T-Rex 3 Stepper Motor Torque Testing

The T-Rex 3 is a large format 3D printer, marketed and priced as an industrial 3D printer. The frame appears ruggedly built aluminum extrusion, especially when compared to the standard I3 format printers that it copies from.

The biggest problem I've experience thus far with this printer is layer shifting, especially on large prints. 

I found that the TMC2208 drivers had inadequate cooling, getting as hot as 115C on the plastic IC case in only a few minutes of printing. I addressed this issue by improving the thermals. That solved the vast majority of my layer shift problems, but I still had a print for a client that would always have a layer shift.

I began to suspect the stepper motors, so I decided to test them with a torque meter.
I purchased this torque meter to do my testing.

I created a part in Fusion 360 to create this adapter for my test, then printed it in PLA.

I tested the left X drive stepper motor. The stepper motor is rated for 1.7A, which is 1.2A RMS. As it turns out, the TMC2208 is rated for 1.2A RMS at its highest, so I set the Vref on the TMC2208 driver to 1.7V.

After setting the current, I engaged the stepper motor and locked it in place. I tested it with my meter, setting the meter to record peak torque. Next, I turned my meter until the stepper shaft turned. 

The peak torque recorded was only 39.7 oz-in (0.28 N*m) of holding torque. 

My second test had me holding the torque meter while moving the stepper motor using the knob on the front of the printer to advance 10mm. 
This resulted in measuring 61.9 oz-in (0.437 N*m) of torque, which is closer to the rated torque of the stepper. (Supposed to be rated at 84 oz-in of torque, but I don't really believe that rating). 

My third test, I took another cheap stepper motor from China (17HS4401), rated at 1.7A, but only 56 oz-in. This one I only did the static torque test, and read out 36.3 oz-in (0.256 N*m) of torque. 

My fourth and last test, I removed the TMC2208 and replaced it with a tried and true A4988 driver. I set it for 1.4A (highest I could set it) and repeated my measurements. With the stock T-Rex 3 X drive stepper motor, I measured 58.6 oz-in static, 54.3 oz-in dynamic torque. Turning the current down to the 0.8A level that I run on my other printers, I measured 39.2 oz-in static, 36.4 oz-in dynamic torque. (Yes, static was higher on the A4988s). 

So, where does this leave us? 
Clearly the TMC2208s have more than enough torque at very low speeds. I can only assume that at higher speeds, the 1/256 stepping causes problems. 

The default acceleration in X for this printer is 1500 mm/s^2. That is pretty high. I'm guessing that turning this down to about 800 mm/s^2 will help solve this problem.