While testing the small 16-button keyboard, connected to the expansion board, I encountered a problem -- half of the buttons haven't registered keypresses at 10 kHz scan rate. When testing with manual scan - i.e. row address set statically through switches, all button presses where visible. After investigating with oscilloscope, I found out that some of the input pins on Tang Nano 9k board had too big capacitances attached, and charge/discharge of them was too long - it took almost a millisecond to reach proper logic level in my setup. This way, the shorter time period I used for scanning was not enough for proper voltage change and thus the registering button press.

Perusing board schematic I found out that several pins on board do indeed have capacitors attached -- those are dc filtering capacitors for hdmi signal lines. As I do not intend to use hdmi connector on this board, I decided that removing these capacitors will only be positive for my intended use -- and thus the surgery was performed:

The eight pairs of square pads in the middle are the connection points on smd capacitors that were used for filtering hdmi signal. They were interfering with inputs of shared GPIO pins (68 through 75), and were removed.

------------------------

Now, the board is not able to output hdmi signal, and in this way it is crippled. But, as I've mentioned above, I do not need hdmi output, and the capacitances were annoingly interfering with inputs, so for me now it is a net positive.

If in the future I should need the hdmi signal, I will have to generate it with other means, or order new untouched board.