nixie clock using trigger tubes

550 views
Skip to first unread message

Pieter-Tjerk de Boer

unread,
Dec 24, 2020, 1:13:10 PM12/24/20
to neonixie-l

I recently completed a nixie clock without any transistors, microcontrollers or other ICs, instead using trigger tubes to implement the digital counters:
  https://www.youtube.com/watch?v=E3PBJUcKeoo

Actually, I'm not even using the 'trigger' function of these tubes, they are just used as neon lamps with a large difference between strike and maintaining voltage. It's this difference which allows one to build counters with them. I'll publish a more detailed description on my website later on.

Some may know that I built a similar clock over ten years ago:
  http://pa3fwm.nl/projects/neonclock/
That clock used regular NE-2 style neon lamps for the logic; unfortunately it became unreliable, as the lamps' properties changed with time. I hope the new clock will turn out to be more reliable...

Regards,
  Pieter-Tjerk

Konstantin

unread,
Dec 24, 2020, 1:54:46 PM12/24/20
to neoni...@googlegroups.com

Cool! Very unusual clock! Do you use  MTX90?  Do you think this version is more stable/reliable than your previous one on  NE-2 bulbs?

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/e9808ce4-c216-4771-a1e0-42faa4db2a3dn%40googlegroups.com.

Mac Doktor

unread,
Dec 24, 2020, 6:23:30 PM12/24/20
to neonixie-l
On Dec 24, 2020, at 1:13 PM, Pieter-Tjerk de Boer <pa3...@amsat.org> wrote:

That clock used regular NE-2 style neon lamps for the logic; unfortunately it became unreliable, as the lamps' properties changed with time.

I have several power strips with flickering pilot lamps in the rocker switches and a couple that are completely dead. I've always assumed that the cathodes in NE-2s aren't the greatest.


I hope the new clock will turn out to be more reliable...

I'd love to know how these Soviet era neon indicator lamps (including the thyratrons) hold up over time and how well the thyratrons would work using a low-voltage trigger. Can they safely be driven from a 5V chip or would opto-couplers be necessary?


Terry Bowman, KA4HJH
"The Mac Doctor"

Q: Should car stereo speakers be pointed to the rear for more thrust or up for more traction?

A. On long trips, the 20- to 30% improvement in gas mileage you might get with speakers pointing to the rear is certainly worthwhile. On the other hand, if you drive on snow or ice, the extra traction of speakers pointing upward gives you added control.

Don Lancaster

Paul Andrews

unread,
Dec 25, 2020, 8:28:58 AM12/25/20
to neonixie-l
That’s a beautiful thing.

newxito

unread,
Dec 25, 2020, 11:27:07 AM12/25/20
to neonixie-l
Great project! Looking forward to reading the details on your website, I would like to understand how this clock works.

Pieter-Tjerk de Boer

unread,
Dec 25, 2020, 1:43:51 PM12/25/20
to neonixie-l
On Thursday, December 24, 2020 at 7:54:46 PM UTC+1 kosbo.com wrote:

Cool! Very unusual clock! Do you use  MTX90? 

Yes.

Do you think this version is more stable/reliable than your previous one on  NE-2 bulbs?

For now, I can only say I hope so!
At least the circuit is much less critical now. With the NE-2, I needed to characterize them and sort them, and in each ring counter only use bulbs with matching striking and maintaining voltages. With the MTX90, I only quickly tested them to avoid out some outliers and a very few that didn't work at all, the rest worked fine in the ring counters without further matching. There's a large spread of striking voltage among them, but the difference with the maintaining voltage is larger still. With the trigger electrode not connectected, the striking voltage is (far) above 200 V, and the maintaining voltage around 60 V. Compare that to about 80 V and about 60 V for the NE-2s I used back then.
But only time will tell if and how this changes with aging...

Regards,
  Pieter-Tjerk

Joe Croft

unread,
Dec 25, 2020, 8:59:04 PM12/25/20
to neoni...@googlegroups.com
Nice clock Pieter! I like your Skeletal' layout. I always liked making wire circuits :)

I designed and have a couple of NE-2 ring counter clocks and yes, every so often, I get to change out a ring to keep the clock running. No fun at all!

-joe

Pieter-Tjerk de Boer

unread,
Dec 29, 2020, 12:02:13 PM12/29/20
to neonixie-l
As promised earlier, more detailed information, including schematics, about my clock is now on my website:

Regards,
  Pieter-Tjerk

Yohan Park

unread,
Dec 29, 2020, 4:43:29 PM12/29/20
to neonixie-l
Very cool clock!
I assume the Reed switches are there so you can use a magnet to set the correct time?

Dekatron42

unread,
Dec 31, 2020, 6:00:07 AM12/31/20
to neonixie-l
Really nice!

Thanks for sharing the circuit diagram!

Did you try to use them as proper trigger tubes and not just as neon tubes?

