Slow ionizing nixies

[Michel]

After making a couple of batches of watches, I noticed that some tubes
(about 1 out of 40) occasionally ionize very slowly when driven by a
low current. This doesn't always happen, sometimes they ionize just as
quick as other tubes, but occasionally it can take more than 1 second
before the gas ionizes. Once they ionize, they look just as bright as
the other tubes, there really is no difference. When I replace the
slow tube with another one, the problem disappears.

Why is that? I was thinking could the pressure of the gas inside the
tube be a bit too low after about 40 years not being used? But
wouldn't that influence the brightness as well?

Michel

[Jeff Thomas]

Good question.
There are a few factors involved. Varying penning gas pressure/mix, process variations in materials assembly and burn-in, and the possibility of residual oxygen. We know that KR85, if introduced to assist in ionization would be long gone.

I've experienced the same in testing volumes of other nixie tubes over the years, and I'd just chocked it up process variations when frit seal failure couldn't be the cause.

Regards, Jeff

[Michel]

Thanks Jeff, I assume there isn't really anything that can be done to
this then.

I could still use them for a clock I think as there isn't really
anything wrong with them once they ionize.

Michel

[jb-electronics]

Hi,

I was just thinking, if outgassing is a major factor here, shouldn't the
pressure inside the Nixie tube increase due to these extra molecules
floating around?

Also, if the pressure was too low the digit would become hazy while
increasing the pressure does not significantly alter the appearance of
the glow. Found this out while experimenting with a broken CD47 under a
vacuum bell and some Penning mixture.

Jens

[Michel]

I actually assumed that the pressure inside the envelope is normally
above atmospheric pressure, but of course if it is below then it would
indeed be the opposite.

Anyway, according to your findings, it wouldn't be related to a too
high or too low pressure and if it was caused by an additive that has
now been gone, the other tubes (from the same batch) should show the
same characteristics.

Michel

[jb-electronics]

Hi,

the pressure in a Nixie tube is roughly 1/25 atm.

> Anyway, according to your findings, it wouldn't be related to a too
> high or too low pressure

How do you draw this conclusion? The ignition process may take longer.
When doing the experiments, I ignited the tube at well-defined
conditions (i.e. 20 Torr) and then changed the pressure with the
discharge still going on. Thus, I can not draw any conclusion about the
igniting properties at different pressures, all I can say is that the
digit - once ignited - will look about the same at higher pressures.

Jens

[Michel]

So basically it could be that if the pressure is higher than what it
ought to be, it is likely to be not visible once the gas has ionized,
but it may still affect the ionization time?

The strangest thing what happens is that if I leave this tube off for
a while (say 15 minutes) the ionization process is usually slow. On
the other hand, if I trigger a time reading just a few times after
each other, the first reading might be slow but all the next readings
will be just normal.

Michel

[jb-electronics]

Hi Michel,

> So basically it could be that if the pressure is higher than what it
> ought to be, it is likely to be not visible once the gas has ionized,
> but it may still affect the ionization time?

yep, that was my point.

> The strangest thing what happens is that if I leave this tube off for
> a while (say 15 minutes) the ionization process is usually slow. On
> the other hand, if I trigger a time reading just a few times after
> each other, the first reading might be slow but all the next readings
> will be just normal.

Is it really strange? Once an ignition has been triggered, there are
much more ionized gas molucules around, thus making the next ignition
much easier.

Jens

[kay486]

I have noticed the exact same thing with all of mine LC-513 (Dolam) tubes. When i pop them in a clock on the seconds place, they all the numbers tend to flicker, some will have stable glow faster than others, it usualy doesent take longer than 30 seconds for all o them to glow fine. I asume that the multiplexing plays some role in this too.
Ive also seen some neon bulbs that have this sort of weir flickering on them (im not relating to that AC flickering) the glow of one of the bulbs in my clock tends to jump up and down on the cathode. If i have them turned off for a longer time before turning on, the flickering is really slow, it jumps up and down only like once per second, but soon after that it starts to excelerate to a poin when it all really unstable, it jumps up and down like six times per second! The weird thing is it happes only on one bulb! The other glows just fine. Ive seen this on some youtube videos too, so im not the only one. Does anybody know whats the problem with that?

[Michel]

Do you also drive them with a variable cathode current? If I drive
them with a reasonable current right from the start, they ionize just
as quick as the other tubes, or at least I don't see any difference.
But when I drive them on a low current it becomes obvious this 1 tube
is sometimes very slow.

Jens, you ask "is it really strange", but even if I do time triggers
with 1 minute intervals, the suspected tube still ionizes quickly.
It's only when I leave it off for say 15 minutes that I see a
difference. I would think that even after 1 minute the ionization will
have come to a complete rest, or not? It seems like it is temperature
related, that the gas needs to cool down for a relatively long time
before the problem shows up.

