Hi Ahmed,
in my opinion your first diagram is wrong and never will work, the second one corresponds to the diagram given in the National Semiconductor LM139... datasheet, March 2004, page 13 - Typical Applications, Two-Decade High-Frequency VCO. Nevertheless, the simulation results don't correspond to the datasheet information (i.e. 700Hz <= fosc <= 100kHz for Vcontrol from +0.25V to +50V) - they are far from it. (Note that the LM139 supply voltage is +30V (V+) with the V- pin grounded.)
'albbg' wrote you cannot use batteries to generate V+ but it's not true, you can - but for the simulation to work properly (oscillations to begin) you have to set the initial voltages of the capacitors to zero (IC=0); if you don't do it, the simulator calculates the bias point first, setting the capacitor's voltages to the calculated value and then it starts the transient analysis from this "steady" state (therefore there are no changes in such a circuit and no oscillations can begin).