TIDA-00366: Tackling EMI Problems regarding Speed Measurement

Hi all! ;)

We are using a LAUNCHXL-F28377S to control a three phase inverter (reference design TIDA-00366). Our current problem is high frequency noise on the eQEP input signals A and B from the digital encoder. We programmed a PLL on the MCU to track the shaft speed of the machine.
Because of the PWM of the inverter, there is much EMI in the environment of the MCU and the digital encoder. If we increase the DC link voltage above 40 V, the noise spikes on the signals A and B become quite big and the PLL loses track of the shaft speed.
This is no problem of the digital encoder: We already tested this with the encoder unconnected, using a signal generator to simulate the quadrature pulses at the eQEP pins of the MCU. The result was the same when we increased the DC link voltage above 40 V.
We think this problem is a ground loop: Every signal on the inverter PCB (the whole MCU, measurement OPAs and the digital encoder) refers to a common ground GND. When we connected a different MCU to the PCB to drive the PWM and used our original MCU to measure the speed via encoder, the noise also occured. But the only connection of the speed-measuring-MCU to the PCB was the GND signal.
Do you also think this is a ground loop? Do you have any ideas to tackle this EMI problem?

Best regards,
Armin