For me, the most beautiful and perfect hydraulic circuit has always been and forever will be the closed-loop hydrostatic transmission. Simple, at first glance, and yet so complicated, when you really look into it... The best solution for infinitely controlled bi-directional motion that must be delivered in a compact package, with the most common application being industrial vehicle propulsion.
Although hydrostatic transmission is an old technology, and there are many shops out there advertising readiness to tweak your closed-loop on your first demand, I still come across off-centered closed-loop pumps more often than I would like to. And, to top it off, I sometimes even get to see intuitive, yet incredibly nonsensical techniques used to "catch the damned zero", showing a complete misunderstanding of how the given servo-mechanism works and what must be done to "make it stop!" This series is, therefore, my attempt to shine some light on the very important matter of adjusting the neutral swashplate position in closed-loop pumps.
Nobody likes heavy machinery moving on its own. Unless, of course, we're talking about a runaway forest crane slowly bulldozing the shiny Boxter that your boss bought right after telling you that "these are tuff times and the raise will have to wait...". (Just make sure that you deliver the news about the cruel hydrostatic malfunction without giggling!) But, in all seriousness - correct null adjustment of a closed loop is very important.
I must warn you now that extreme care must be taken when dealing with hydrostatic transmission neutral adjustment. If you are new to this, you should NOT do it. Period. Don't even think about it! Although setting your machine in uncontrolled motion will be fun to watch, it won't be fun to be responsible for, believe me! If you're not sure whether you are up for the task, call someone who is. And if you're brave enough to take any of my advice - read the disclaimer before you reach for the spanners.
Now that we're done with the safety drill, let's go ahead and talk about them transmissions.
Let's ask ourselves - where do the "crooked" transmissions come from? Well, the sad truth is that most closed-loop pumps don't go off-center "by themselves", but are the result of "craftsmen intervention" normally performed "to improve something". The pumps that do go off-center by themselves either were left with adjustment screws unlocked, or have gained a whole lot of internal play due to wear, and, in the latter case, aren't null-adjustable any more... From what I've seen so far, the most common reason for a hydrostatic transmission to exhibit the "drifting null" symptom is the recent overhaul, during which the "overhauler" skipped or simply screwed up the adjusting procedure.
Before going into any adjusting techniques we must define the best way to check if the null swashplate position is set correctly. I consider the best and easiest way to be the comparison of the pressures in the loop legs, which, with no control input and the actuator not moving (or the load inducing needle valve on a test bench closed) shouldn't differ more than a couple of bar. The null check-up and adjustment should be performed at normal system working temperature and at high idle. It is very important to make sure that the pump goes back to zero pressure differential after stroking and achieving maximum system pressures in both directions, and takes the same amount of time to de-stroke in both directions. It is also important to verify control sensitivity - the response to the input signal should be symmetrical for both sides (although this may not be imperative and can depend on the type of control used). In case of closed-loop pumps driven by relatively quiet electric motors, you can actually hear the pump go off-center as it builds presure in one of the legs, something an experienced mechanic can use as a covenient reference, but checking those two gauges is still necessary.
It is also possible to check if the pump is off-null by reading the charge pressure. This technique can only be used with closed-loop circuits equipped with a loop flushing system. Any unwanted pressure differential will trigger the loop flushing spool, which will cause the charge pressure to drop. The advantage of this method is the fact that you use only one pressure gauge and, if you use an analog needle-type gauge, the pressure drop can be easily seen from a distance. However, this technique will NOT work for all circuits, because the charge pressure drop depends on the type of valves used in the flushing block, and for certain arrangements can be hardly noticeable at all.
Now that we know that the best way to check the correct neutral swashplate position is the work-line pressure comparison, let us look closely into typical closed-loop controls and servo-mechanism designs, and the best ways to null them.
The most common type of closed-loop pump control is proportional displacement control (hydraulic or electric). It is also normally the one that gives most "zeroing troubles", because it uses two independent systems for the swashplate centering - mechanical (springs) and hydraulic (servo pressures).
Mechanical positioning can come in many flavors - it can be done by means of one spring (example), two or more springs using the same design principle (example), two independent pre-loaded springs (example, feel the nostalgia), or one or two pre-loaded springs, constructively placed inside the servo-piston (example, used on Danfoss H1 series).
Hydraulic positioning is done by the proportional displacement control, which defines the position of the swashplate through servo-pressures.
Both of the systems have deadbands, and when they don't overlap, an unstable null condition is created, because the hydraulic positioning system overrides the mechanic "forcing the springs" into the correct null, but, like any hydraulic feedback positioning system, it is subject to small fluctuations (originated by a number of things, like charge pressure deviations, shaft rotation speed, vibrations, case pressure fluctuations, etc..) which result in unwanted swashplate movements when the spring-centering mechanism is allowed to "push back to its own zero" and, consequently, cause the null to drift. The only way to create a stable neutral is to make sure that both mechanical and hydraulic positioning deadbands overlap symmetrically.
This can ONLY be achieved by adjusting one system at a time and ONLY in the following order - first - the mechanical zero, second - the hydraulic zero. If proper tooling is available, correct mechanical zero can be adjusted during the assembly of the pump, if not, you will have to do it during the testing phase. The most common course of action would be to eliminate the action of the hydraulic positioning system by by-passing the servo-cylinders with an external line (the by-pass hose should be as short as possible to minimize the pressure drop. I use a ball valve in the middle of the line to conveniently turn the by-pass on and off) and adjusting the mechanical zero first. Then, after the mechanical zero is adjusted (note that the best way is to always look for the middle position of the adjusting screw within the deadband, not just the first position when both of the loop leg pressures level out), you remove (close) the by-pass and adjust the hydraulic zero, again aiming to leave the adjustment in the middle position within the deadband. Skipping this procedure, and trying to adjust the neutral position with both positioning systems active produces unreliable results.
Note that for some pumps adjusting the correct mechanical neutral position of the swashplate can be more challenging than simply turning an adjustment screw. In Sauer Danfoss series 90 pumps, for example, the mechanical zero is defined by the position of the round side cover that guides the centering T-bar (drawing) and has a small rotational play. To adjust the null you must loosen the six screws and then turn the cover into correct position by hitting on the machined pit (one of the few cases you may actually need a hammer to adjust a hydraulic setting).
Again, some series 90 pumps use an extra spring to hold the swashplate in neutral (like this example here). This design is used to create additional force in the servo mechanism. The best way to precisely adjust the correct mechanical zero in such a pump would be by adjusting it during the assembly process.
The described technique of servo-cylinder by-pass works well on closed-loop pumps that have spring centering systems with fixed swashplate free play (usually insignificant, existing only due to small clearances in the servo-mechanism). But what about servo systems that can allow, when set incorrectly, a significant swashplate free play. For example, on pumps that use this servo-mechanism design, or this, a "fiddler" can unscrew the servo-cylinders and leave the swashplate loose. If hydraulic zero is set correctly, there will be no pressure differential in the loop, but the response of the control will be asymmetrical, and the displacement control fluctuations, which in case of correctly adjusted mechanical zero would have been absorbed by the springs, will cause the null to drift.
Obviously, a different adjustment technique must be used for such pumps, which, along with other null-adjustment-related stuff will be discussed in Part 2 of this series.