Nixie tube driving problems

[maros]

Lately encouraged by Dave's Nixie display project I've managed to finish my own nixie clock which I was planing to build for a couple of years.

I've built high voltage power supply based on a famous Pile-O-Poo project which gives me a nice and stable 180V output voltage.

For driving nixie tube cathodes I've used exactly the same IC like Dave used: TPIC6B595
 8-bit shift register. Output drains are connected straight to nixie cathodes. Nixie (IN-8) anode is connected through a 30k resistor to high voltage power supply. The shift register clock, data, enable, VCC and ground lines are connected to Arduino.

Unfortunately not all things went as I expected. I'm facing with an issue. When I set the shift register output the nixie tube number is glowing nice and bright (the current is under 1mA), but when I turn off the output the number is still partially glowing. The voltage difference between ground and drain is about 50V. It looks like there's a current leakage between source and drain (Idss). How to prevent it? I've tried to rise the resistance on nixie anode, but it results now with partially glowing cathode when the output is set.

I'm surprised that Dave didn't have such a problem with his nixie display because I did everything the same like he did. Only the tube I used is different (IN-8). But they seems to have similar parameters.

Can someone give me a clue what do I suppose to do to fix this issue?

[Rbastler]

I had the same problem when I made tests with BC546 as switches. The problem was solved with a pulldown resistor at the base.

Sent from my A0001 using Tapatalk

[Ian.M]

A TPIC6B595 has avalanche diode clamping from each drain to ground with a clamping voltage slightly above its output MOSFET max rated V_DS of 50V.   If you apply more than 50V to any drain it *WILL* conduct current to ground, even when 'Off'. Therefore if you turn off all the digit cathodes, you will still have up to 130V across the tube (180V -50V - resistor drop) and that is enough to sustain the discharge to the last digit cathode that was on .   The *ONLY* way it can work is if there is less than 50V differential between striking and running for your tube, and you *ALWAYS* have one digit on.

If that's not the case you'll need HV (at least 200V V_CEO or V_DS) NPN or NMOS transistors in cascode with all the TPIC6B595 drains.   Their emitters or sources go to the TPIC6B595 drains and their bases or gates go to a common bias rail.   For cascode MOSFETs, the gate bias rail is usually ground. For cascode NPNs, the base bias rail is positive - you can use your logic supply, but then any transistor failure can take out all your logic, so its better to use a 5V1 Zener fed via a resistor from your unregulated DC in.   To provide enough base current, the resistor should be chosen to pass  a current of at least the total load on the HT supply divided by the cascode transistors' minimum h_FE

[mbless]

I suggest you rewatch Dave's videos then. He used a lower voltage, 160V IIRC, so the voltage across the Nixie was less than the sustaining voltage when turned off.