/Martin

Pieter-Tjerk de Boer

unread,
Dec 31, 2020, 8:12:34 AM12/31/20
to neonixie-l
Indeed. Since the magnet works through the glass, that's an easy way to safely add switches to the circuit (which is directly connected to the mains).

Pieter-Tjerk de Boer

unread,
Dec 31, 2020, 8:21:16 AM12/31/20
to neonixie-l
Yes, at first I tried to use them as proper trigger tubes, in multiple ways, but I couldn't find or come up with a circuit that worked reliably over a range of supply voltages (which is also an indication for how robust the circuit is for variations of the tube properties). So I went back to the neon tube counter circuit, and found that to work more robustly, thanks to the very large difference between striking and maintaining voltage of these tubes. Somehow, this seems "wrong",
as the trigger electrodes are there precisely to make this kind of circuits easier, but...

Regards,
   Pieter-Tjerk

Mike Mitchell

unread,
May 2, 2021, 7:59:44 AM5/2/21
to neonixie-l
I've built two clocks out of the MTX-90 tubes, following Pieter-Tjerk de Boer's schematic.  The longest I've gotten one to run is about a week, at which point I have to change out tubes.  The clock runs fine in the daylight but some random tube will stop firing in the dark.  I'm thinking about sprinkling some blue or green "neon" tubes throughout the clock just to provide some extra photons.  Something like these:  https://www.amazon.com/Othmro-Pieces-6x16mm-Bright-Indicator/dp/B07WFNSKSM/
I do have some near-UV LEDs (400nm) but the clock is line powered and I don't really have space for a buck converter.  I hate to drop nearly 300 volts across a resistor just to light a string of LEDs.

Does anyone have any other ideas?

Thanks!
Mike



NeonRing.jpg

Dekatron42

unread,
May 2, 2021, 8:09:23 AM5/2/21
to neonixie-l
Have you made any experiments with putting them closer to each other so that the one that is lit always shines on the next one in turn to light up - this would need an extra tube at the first position connected to the last tube so that it would trigger at the same time and so that there would be some glow on the first tube when the last one is lit, or you'll have to put them in a proper circle so that there is always glow falling on the next tube in turn.

/Martin

Dekatron42

unread,
May 2, 2021, 8:09:55 AM5/2/21
to neonixie-l
Forgot to say that it is a really nice clock!

/Martin

Pieter-Tjerk de Boer

unread,
May 2, 2021, 3:45:12 PM5/2/21
to neonixie-l
Nice that you managed to duplicate my MTX-90-based clock and get it to work!

If I saw it correctly, your clock has 59 tubes while mine has 56. So apparently you changed the circuit a bit?

My clock became unreliable in darkness a week or so after completion. Adding some blue LEDs fixed this, but not for long. One tube needed to be replaced. It turns out this tube was an odd one out: it already at the start had the lowest striking voltage among all tubes in that particular ring counter, and this had drifted down by some 10 volts, while the other tubes had drifted less and mostly up. Since replacing that tube, the clock has worked reliably, except for one (so far) hickup.

The blue LEDs are still in the clock, it seemed prudent to leave them in even if they were perhaps not needed anymore after the tube replacement.

As for powering the LEDs, the easiest is to put them in series with the entire clock circuit (with some protective components, otherwise they may not live long - I learnt this the hard way in my earlier NE-2 based clock). I've updated the schematic on my website to show how I connected the LEDs, see https://www.pa3fwm.nl/projects/neonclock2/

Regards,
  Pieter-Tjerk

Pieter-Tjerk de Boer

unread,
May 2, 2021, 3:50:52 PM5/2/21
to neonixie-l
I tried this back when building my NE-2 ring-counter based clock, and there it didn't work. The red neon light simply didn't help the striking process, in contrast to blue light.
Note that we're talking about two totally different processes here. The light emitted by the neon gas has a wavelength (colour) determined by the physical properties of the neon atoms, while the light needed to help the striking process must have photons with sufficiently high energy (sufficiently blue colour) to kick an electron loose from the electrodes, which depends on the physical properties of the electrode material.

Regards,
  Pieter-Tjerk

Grahame

unread,
May 2, 2021, 3:53:03 PM5/2/21
to neoni...@googlegroups.com

Hi

Consider using a capacitive dropper if you have AC easily available.

My first ring clock used XC18 tubes which were originally radioactive doped. It failed overnight if left in the dark but ran ok if lit. The second clock uses Z700U which have a priming electrode and that runs in the dark. I have found that some tubes are initially out of spec but come in to spec when run for a while. Some tube drift out of spec and the rings can then fail. I think some ageing and measurements is a worthwhile step ahead of soldering them in.

Well done to you both for an interesting ring clock.

Grahame

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.

Mac Doktor

unread,
May 2, 2021, 4:50:02 PM5/2/21
to neonixie-l

On May 2, 2021, at 7:59 AM, Mike Mitchell <mmit...@gmail.com> wrote:

Does anyone have any other ideas?

