I tried this out on a SLC5/05 with unexpected results.
I tried my previous suggestion out on a SLC 5/05. It appears to work, but it in fact doesn't work.
First thing I did was to first program timer T4:0 with a 1.0 second time bas and then with a .01 second time base and look to see which bits were setting the time base. Bit 9 appears to be the one that sets the time base on the SLC5/05 (A Micrologix will need two bits because of the additional time base options) Here is a composite screen shot of the two timer configurations.
When bit 9 is clear, the time base is .01. When its set, it is 1.0.
One thing you will notice if you try this yourself online is that the low order byte of N7:0 is always changing. This is refleccted in the screen shots I made. This is a copy of the reference time stored in the timer. It is always updated, even when the timer is not running. See the technote 21729 for more information on this.
Next, I expanded the program to include logic to mainpulate bit N7:0/9 and write back to the timer.
Notice how bit N7:0/9 is set, and when copied back to the timer by toggling B3/201, the timer time base is 1.0. And when B3/202 is toggled, the time base is .01. This is updated and reflected in the data table for the timer. However, it is not reflected in the ladder. This didn't surprise me as I figured the logic in the display would be static.
As you can see, I first programmed a 10 second timer with a 1.0 time base. When I enabled the timer, it counted for 10 seconds and set the /dn bit as expected. Then I disabled the timer and toggle B3/202. The time base changed to .01 in the data file as expected, as shown in the third image. But when the timer was re-enabled, it still counted for 10 seconds. This was unexpected.
So I edited the ladder so that T4:0 had a .01 second time base. When enabled, the timer counted for .1 second and set the /dn bit as expected. And when I toggled B3/201, the time base was changed from .01 to 1.0 and the change was reflected in the data table. However, the timer still operated for .1 second, as programmed.
So I am left to conclude that in the case of a timer, the time base bits in the data table don't mean a thing, the instruction operates as programmed, not as the data table indicates. I doubt that any of the other methods suggested such as copying two timers to a third or using indexed addressing past file boundaries will work either.
So next, my question to the origianl poster is what is the customer trying to accomplish in chaning the time base? There are some timer tricks that can be done to create accurate long duraion .01 second timers, possibly eliminating the need to change time base.