yup, two changes.
first one was a while ago, we switched from a pnp to an npn on transistor T1 and bypassed the regulator, less load on the regulator and easier to relocate the backlight resistor outside the case and/or drive it from the instrument panel dimmer. It also meant that any energy stored in the capacitors would only go to the cpu/display and not to the backlight if there was a power dip, a-la "protect the brain", and that should help stability.
second we went to the low power lcd, which has a greener look to it. The lower power meant the backlight resistor (R1 below) was changed to about 450 ohms from about 81 ohms, IIRC.