Yes, that's what I meant. Change the charge amps (current) to a lower value. It will take longer, but hopefully the voltage will be under the max value of 1.9-2 volts. Also using short fat charge lead will help and good connection to cell.

If you want to do it the other way (or even use a higher rate) then leaving the cell count to 2 is ok and use 3 mv instead of 6 mv for the peak detection. The charger SHOULD still look for a total drop of (2x3mv) 6 mv. In fact for a 2000 nicad you could go up to 10-15 mv per cell so leaving it at 6 mv wouldn't be bad either.

Watching the graph during the charge (especially right when it kicks off) may give you an indication of what's going on. The graph should be super smooth and if wiggling the battery or leads causes spikes in graph then look into bad connections of some sort.