Crossfade

[Soren Kuula]

Hi - I am a complete newbie to nixies.

I have seen some vids of a nice cross fade in tubes, where the next digit lights up as the previous one fades out.

Any implementation hints? I can imagine/guess:

- A series RC circuit from each digit to anode or ground would be expensive

- An RC circuit on the base/gate of each driver might have issues with heat?

- Will PWM work? In that case, are the 2 digits driven in phase?

Regards, Soren

[gregebert]

I would not use an RC circuit; not only would it be expensive, it would require a lot of trial-by-error experimenting and it likely would be susceptible to tube-aging and inherent differences between tubes. Stick with PWM.

I know for a fact that tubes of the same type, and even with same/similar date-code, have different I-V characteristics from data I've gathered on my own tubes.

Heat shouldn't be an issue for your drivers, unless you use a really high supply voltage or you use multiple devices in a single package. It's easy to calculate. For example, if you have a +200V supply, and the tube requires 150V to illuminate at 5mA, your driver will dissipate 250mW maximum. In reality, you will have anode and/or source/emitter resistors that will dissipate some of the ~250mW, so the heat dissipated by the driver will be even less.

My first clock is an extreme example: +340V supply, 160V across the nixie, leaving 180V across the 75K  anode resistor. That equates to 400mW; I played it safe and used 1W resistors instead of 1/2W devices. The driver, however, had essentially zero power dissipation because it's either on or off.

[Soren Kuula]

Hi, OK thank you very much! 

With heat problems, I was just thinking of a transistor not being used as a switch but linearly. For example, with a 50V drop across the transistor for some milliseconds * maybe 7mA = 350mW. Well that is not too bad over just milliseconds anyway.

I'll just try PWM.. an MCU with a lot of GPIO outputs, and a big pile of NPN transistors. They'll look nice in TO-92's if installed carefully. No  muxing or anything. Depending on firmware, it could be a clock or a general purpose display.

Regards, Soren

[gregebert]

I use shift-registers to control the individual driver-transistors in non-multiplexed mode. Then, I tie the shift registers into a serial chain so it uses only 2 pins on the controller (clock, data). So far I've used FPGA's for control, but I'm switching to a RasPi for the next project. Some shift-registers have output-enable pins, so you could use that for PWM, rather than trying to serially shift-in a PWM pattern.

[threeneurons]

On Friday, January 20, 2017 at 12:45:09 PM UTC-8, Soren Kuula wrote:

Hi - I am a complete newbie to nixies.

I have seen some vids of a nice cross fade in tubes, where the next digit lights up as the previous one fades out.

Crossfading was covered a couple of months ago:

 https://groups.google.com/forum/#!topic/neonixie-l/bh7kymv2xQA

I use PWM, but the fade pattern follows RC decay, encoded in a look-up table.

I do this because visual perception is not linear. My early display designs did use linear crossfading, so that the sum of old and new "ON-times" was constant. Using the RC model is non-linear, so that the sum of the old and new, is minimum, at the mid-point, when both are equal, but the blank period is maximum. Visually, this scheme looks nicer.