my component level models perform basically the same while .cir model doesnot quite get "there" (what's expected)
googling for substitutes didn't revealed any credible results so - going to error~trial ...
the following figure sows a TOP - dn list of most possible substitutes for the LM308
although the LTC1047 looks a better match by gain curve - then after reading the datasheets of the bad boy (LM308) itself and it's other 3 top substitute candidates the LT1366 seems to be preferred coz LTC1047 needs extra feedback capacitors and LT1880 is referred to as super ß Op Amp (the input sensitivity of such may be a likely source to some instability . . . maybe . . . . . . . . . the LM308 is supposedly not so calm as it's macro-model lets it show ???) - so the LM1366 may be too calm coz it's internal capacitors maybe not - anyway it performs near excellent at the following quick check !
semi theoretical - mathematical application test ...
-- just a quick pileup from scratch -- to get some idea of it's behavior / performance (if it's a right chip for the task and/or substituting . . .)