very interesting thread, thank you for that
I was aware of the microstepping issue and the fix by switching to fast decay mode with the possible risk of introducing more noise, but did not yet know about the solution with antiparallel diodes.
So I joined the game with my Vertex K8400, as all Stepper Drivers are DRV8825 there and it also runs on 15V like the K8200 (correct me if I’m wrong here)
Halfway through the soldering I noticed something that might be crucial for the success with this mod: The printer in the original linked blog post runs only on 12V and this might turn out to be a problem (not sure yet). The diodes drop 1.4V, but with a 15V PSU 12% are 1.8V, so it would not be enough to eliminate the problem.
So far I only looked at the voltages on my scope as I’m apparently very good at hiding my shunt resistors, so I had to order new ones today.
This is with the mod. The most visible part is of course the switching between low and fast decay mode, [color=#FFFF00]but I think the actual microstepping problem is in the gaps where the sine wave crosses zero[/color](EDIT: wrong part of the curve), but of course I still have to measure the currents. It could be necessary to put six antiparallel diodes per coil which would be even more space heating (still cheaper than new stepper drivers with their own issues).
To see if it made any difference I connected a 12V brick instead of the normal 15V PSU, but the curve on the scope was the same, so it’s possible that measuring the voltage is just useless in this case.
I think the printer sounds a bit different at low speeds and I can’t see any artifacts in the prints, but I only finished the mod late last night so it might just be sleep deprivation hallucinations
EDIT: If my math is correct then with a 15V PSU, 3.1V nom. 2Ohm motors and 2Amp full scale current of the driver, there are actually 10 microsteps missing in 1/32 mode. The diode mod with 4 antiparallel diodes in series with the motor reduces this to 2 missing steps (maybe 1 if lucky with the diodes)