I recently acquired (free!) an obsolete frequency counter which uses a Nixie Tube display. I don't really have much need for a counter, so I decided that I would rip out the tubes and any other possible circuitry and make myself one killer desk clock. I now have read a ton on both Nixies and digital clocks, in preparation.

I was thinking to build the clock using basic decade counters with a crystal oscillator divided down to 1Hz- I have that part of the circuit pretty much figured out.

The Nixies require ≈200vdc to run. I was thinking that a voltage doubler from mains current would be the easiest way to get there- however I have no experience with them. Does anyone have any ideas for appropriate capacitor values? Any equations I can use? Current draw will never exceed 100mA, with 25mA typical. I can accept up to 25V ripple.

My other question is about using about supercapacitor backup for the timekeeping circuit, so I don't have to reset every time AC is lost. 1 Farad supercapacitors are actually pretty cheap, and I never want to have to replace a battery. I'm looking at powering about 10 CMOS chips at less than 1Hz, plus a few LEDs.
Does this power supply circuit make sense?

(in case you can't see, that's a 10microFarad filter cap.)
The notion is that when the power is on, the top PNP transistor stays off, the 1F cap charges, and everything goes happily. When the AC power goes off, the top transistor polls low, causing the LED to light as an indicator that the circuit is on backup. It also causes the bottom transistor to turn nonconductive, locking the ground rail and preventing the cap from just draining through the rectifier.

Thoughts?

 

Check out Ronald Dekker's flyback converter page here:
http://www.dos4ever.com/flyback/flyback.html
Nice design, and great info.
Check out his home page:
http://www.dos4ever.com/index.html

He has not only built a Nixie-tube clock, but a talking Nixie-tube alarm clock Complete blow-by-blow instructions, too.

 

You might consider just keeping the counters going and let the Nixie's stay dark while power is off. That should cut the energy budget.

 

You might consider just keeping the counters going and let the Nixie's stay dark while power is off. That should cut the energy budget.

That was the plan- sorry I forgot to mention it. The project will also include a "display off" switch for nighttime.

The flyback transformer is pretty cool, but again, doesn't a voltage doubler seem simplest?

 

I've run them off half wave rectified line voltage. Any size cap will charge to 175 volts, plenty for a Nixie. Aren't most of them spec'ed at about 180 volts?

I also seem to recall somebody (sorry, too long ago for a reference) placing a CDS cell in each Nixie's anode supply line. The increase in resistance caused them to auto-dim at night.

 

As for brightness, I much prefer full off control. I hate night lights so much that I reversed the polarization of my cheapo LCD clock. (Note: boredom might have also contributed.)

The frequency counter I'm scavenging this from has all the tubes and driver ICs on one board, which has BCD, 5v, 20v and 200v inputs. I would make the power circuits and clock, then just sent the BCD to the driver board. Therefore, the ideal is 200v for the high supply.

Where I live, line voltage alone is insufficient.

 

If you're using cmos you could lose the voltage reg as this will only waste power. The on-backup LED will probably consume more power than the clock. Also, from the schematic it's impossible to tell how the on-backup led is supposed to work.

It's probably worth adding a circuit to reset the clock when the cap voltage falls too low, otherwise when power is restored the clock may lock up.

 

Actually, what kind of power draw am I looking at? I just did some math which says that if I draw 10mA, the cap will only hold enough voltage to keep the ICs running for about three minutes. .

 

You can get an idea from the data sheets. Each device should have a figure for current usage, but the 1 ma per chip is pretty good for a rule of thumb. You can always add a cap. 4000 CMOS is very flexible - it works from 18 volts down to 3.

 

Some (hopefully) final questions:

One of the pins for the driver board is showing +20 volts. I figures it's a cathode pre-bias, even though the datasheet lists desired bias as 60v. Would it be ok to use a simple voltage divider from the +230, seeing as it won't actually draw any current?

Secondly, another pin show -0.01 volts. Any idea what this is?

Finally, would tying the grounds of the two power supplies together be fine to do?