Insane Hydraulics
Bold and audacious blog about fluid power
This is the pump aggregate from the Komatsu PC88MR-6 excavator (the very same excavator we were troubleshooting last week), and today I'll be striping it down and looking into the details of its intricate design. The function of this variable displacement pump is simple - it is a classic open-loop unit with a closed-center load-sensing and a summation torque limiter, but we, as techs who actually use our hands when we work on pumps and the machinery they are attached to need to know the "hands-on specifics", namely:
Usually, when you back-engineer a pump, you begin with parts and end with a diagram, but I want to do it differently this time - I'll start with the diagram and discuss the operation of this pump before looking at "hard implementations." So, without further ado, here's the hydraulic schematic of this unit:
This four-pump aggregate is composed of a double (45 + 45 cc) variable displacement pump that feeds the main DCV (the arm and the tracks), a 33 cc gear pump that feeds the turret rotation circuit, and a 7 cc pump for the low-pressure (32 bar) pilot circuit.
First and foremost - the variable displacement pump is a double pump with a shared swash-plate. This is achieved by using a ten-bore cylinder block in which alternating bores are routed to two concentric kidney port rings. This means that the two equal and independent "pumping units" are always locked in their displacement. I'll show you how this is implemented in a minute.
There's no bias servo-piston in the schematic and this is not an omission - this unit employs the offset swash-plate axis design. The force projected on the swash-plate by the pistons exposed to the system pressure causes it to swivel towards maximum displacement, and the servo-piston needs to be pressurized to reduce the swash-plate angle. Constructively, there are two servo-pistons, connected in parallel, and each of them has a relatively small spring pressing it towards the swash-plate, so, mechanically, this pump is biased towards the minimum displacement, and it needs system pressure to be able to displace fluid.
What does this mean? This means that, theoretically, this is not a self-priming pump, which is why it is imperative that all the air is bled from the suction line before it is started, and ideally its minimum displacement is set above zero. Fortunately, on the actual excavator, "this is taken care of" - the pump sits below the oil level of a pressurized tank (which is how all tanks should be, IMO), and there's an air bleeding valve on the side of the pump control which allows you to bleed off all the air when you need to, and the minimum displacement is set to 5+5 cc, so this pump couldn't have priming issues even if it wanted to (that is - when it is adjusted correctly). This is something to bear in mind when you rebuild (and test) such a unit, by the way. And if you are used to classic closed-center LS systems - don't worry about the positive minimum displacement "ripping through" the closed-center when no functions are called - the DCV of the excavator is "false closed-center" and actually has 29-bar unloading compensators in both inlets.
The load-sensing spool that balances the delta-P exposes both of its ends to the "outer word" via the LS and the PP ports, which is common for Komatsu systems but not common at all to industrial pumps, which usually have the high side of the delta-P spool channeled internally to the pump outlet.
But the most interesting part, in my opinion, is how the high pressure from the two pump outlets is "communicated" to the actuator of the torque-limiting spool. If you look at the diagram, you will see that it is done via a combination of two orifices connected in series (part Nº 2 in the diagram), which essentially creates a pressure that is an average of the two pressures in the pump outlets! This is such an elegant way of making a single area of the toque-liming piston work with two independent pressure sources. The classic alternative would be to use a piston with two distinct areas and individual feeds, which would avoid the "clog-able" orifices, but would also make the design more complicated - so this is a valid (and interesting) solution.
The servo-pressure, on the other hand, is supplied via a combination of two check valves (Nº3) and, therefore, is always the highest of the two outlets.
Did you notice the "sabotage screws" in the schematic? I am talking about the variable orifices marked with Nº1, and I call them "sabotage screws" because any intentional (and permanent) leakage path to the tank sounds like sabotage to me. They are supposed to be fine adjustments for a perfect straight travel - something tracked excavators often struggle with. To be honest - I am not sure if this is dumb or not - I would most likely keep these closed in my excavator. But they are there - so we all must know they exist.
I tried my best to represent the "real life" in the diagram, and I hope that it is a good blueprint for determining "potentially troublesome spots":
Now that we already know the theory behind the operation of this unit, it is time to dive into the implementation (in other words - let us open the pump and look inside). First of all, let us remove the gear pumps:
As you can see, both of the gear pumps connect to the main pump via internal channels and o-ring sealed ports on the flange face (fancy!..) Let us see where the holes lead to:
It is interesting to note that the test ports for the pressures of the pilot pump and the rotation pump are located kind of next to the pump outlets, which may lead you to a false assumption that these could be the main pump test ports.
Let us remove the "sabotage screws" now (so sorry, but I could not resist):
At least the restrictors are tiny, so I suppose their sabotaging capacity is somewhat limited. Now, let us remove this "thing" that looks like a cartridge valve:
Aha - so this is where the double orifice lives! Note the tiniest holes (can get clogged) and the seals (can fail). Now - I wonder what these two screws between the pump outlets might be? They look like adjusting screws, or, maybe, by-passes, don't they?
Oh, I see now - these are the check valves for the servo-pressure supply! Let us move to the control valve now. By the way - here's another "hidden trap" for you - it is not exactly in the pump, but it is close - just look at the orifice hidden in the LS fitting:
I like the presence of an air-bleeding vale on top of the LS fitting - that's a great touch! Let us remove the control valve assembly now:
Komatsu loves these small wire-mesh filters and, I must say, they make sense. Note the (very important) air-bleeding valve on the side of the control valve. Bleed off all the air from it every time you do an overhaul or an oil change! Let us put the control valve away (for now) and continue with the disassembly:
It is good to know that the shaft seal carrier comes out easily, and you can replace the shaft seal without disassembling the pump. You can see that the swash plate is pivoting on two large balls, and - quite surprisingly - there's no pressurized lubrication here! I think larger models do come with pressurized lubrication of the bearings - but this one has none! I suppose the wear on the ball and the cavities (especially on the pressure side) prove that it is not the best of ideas. Let us continue with the rotary group:
Now you can see how you can fit two equally sized pumps in a single barrel. If you carefully look at the valve plate and then inside the pump casing, you will see a lot of interesting things. First - you can see that the two holes at the beginning of the suction kidney port align with a machined groove under the valve plate. This means that there's a restricted connection between the casing and the suction line, which in its turn means that it is very easy to fill the casing by simply opening the bleeding screw on the side of the control valve. Then, you can see that the two larger holes in the middle of the pressure kidneys lead to the check valves, and the smaller holes at the beginning of the timing grooves lead to the double-orifice assembly. Quite interesting!
This here would be the magnetic plug:
And now we can move to the internals of the control valve:
Notice the detail of the eccentric axis of the displacement feedback lever that allows you to adjust the torque limiter threshold and the variable rate spring of the torque limiter. Also - check out the design of the load-sensing spool - the pressure differential is balanced over the area of the small-diameter "pin" on the right side, because there's a hole through the spool that connects the left side to the orifice that you can see in the picture.
What else can I say about this pump? It is very compact, and I suppose is good enough for this system that works at a relatively low max pressure of 290 bar. But the failure of the servo piston that we saw last week and the wear of the swash-plate ball bearings is quite substantial for the 6000 hours she clocked so far. It almost feels like this is a medium-duty pump in a heavy-duty application. It's OK if you look at this from the commercial standpoint (hell - even we are profiting from the overhaul) but I suppose it could be much better if the swash plate bearing system and the servo pistons were beefed up a little (or, you know - redesigned altogether). I can also tell that this pump is far from being intuitive for a hydraulic tech and you definitely need a manual to know "where stuff is" when you intend to work on it or troubleshoot it. Hopefully, this post can help with that!