"I thought that for an input below Vref I should see zero at the output 6..." The 741 opamp is a very old design which does not have a rail-to-rail output swing. A low level output will be a volt or more above the negative rail. "...and for an input over Vref I should see the input voltage (less a bit)" And for the same reason a high level output will be a volt or more above the positive rail. However,  your circuit will not work reliably because when you open the switch,  you leave the non-inverting input floating. You cannot do this because the two inputs require a small but non-zero bias current. Offhand I cannot remember if the current into

the inputs of the 741 flows into or out offthe input terminals. Either way you must return the non-inverting input to something like a mid-rail voltage through a resistance It is an unusual device in several ways an in fact is quite far from being a classical opamp, especially its input stages. There are some interesting articles about the original design and how it came about on the web. Please look up the 741 on wikipedia and study the datasheets for it. Please also note that although it will function, the 741 is not intended to be used as a comparator. :)

Thank you for that. Will read up and try and find a more suitable item.

Bizarrely some of the dedicated comparator chips cannot tolerate a differential input of more than some specified amount - which is a but counterintuitive for a comparator - so you may have to check the differential  (the difference between the two inputs) and the common mode (the common bit of the input voltage that both inputs see) range over which that differential input spec applies. You may also therefore have to pot down the input swings/voltages to fit inside both ranges.

Thanks. I have some LM393, LM311,LM358s coming. I thought the LM311 looked promising for what I want. I need to read 3.1V output from a current sensor when 4A is flowing, with 2.5V output for 0A flowing, so intend to set the threshhold at around 2.8V with the divider