I ran some tests this weekend and while not conclusive they seem to generally agree with what is posted here I tried 4 configurations each for one test run and here are the results.
Duckbill type check valve, installed on line between tank and dispenser (out line) and input line ziptied to output line 1" above dispenser- no drain back.
Duckbill type check valve, installed on input line, input line also ziptied to output line 1" above dispenser- no drain back
Spring type check valve, installed on output line, input line ziptied to output line 1" above dispenser, no drain back.
Spring type check valve, installed on input line ziptied to output line 1" above dispenser, slow drainback of the volume of the hose from the dispenser to the tank.
The check valve mounted above the dispenser works basically like holding your finger over a straw and it seems to work better because as long as not so much powder is used (I use 6 teaspoons) no powder enters. I think on the back end some powder drains back and this would be unsoluble calcium carbonate most likely and it jams the valve open slightly. We have used three check valves in the 4 years the 5074 has been available and each had it's draw backs and I still think the spring is the best so far. The first one we used was a rubber flapper valve and it greatly reduced flow. The second was the duckbill type and the problem is that while they work great for 6-18 months the kalk hardens the rubber and it gets stuck open. The spring type closes slower and on the back end it seems powder can jam it open. I think the critical point in all installations is zip tying up the inlet hose above the top of the reactor.