Michel

[kay486]

You can see the neon bulb flickering here for example, its not as fast as on my clock, but its the same thing. http://www.youtube.com/watch?v=5DzJSscfvtI

[threeneurons]

What supply voltage are you using ? Nixies will ionize faster at when subjected to higher voltage. With an externally (non-battery) powered nixie clock, you can adjust the supply voltage up, and compensate by varying the anode resistor appropriately. With your watch, monitor the current. Start with a high voltage (200V or so), then once you see current flow, then drop the voltage.

A few years ago, I ran an experiment, simulating multiplexing. I noted the "turn-ON" time versus the supply voltage. I have the real numbers somewhere, but the "turn-ON" time could be cut dramatically, by raising the voltage. If you just set it at 170V, you could be in trouble.

There are published charts of ionization probability versus supply voltage. That and my experiments, are telling the same thing. Up the voltage.

[Michel]

My circuit is designed in a way that should the tubes not ionize, the
voltage will rise to 200V. It could actually go higher than that but I
got my backlight LEDs in series with a 200V zener diode, so when
current starts to flow through the LEDs and zener diode, the voltage
no longer rises. When this specific tube doesn't ionize, the backlight
LEDs are on, so I know the anode voltage must be at least close to
200V.

Michel

[Michel]

I'm just thinking about something, is it possible that the speed of
the rising anode voltage voltage also has influence on the ionization
time? In other words, is the ionization influenced by dV/dt? I know
for a fact that in my circuit dV/dt will be lower when the target
anode current is lower, so climbing to 200V will take longer than when
the tube's target current is higher.

Michel

[jb-electronics]

Hi,

so 15 minutes is the cutoff where the inonization has become too low? I
just remembered I have the same problem in my very first IN-14 Nixie
clock, some digits usually take a few seconds to ignite when powering it
up after a longer standby time. I have no idea, though, how long the
minimum standby time is in order to reproduce this effect.

I do not know if temperature could be responsible. Room temperature
corresponds to some 25meV. Typical ionizing energies in Neon are in the
range of eV, see here:
http://en.wikipedia.org/wiki/Ionization_energies_of_the_elements_%28data_page%29

The outermost electron needs almost 41 eV to get ionized.

Jens

[marta_kson]

Just a thought as the off-time before the problem reappears seems to
be very long, may the mercury in the tube be involved in some way?
That would take minutes to condense. The loss of ionization is a
microsecond process, so the explanation must be something else. The
mercury insertion is also something that could had some process
variations at the manufacture making some tubes worser than other even
in the same batch.

[Michel]

I did some extensive testing today and made the conclusion that these
slow tubes need much more ambient light to ionize than other tubes.
Raising the anode voltage to 200V rather than 170/180 is not doing
much at all, I only see a dramatic improvement when I expose them to
more ambient light. Could that be related to the amount of mercury
vapor then? Is it then just a badly manufactured tube or is that due
to its age?

Michel

[jb-electronics]

Hi,

it looks to me that the 20V difference does not do much concerning
ionisation seeds, which seems plausible. Ambient light, however, carries
energies of about 2-5 electron volts per photon. When the intensity of
ambient light is large enough, these photons can do more together than
the 20eV acting on the gas by the higher voltage. This way, the ambient
light photons create ionisation seeds.

Once there are seeds available, the potential dofference of 20V kicks in
remarkably, of course, just as you experienced.

So it all comes down to the fact that your "bad" tubes apparently do not
have as much default seeds as the good ones, which might be due to a
smaller concentration of mercury.

Jens

[marcin]

It is necessary to somehow ionize the gas a bit. Light is good, radiation is. Some people - eg. this guy http://eeberfest.net/gallery.php?set=thebox use UV LEDs to ensure ionization in dark conditions. Maybe it would be possible to somehow use your blue LEDs? I have no idea if they would be sufficient and how much time they would need to sufficiently ionize the gas. Just an idea.
Marcin

[Michel]

Hi Marcin,

That is exactly right, the blue backlight LEDs can indeed be used for
this purpose. They do already come on if the tube doesn't ionize but
the LED current in low ambient light is very low so there's not many
photons emitted. I changed my code a bit yesterday so that I can
inject a variable boost current for the initial 160us, 320us, 640us or
1280us. I am trialing this at the moment and so far the results are
really good.

Michel

On Sep 12, 6:40 am, marcin <marcin.r.adam...@gmail.com> wrote:
> It is necessary to somehow ionize the gas a bit. Light is good, radiation

> is. Some people - eg. this guyhttp://eeberfest.net/gallery.php?set=thebox

[Michel]

This does pretty much do the trick, so that point is proven. I just
don't feel happy with the solution knowing the tube doesn't perform as
good as the other tubes, so still decided to replace it :-).

Michel