Can anode current limiting be replaced with cathode current limiting?

[Allen Dutra]

Greetings Neonixie group,

Anode current limiting seems to be a fundamental design element when using Nixie tubes. Yet is there anything fundamentally wrong with limiting cathode current instead of anode current, assuming a design will reliably turn on one cathode at a time?

If my math and understanding checks out, one should be able to accurately limit cathode current with a common collector transistor circuit and eliminate the anode resistor entirely. The circuit shown bellow controls current based on the equation: Ic = Ie = (Vb - Vbe)/Re

Controlling a Nixie this way should allow a designer to, eliminate the anode resistor, ignore the voltage drop of the Nixie tube, and not be concerned with the exact value of anode voltage. The only variables that really matter are base voltage (Vb), base-emitter voltage drop across the transistor (Vbe), and emitter resistance (Re). Control these three variables and Collector, Emitter and Nixie currents should all be limited to the same value. The only requirements for anode voltage being; anode voltage needs to be greater than the Nixie tube striking voltage, the collector emitter voltage rating of the transistor needs to be greater than the anode voltage and the transistor needs to dissipate the heat generated by the voltage drop. Overall, keeping anode voltage somewhat close to the Nixie tube striking voltage will improve efficiency. 

Are there any fundamental issues with the design direction I'm considering?

Thanks,

Allen Dutra

[David Forbes]

Allen,

This should work. It's not done because it uses more parts than does anode current limiting. 

My Nixie watch design uses a current limited power supply, which is a more efficient way to do this, but only in a multiplexed design.

[gregebert]

No difference; current is current, and that is what determines how bright the glow will be. However, the circuits you posted are quite different. The anode-resistor provides current-limiting, but no current regulation: If the supply voltage or the tube's voltage drop change, so will the anode current. This can be undesirable, especially if it results in higher-than-spec anode current which can lead to premature failure.

The cathode circuit you showed will regulate the current, irrespective of the anode voltage or tube condition.

You can also use a similar regulation technique at the anode; you would use a PNP or PMOS (my preference) device.

I've used all of these techniques in my designs, and for the 7971 nixie clock I recently finished, I use both anode AND cathode current regulation.

If you are using plentiful and inexpensive tubes, such as IN-1's, IN-8's, an anode resistor is fine. But if you are going for larger tubes like IN-18, 7971, or Dalibor's I would definitely use a current regulator.

[Robert L]

Individual cathode current limiters work great... Stole the idea from Greg's B7971 design. Note the transistors and current set resistors lining the display in the photo below. First light today on this Sperry SP-151 Panaplex.

In this case, I'm using individual cathode limiting due to widely differing amounts of current needed by the various segments for uniform intensity... not to mention the impact of varying numbers of segments for different displayed values using basic anode limiting. Note the uniform glow? :-)

These are running a test with a single one and 31 zeros circulating at high rate around the 32 element shift register that drives the transistor bases. 

Will use the same design with a Telefunken ZM 1350 display that's also in the works. Same issues and concerns a the B7971.

Best regards,

Bob

[Robert L]

One other thought Allen...

What I believe is a simulation schematic shows 1 meg pull-downs on the base lines. These may not be necessary depending on other elements of your circuit in the real world.

My current design uses 595 shift registers and the outputs are high-Z when disabled (Output Enable not asserted). There is indeed a small amount of leakage current driving the transistor base lines when the 595 outputs are in high-Z - the display has a very faint glow on just a few segments that is not really visible in any but a dark room.

The thing is, the faint glow won't matter in my design. The only time the outputs are in high-Z is during duty cycle modulation to control display intensity. The faint glow during the high-Z portion of the driving signal makes no difference. It's just not visible.

Segments are solid on or off when the 595 outputs are driven. My circuit drives hard zeros when the display is dark but powered. I disable the HV power supply when I want the display dark and in an energy saving state.

I opted not to include the pull-downs.Your mileage may vary!

Again, best regards,

Bob 

[gregebert]

If you want to make this rock-solid, the pulldown resistor can be calculated knowing the  max tristate leakage current (usually 10uA) and ensuring that 10uA of leakage does not produce enough base-current to turn on the transistor.

Worst-case, you will have a transistor with infinite current-gain (beta). Then the problem is to prevent voltage at the base to start turning on (0.7V). That works out to 70K. This is approximate, because there is additional leakage current at the transistor itself (collector-base leakage), which is typically much less than 1uA.  This is just a quick-and-dirty calculation and in a real circuit the pulldown resistor will have a  higher value; if you factor-in transistor beta (lets pick 100) and take a swag at how much current is required before the nixie starts to glow (my experiments with b7971 tubes show 100-150uA is needed), an emitter resistor of about 1.5K, the pulldown resistor works out to 94K, which is pretty close to the quick-and-dirty value.

[Allen Dutra]

Thanks all for the wonderful input and confirming cathode current limiters is a valid replacement for passive anode current limiting, even though the latter is far more common. Despite the active cathode current limiters requiring more parts and cost, the benefits are worth it for the projects I have in mind.

Side note, good looking Sperry SP-151 display you have there Robert L. The rectangluar neon segment displays always looked awesome to me.

Additionally, good call on the pull-down resistor, that was indeed an extra part added to make sure the simulation worked correctly. In reality 595 outputs will a controlled high or low and never set to the High-Z state. Hence pull-down not needed in reality as pointed out.

To give some added background, I want to add IN-12 or other common Nixie tubes to devices that already have high voltage supplies. In practice if some vintage hardware is already rectifying grid power to create high voltage DC source generally 25mA can be siphoned off for a Nixie display and not affect the vintage device. Of course exact implementation will assessed on a case by case bases for best results.

Thanks again for the input everyone,

Allen