I had to think a little about that one. The easiest way to drive them, is using an NPN, with an emitter resistor, and sending the input to the base. Then the emitter will follow the base. The emitter current will follow that signal, as a function of voltage across the emitter resistor. The collector is roughly the same as the emitter current, less the base current, which can almost be ignored.
The ramp, on this circuit, is always negative, relative to the analog portion's common (AREF). Maximum signal, is most negative, relative to AREF. The bar length is roughly proportional to the current flowing thru them. So instead of driving the base, I tied that one to near AREF, and fed the signal to the emitter. But the emitter has a lower impedance, and needs more drive current. That's why I added the PNP, as an emitter follower. The signal is fed into the base of that PNP. Its emitter follows the input. The base of the HV NPN, is fixed at roughly two diode drops (~1.2V) higher than AREF. The voltage drop, then, across the 1K resistor between these two transistors, roughly equals the ramp voltage. The current then thru both, and thru the bargraph is roughly that voltage divided by 1K. I designed the integrator to peak at -5V (relative to AREF), with 1.5V in (or 150 F) from the LM34. That translates to 5mA, which will "peg", the IN-13.
I like the IN-13 over the IN-9, mostly because it uses less current. The aux cathode helps keep the bar at one end. Its no guarantee, however. A friend of mine has a simple IN-13 winker, which now has its glow floating around the tube. These are odd tubes, which are designed to partially glow. It jut looks like a recipe for cathode poisoning.