Put a strong beta or emitter in the center of the device or dust the tubes with some thorium-laced lantern mantles. X-rays might work as well but then you need a power supply over 15kV.

Tidak Ada

unread,
May 2, 2021, 5:07:11 PM5/2/21
to neoni...@googlegroups.com

Did you already try UV-LED’s? Banggood and Ali sell COB-chips up to 365nm wavelength 10 to 50W power and also a 365nm cut-of filter. That gives you the advantage of no visible light for small money, However, you don’t need the full power of a 10W COB-chip of course, what saves power if used with a lower current

 

Cheers,

eric.

--

You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.

Toby Thain

unread,
May 2, 2021, 5:41:33 PM5/2/21
to neoni...@googlegroups.com, Tidak Ada
On 2021-05-02 5:07 p.m., Tidak Ada wrote:
> Did you already try UV-LED’s? Banggood and Ali sell COB-chips up to
> 365nm wavelength 10 to 50W power and also a 365nm cut-of filter. That
> gives you the advantage of no visible light for small money, However,
> you don’t need the full power of a 10W COB-chip of course, what
saves
> power if used with a lower current

I suspect there might be some safety implications of high power UV leds
in an appliance project...

--Toby



Mike Mitchell

unread,
May 3, 2021, 7:31:36 AM5/3/21
to neonixie-l
There are more tubes in my clocks because I'm using 6-tube and 10-tube rings to divide 60 Hz to 1 pulse per minute.  My cloches are not as tall as yours so I used fifteen rows of five tubes instead of seventeen rows of four, giving plenty of space for more tubes.  I figured the life of the tube was related to its on time, so having more tubes in a ring should increase the time before replacing tubes.
I'm using four brass posts for support.  The ladder in the front is +300v, the ladder in the back is ground.  The rungs in the back are offset vertically so the bottom of the tube rests on the front rung and the cathode is connected to a slightly higher rung in the back.  Two front-to-back diodes interconnect the front ladder and the rear ladder, a pair of diodes on each side.  On each side is a filter capacitor between the two ladders.  High up between the two ladders are discharge resistors, one per side, giving the four posts more stability.  The AC mains are connected through a fuse and a current-limiting resistor to the center connection of the front-to-back diodes on each side.  I left off the bottom two rungs of the rear ladder and built a third ladder slightly narrower with only two rungs.  The smaller ladder in the rear is connected to the current-limited AC line.  The first divide-by-ten ring sits in the bottom, using the AC line as its counting input.

I did several tests to see if having 300v near the front of the tube would affect firing.  300v near the back of the tube did cause the tube to fire earlier, but had no effect at the front.  I did find that the front ladder isn't needed to support the MTX-90 tubes.  The rear ladder is sufficient for the MTX-90 tubes, their wires are stiff enough for support.  The Nixies at the top do need the support at the front.

Mike   

Mike Mitchell

unread,
May 9, 2021, 7:34:19 AM5/9/21
to neonixie-l
I might have gone overboard.  I added fourteen blue "neon" bulbs, seven on each side, placed horizontally every other row of the MTX-90 tubes.  The clock kept good time over night in the dark.

Mike

NeonRing+Blue.jpg

Pieter-Tjerk de Boer

unread,
May 13, 2021, 4:25:19 AM5/13/21
to neonixie-l
Nice that this has solved the problem! I do like the use of the blue "neon" bulbs for this, although 14 seems quite a lot. In my clock, I have (so far) only 2 blue LEDs. I have four columns of MTX-90s, and each led is strategically placed at the top, in between two such columns.
B.t.w., thanks for your (earlier) detailed description of how you built yours. I'm not sure your cloches are really less tall than mine, as my MTH-90s are closer together. They actually touch each other, as I don't have ladder rungs in between them. Other than touching each other, they don't have any support on the front side, they just hang off their (rather stiff) wires at the back. My clock's frame consists of only three vertical brass posts, arranged in a triangle: two to which the MTH-90 cathodes are soldered, and one even farther back mostly for mechanical strength.
Regards,
   Pieter-Tjerk

Mike Mitchell

unread,
May 14, 2021, 7:14:07 AM5/14/21
to neonixie-l
It only solved the problem for two days.  Right now I have two near-UV LEDs shining on the cloche in addition to the blue "neon" bulbs and that seems to have fixed the problem.  Next I'll try increasing the current to the blue "neon" bulbs.  The vendor says to use a 150K resistor with 220 vac, I'm using 510K with ~300 DC.  I'll try 390k in a few strategic locations.
A few years ago I took apart a faulty Philips LED Edison-base bulb.  It had three modules of three blue LEDs and three orange plastic diffusers, like this one: https://www.bulbamerica.com/products/philips-enduraled-8-watt-a19-dimmable-light-bulb-40w-equivalent.  I'll try a scavenged LED module at the base if I can't get the the blue "neon" bulbs to work. 

Mike
Reply all
Reply to author
Forward
0 new messages