The program takes the exponential of the existing ticks. It then adjusts them to get a sequence of 'pretty numbers.' Because of this, the spacing is never perfectly uniform, and sometimes this non-uniformity can be undesirable. In those cases, it helps selecting a larger minimum decimal position, because this will allow the program to choose more uniformly spaced ticks on a narrower grid.
Examples:
t = 0:30; y = exp(-t*0.03 + randn(1,31)*0.2);
plot(t,log(y)); grid on; logticks()
figure; plot(t,log(y)); grid on; logticks(gca,2)
t = 0:50; y = 1e6 * exp(t*0.05 + randn(1,51)*0.2);
figure; plot(t,log(y)); grid on; logticks
figure; plot(t,log(y)); grid on; logticks([],[],1e-6)