The HAM1 model was restarted this morning to get fix for the issue I found on Friday. After bringing the HEPI back up I tested some configurations and adding the 3dl4c ff path to the twist sum reduces the measured low frequency bending motion cause by the ff drive. Attached plot compares a segment of time from last night (red and blue) before the model restart and this morning after (green and brown) while TJ was relocking. The blue trace shows the excess noise caused by running the 3d ff around .1-.5hz. This excess is gone in the green, and is basically the same running without the ff now. This isn't really an issue for the other HEPI, but it would probably be a good idea to extend this fix from the friday alog to all the other HEPI models.