I have been having issues with TMC2100 drivers also. My first ones from BigtreeTech behaved inconsistently, so I purchased some Watterot versions. I followed the recommendation not to solder the pins to the 3 config sockets, which puts the motor in 16 microstep, 256 interpolation stealth chop mode. The motors are 0.4 A, 0.9 degree step angle.
The motors run very smoothly and quietly, but slew extremely slowly. The mount also only moves a fraction of the distance it should (e.g. a command to move 90 degrees results in 3 - 4 degrees of travel.
EQMOD runs happily enough (no errors reported).
One thing that concerns me, is that each time I configure the board (a new genuine Arduino Mega 2560 running 8.11), the 'Goto Rate (x sidereal)' parameters show different values. These vary from 102 through 460 to 507, despite the input parameters beingthe same (I loadthe EQ5 modified pulley config and change the driver to TMC2100, 16 ustep, and 0.9 degree angle, then update both axes. I am going to try this on a different PC to see if that helps, or possibly set the angle in the config to 1.8 degrees to see if that speeds it up. Sidereal tracking rate looks to be about right, but have not yet tried it on a star as there is obviously something amiss at the moment which needs to be rectified.
Using jumpers, I can set the drivers to 16 ustep, no interpolation mode - will give this a try when I get a chance.
Meanwhile, my 8825 setup with diode matrix runs happily in the observatory . . . .
Dave