I would think more sensors, regardless of type. In a linear string lowest to highest.That would give you a finite number of valid logic states and some discernible error states. Any individual sensor failure could then be recognized and you can react accordingly. Given the inherent risks of pumping gasoline, I'd also consider a more involved system to turn on the pump. Possibly two separate processors driving separate relays in series with some sort of data protocol between them. That would allow them to watchdog each other. Also, many years ago I worked with a guy that was developing an new no-moving-parts fuel level sensor for cars. From memory, it was a thermistor (PTC/NTC?) with some current run through it (Constant?) and a comparator to measure the voltage across it. The mechanism is that the self heating of a sensor submerged in fluid is distinct from one that is not submerged. Fuzzy memory says there were some interesting issues associated with decoding it and dealing with the asymmetric time constants involved (slow heat up, fast cool down when the fuel splashed around) -Denny On Wed, Oct 16, 2013 at 6:22 AM, Carl Denk wrote: > For my standby generator 16 gallon gasoline tank, there is a nearby 175 > gallon gasoline tank with a 12 VDC transfer pump to keep the 16 gallon > full. The pump ground has a 12VDC 1" black cube automotive type SPDT > relay that is using the NC contact that opens when the relay is > energized. The relay coil is energized with a IRF9530 P-Channel Power > MOSFET, which is activated with a H11B1 opto-isolator. The reed switch > is a float switch to ground, NC at normal fuel levels, and NO at high > level activates the opto-isolator. There is a suppressor diode across > the opto-isolator input and dropping resistor from the 12 VDC power. > There is also one section of a PS2501-4 opto-isolator driven by the > switch that goes to a PLC that turns on the pump (there is also a manual > pump switch). > > Recently the reed contacts welded together causing a possible unsafe > condition of overflowing gasoline. I have new switches ordered, the > question is, is the current configuration of the usual alarm circuit, > normally powered, and a fault (could be a broken wire) open is detected > and shuts thing down. But with welded contacts that's bad. :( It > wouldn't be hard to make the reed switch NO, change the relay contact > used to NO, and program the PLC for inverted action to make the circuit > powered is alarm. Think it's about which is better scenario. > > Oh the system is 7 years old and working OK until now. > > Comments appreciated. :) > -- > http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive > View/change your membership options at > http://mailman.mit.edu/mailman/listinfo/piclist > --=20 http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .