I'd rather have 13 ULN2003's + zeners
Funny, this was my first thought when I finished watching Part 1. Then I asked myself: why Dave didn't mentioned this approach? There must be some issue with that. And then I realized what could be wrong with that.
As far as I understand the physics of gas discharge tubes, you have a sort of hysteresis there. To start conduction you need to apply voltage greater than
firing voltage (which is 170V for these particular tubes). Than the conduction won't stop until voltage drops below
minimal sustaining voltage (which is guaranteed to be not less than 120V according to the datasheet).
Unfortunately, this window (170V – 120V = 50V) is the same as ULN2003 max voltage and therefore is equal or even greater than our clamping voltage.
So, it is possible, that since we "fired" a digit in the tube, we will not be able to turn it off just by switching off corresponding channel of ULN, isn't it?
I mean that when using clamping approach, you are limited to have either 0V or 48.6V on a cathode pin of the tube so that voltage drop is either 170V or 121.4V, and the lowest one is above the
minimal guaranteed turn-off threshold.