Horrible signal reflections and timing violations in DDR3L HyperLynx simulation of the STM32MP157 Discovery
I'm running the STM32MP157A-DK Discovery board through HyperLynx and I'm getting some pretty horrible ringing on the address lines, along with timing violations when I use the DDRx Batch Simulation, and I'd like to verify my setup is good.
First, I exported the board from Altium using the HyperLynx exporter, and then opened it up in HyperLynx, added my IBIS models, set my pins to MSD_D3RP3L_48, set all the address / command termination resistors to 56 ohms as indicated in the schematic, and ran an eye diagram, as seen at the memory chip die:
There's substantial ringing on the bus, which appears to be caused by the long stubs that run off to termination resistors.
To check that theory out, let's focus on the yellow DDR_A7 signal (one of the shortest stubs on the PCB), and DDR_A12 in blue (one of the longest):
Clearly, the stubs off to the VTT resistors appear to be causing the problem, as the longer-stub signal is much worse. I reran the simulation with very very large values of resistors, just to ensure that my simulation was correctly using the 56-ohm resistors (if they were DNI'd, the unterminated stubs would definitely cause issues). I got way worse data, as expected:
So, my first question is whether the above results are consistent with other people's simulations of the design? Is there something I'm missing?
Next, I ran the DDRx Batch Simulation. HyperLynx flagged every pin with some sort of read operation setup time margin failure.
I've attached my simulation project with the DDRx Batch Simulation results.
Is there something I'm doing wrong? I'd appreciate any insight! Thanks!