We think we are starting to understand the air issue. It seems that when our leg is lowering at speed, the lowering speed can reach a point that the either the flow rate of the pump is exceeded or the flow rate through the valve limits flow to a point that negative pressure is produced. This we believe is releasing air in the hydraulic oil on the bore side of the ram, then when the ram changes direction at the bottom of the stroke, the air is re-compressed back into the oil causing the jolt we are experiencing.
The other issue is that when the leg is switched off, due to the leakage of the fourth position on our proportional valves, the leg slow sinks, as oil is compressed on the annulus side of the ram and squeezed past the valve. This seems to drag air in some where around the manifold and into the bore side of the ram. So what is needed is a zero leakage solution to lock the leg when power/pressure is switched off. For this we can use pilot operated check valves on both the bore and annulus sides of the femur ram. However, in a closed loop system, the AB ports should not be piloted from each other, they must be piloted from the system P port so as not to disrupt the control loop. This was working pretty well, and indeed has stopped ingress of air into the femur ram by not allowing any movement when the system is switched off, however, we found the bore side of the ram was being restricted by the check valve, so we have removed this valve leaving just the one check valve on the annulus side, which still hold the leg still when power is remove, providing of course the load is pulling the leg down.
We are trying to find a cetop 3 sandwich valve solution, which we will install on every leg proportional valve, so far we can only find pilot operated check/poppet valves that are cross piloted from AB and not from P.
We have found a slice valve that could be installed under the prop valve, but this requires two solenoids to block both A & B, which of course is a little power hungry and seems excessive.