Today's case is another example of how bench tests can fail to simulate real-life conditions.
Back then I was employed at a company that sold LOSI steering units and hydraulic motors ("reasonably priced" Danfoss clones of acceptable quality).
It was a warranty claim - a client complained that a brand new hydrostatic steering unit was causing the oil of his tractor to overheat, even though the power steering function in itself seemed to be working fine. He brought the unit over to our shop, and it was immediately hooked up to a bench test. The steering valve was of a very basic type - simple open circuit, non-reaction, a relief valve and no anti-shocks - the type any hydraulic shop sells a dozen a week... The test itself was also the classic no-brainer - the unit was connected to a fixed displacement pump (P port), tank (T port), and a hydraulic jack (ports A and B). The steering valve functioned flawlessly - everything worked one hundred percent ok without anything even remotely suggesting overheating. The client, as it would be expected, was advised to look for a problem somewhere else in the hydraulic system of his machine.
A couple of days later the man came back and said that he was 100% sure it was the new steering valve that was overheating the oil, because when the new unit was replaced with the old one - the overheating disappeared. The power steering function was fine - the wheels turned and the effort on the steering wheel was normal, yet the oil would overheat badly. Understandably the man was pretty upset because in an honest effort to "fine-tune" his tractor he'd paid good money for the new steering, which was now causing him downtime without any fault from his side.
So, the unit was reconnected to the test bench, a steering wheel was attached to it, and a mechanic spent a good hour "playing Schumacher" in an attempt to see if any signs of malfunction or overheating would appear - and yet again to no avail - the damned steering was working fine... The working pressure was normal, the relief valve was functioning fine, the delta P between the P port and the T port was normal - nothing, absolutely nothing suggested that the unit was problematic in any way. However, the fact that when the owner replaced the unit with the old one the overheating problem went away, and when he re-installed the unit back the problem reappeared, was pointing to the unit as the source of overheating.
Luckily fate intervened - the frustrated Schumacher gave the steering wheel a violent kick, and suddenly the unit went "s-s-s-s" and the stand-by P to T free flow differential went from 0.5 to 40 bar!!! Obviously, something was strangulating the oil passage. When the pump was stopped and re-started - the "s-s-s-s" disappeared, and when the wheel was turned violently again - reappeared! Hurray! If you see the problem you can solve it!
It turned out that the "inverted umbrella" check valve in the P port was incorrectly left too low, which allowed it to function as a speed fuse of sorts. Above a certain flow rate the poppet would "glue" itself against the bottom, strangulating the flow and causing the overheating without jeopardizing the assisted steering function, as the oil was still passing through, because the bottom hole was drilled at an anlge. When the operator started the tractor, he would rev it up for a moment, causing the flow spike large enough to put the check valve in the "heating mode", which would then remain stuck in that position till the engine was turned off. The problem was quickly solved by fixing the check valve in the correct place.
It is interesting to note that the flow rate of our test bench was just about "right" to be opening the valve without bottoming it out. The violent steering inputs would pressurize the 2-meter long pressure hose and then relieve the pressure abruptly, so I believe it was the hose accumulator effect that gave the P line flow an additional boost.
Morals of the story: