Testing - Better late than never...

Crap! I have been toiling away on a test rig to develop the damping feature of the valve. The damping is one of the most critical features to make sure that the valve performs reliably in service, but was not part of the original prototype so the damper design was never tested. It seemed simple enough. I put in the necessary design elements. I even planned for the damping fluid to be misplaced when the valve was turned upside down and provided drain holes within the valve for the fluid to go back to where it was supposed to be. I didn't test it earlier for three (bad) reasons. 1) I was confident it would work. 2) I reasoned that even if it didn't work right off the bat, it would be easy to tweak on the production design 3) Building something representative was difficult due to the design. The parts I can machine do not match the molded parts.

All along I had planned to develop a test specification for the manufacturer to develop the damping performance. In order to do that, I reasoned that I would need a test rig that simulated many types of irrigation system installations (in terms of air entrainment...the source of all my grief). Having the rig would allow me to develop a "profile" of all the types of sputtering and spitting that my valves would have to tolerate. Then I would turn this profile into a specification and have the manufacturer duplicate the test rig to test the damping. That was the plan. Here's what really happened...so far.

I built the test rig. It didn't work. Even though I provided for multiple flow paths and all types of high points and drains so that air would be trapped, the ($300, 3 days to build) rig never produced anything like the sputtering I see when I turn on the sprinklers in my lawn. I realized this was due to the fact that simply trapping air was not enough. There was a delicate balance of slope vs spray head distribution in my lawn that caused the sprinklers to sputter, so just trapping "slugs" of air in my test rig didn't match what I saw in the lawn at all. After much trying, I decided the test rig was a bust and I would need a different approach for the damping tests. I took a standard sprinkler control valve and disabled the main valve so that it was acting only as a direct control solenoid valve (the pilot circuit, for those in the know). Then I opened up the solenoid flow path so that I could switch significant water flow on/off very rapidly. I hooked it up to a controlled leak and viola, instant pressure fluctuation. Anything I wanted, all electronically controlled. The test rig would still be useful for endurance testing, but I had my configurable test apparatus.

Then I set out to build a damped valve to test. If I left out the switch details, I could get a pretty good representation of the damper function. When I built it I was shocked. The damper didn't work worth a damn. The drain holes that I had so thoughtfully placed were acting as pressure equalization ports between two sections of the valve (no, not directly in line with the damper...I'm not that stupid) and causing air to sucked in by the damper piston. I made some improvement by opening some vents in the top (one mold change already...and the molds are almost done) but it still wasn't enough. The drain holes occasionally pulled air into the damper even if I filled the valve up with so much fluid that the holes were never uncovered. Did I say CRAP yet? The valve has to have air in it to compress or the whole concept of operating off of the main line pressure doesn't work. High tech dampers have the air and oil separated with a floating piston. Doing that would add considerable complexity to my design (more expensive mold changes). On the other hand, many dampers work just fine without a separator as long as they stay in the right orientation. I tested my damper without the (needed) drain holes and it works fine. I can't possibly ask the user to never turn the valve upside down, so I need the drain holes. So I decided to write. I have no solution. I have molds in process in China that should produce their first parts in about a week. Parts that I know now will not work as designed. If I tell them to stop, I'll have to get in line to get back in the queue. If I keep going, maybe I can salvage the tools in the development phase. I need a solution...and quick. CRAP!