doomed to fail.
I have been in touch with Catalex, now rebranding as Open-Smart, and they have changed their circuit, on my suggestion. They expect to run out of their current stocks next month and expect to start shipping the new cards in February 2020.
A simple test revealed this. I applied the power to GND (0V) and VCC (5V), and took CS to VCC. Two 1K resistors were wired to the MISO pin, one connected to GND and the other to VCC. The voltage on the MISO pin was therefore expected to be half-rail, 2.5V. But it was held down to 0V, showing that the tristate buffer LVD125A was not in the tri-state mode. The appropriate control pin was pin 13 on the chip. Pin 13 was soldered to GND. So the card was permanently selected. The chip select (CS) had no ability to deselect the card.
The new microSD card will correctly allow the CS to deselect, tri-stating the MISO line, allowing other cards to use the SPI bus.
Thanks very much for getting this done. So the gate enable will be connected to CS after CS has been converted to 3.3V (i.e. - at the card, not at the header)? Is it at all practical to make that change on an existing module? If the enable is tied to ground, you would probably have to cut the pin at the board and bend it up. Or maybe cut 13 and 14, then reconnect 14.
What I did was to float the chip off with a heat gun, bend up pin 13, then re-solder. Correctly a wire should be soldered between 13 and pin 9 (the chip end of R1) but I found it easier to solder the wire directly to the CS header pin, after checking that there was zero current being drawn from the 5V drive.
I guess I would want to connect the enable pin to pin 8 of the 125, or to pin 2 of the MicroSD holder, so the enable pin would see a 3.3V input instead of 5V. But it appears this chip will tolerate 5V at the inputs even if its Vcc is at 3.3V, so I guess it doesn't matter. But as a general principle, you wouldn't want an input to see a voltage that's higher than Vcc. Do you know how Catalex is going to do it?
