When you say that you are not getting what you expected to get, presumably you are saying that a simulator is not giving you the answer you expect. Are you using behavioral models or circuits constructed from transistors? Are bandwidth effect modelled at all? If you used a behavioral model, did you model clipping?

You may getting a mathematically valid but impractical answer. To see if this is the case, try skipping the initial DC solution (skipdc with Spectre, UIC with SPICE (you will need some capacitors in your circuit)). Alternately, decrease the transition timeof the input source to the point where you would expect to see bandwidth effects. Once you do this, I expect that things will start working as you expect them to.

-Ken

There seems to be a basic misunderstanding of the effect of positive feedback on the operation of this circuit. Applying positive feedback increases the gain of the system and any significant amount can make the gain greater than infinity. The result of this is that the output will be driven to either limiting value, close to the supply voltages. Thus assuming supplies of +/-15V the output will be at one or other of these depending on the previous input history. Since the output is in a saturated state the gain will be very low and these two states will be stable. Thus the stated result that the system appears to act as an amplifier of gain 3 is simply a function of the supply voltages relative to the applied input. For an extended discussion of the operation of the circuit see the book An Analog Electronics Companion, (Cambridge University Press, ISBN  9780521687805, p252-255). My LM324 datasheet does not include the formula quoted in the post but I think the terms Voh and Vol are not the desired output voltages as assumed (i.e. 2V) but the limiting voltages as outlined above i.e. ~ +/-15V. To obtain the desired +/-2V output swing additional limiting will be required. Inputting a square wave as shown in the circuit diagram is not useful in showing the hysteresis and switching points; use a triangle or sin wave and you will be able to see the input switching levels and hence find the hysteresis. The use of such low value feedback resistors will mean the opamp output must supply a substantial current.