[MrFixr55]

+1 on Terry's (RadioRoslyn's) comments.

Generally speaking, the local oscillator frequency runs higher than the desired RF frequency by a value equal to the IF value. For example, for your set with a 450KHz IF, the oscillator has to run 450KHz higher than the desired RF signal.

You state that you "get strong IF oscillations from antenna to speaker". Do you mean that you injected a 450KHz signal with 400Hz modulation from a signal generator?

Does the oscillator section of the tuning cap have the same amount / size of plates as the RF section? I am not sure what range of capacitance the Miller 70 coil is supposed to run at, but if designed to run on a 365pF (with plate closed) cap (the older standard where the osc and RF sections of the tuner and the same size and capacity) and your oscillator section of the tuning cap is smaller than the RF section (the newer style), this may be the issue. With the smaller section of the tuning cap, the oscillator freq may be too high. The reverse may also be true. (I don't know if a lower capacity padder would fix that.

You state that when measuring the grid voltage, the oscillator shuts down. Do you mean G1, the Osc grid on the 6A8, Pin 5 or do you mean the grid cap, G4 (the RF grid). It is possible that measuring the bias on G1 (The oscillator grid, Pin 5) could shut down the oscillator.

Bottom line, your oscillator is running about 700KHz higher (or lower) than the desired signal, instead of 450KHz higher, which it should do. You need to change parts (padder?) to drop this to a value that is 450KHz higher than the desired tuned value.