Folks,

Here is a repeat of a message I sent to this list early on in its life...
-------------------------
I've been looking for the answer to the Nixie lifetime qustion - is
multiplexing going to shorten or lengthen tube life?

I finally found an answer, by reading about neon lamp lifetime.

According to several sources, the standard neon lamp has a lifetime
proportional to 1/I^3.3 where I is the operating current. This means that
if you run the lamp at twice the recommended current, then the lifetime
will be reduced by a factor of ten.

For a multiplexed clock display, if you run a 6-digit display at 1.7 times
the recommended current, then the display will last just as long as a
non-multiplexed display running at the recommended current.

I have found that a current of 1.6 times the recommended value (using
ZM1010 or ZM1020 or ZM1040 tubes) puts out enough light for a multiplexed
6-digit display used indoors.

So there you have it. The tubes in a multiplexed clock should last the same
number of years as those in a non-multiplexed clock.


--David Forbes, Tucson, AZ
http://www.cathodecorner.com/

##### level-1 quote by David Forbes (on) 01.04.02 (at) 09:19 -0700

Folks,

Here is a repeat of a message I sent to this list early on in its life...
-------------------------
I've been looking for the answer to the Nixie lifetime qustion - is
multiplexing going to shorten or lengthen tube life?

I finally found an answer, by reading about neon lamp lifetime.

According to several sources, the standard neon lamp has a lifetime
proportional to 1/I^3.3 where I is the operating current. This means that
if you run the lamp at twice the recommended current, then the lifetime
will be reduced by a factor of ten.

For a multiplexed clock display, if you run a 6-digit display at 1.7 times
the recommended current, then the display will last just as long as a
non-multiplexed display running at the recommended current.

I have found that a current of 1.6 times the recommended value (using
ZM1010 or ZM1020 or ZM1040 tubes) puts out enough light for a multiplexed
6-digit display used indoors.

So there you have it. The tubes in a multiplexed clock should last the same
number of years as those in a non-multiplexed clock.

That's the one, David, thanks for re-posting it. (Sometimes classic posts are worth re-posting rather than sending multiple people on a lengthy search of the archives, especially when it is topical.) --ray

--- In NEONIXIE-L@y..., Raymond Weisling <fire.mountain@p...> wrote:

##### level-1 quote by David Forbes (on) 01.04.02 (at) 09:19 -0700

So there you have it. The tubes in a multiplexed clock should
last the same number of years as those in a non-multiplexed clock.

That's the one, David, thanks for re-posting it. (Sometimes classic
posts are worth re-posting rather than sending multiple people on a
lengthy search of the archives, especially when it is topical.)

It might not be so bad if doing an archive search would paginate
based on number of results returned from the entire archive, rather
than arbitrarily breaking it up into (400+ message) chunks beginning
with the most recent message. If you do a search for something which
only appeared in an early message, you have to click through page
after page CONTAINING NO RESULTS until you finally find the message
you wanted. Do you think there's a snowball's chance that Yahoo
might do something about this if we petition them?

A.J.

--- In NEONIXIE-L@y..., David Forbes <dforbes@a...> wrote:

Folks,

For a multiplexed clock display, if you run a 6-digit display at

1.7 times

the recommended current, then the display will last just as long as

a

non-multiplexed display running at the recommended current.

So there you have it. The tubes in a multiplexed clock should last

the same

number of years as those in a non-multiplexed clock.


--David Forbes, Tucson, AZ
http://www.cathodecorner.com/

But, to my Experiences with Muxed LEDs, the Current has to be much
higher: if you multiplex 6 Digits, the current is on for max. 1/6th
of the total time, so has to be 6 times normal current. (Same applies
for 4-digit: replace every 6 with 4) The effective Current is then
the same!
I asked a few 'Experts' about this, but none has Experience with
tubes. Does this rule of thumb not apply to them????

At 2:50 AM -0700 4/2/02, ptsearcher2001 wrote:

--- In NEONIXIE-L@y..., David Forbes <dforbes@a...> wrote:

Folks,

For a multiplexed clock display, if you run a 6-digit display at

1.7 times

the recommended current, then the display will last just as long as

a

non-multiplexed display running at the recommended current.

So there you have it. The tubes in a multiplexed clock should last

the same

number of years as those in a non-multiplexed clock.


--David Forbes, Tucson, AZ
http://www.cathodecorner.com/

But, to my Experiences with Muxed LEDs, the Current has to be much
higher: if you multiplex 6 Digits, the current is on for max. 1/6th
of the total time, so has to be 6 times normal current. (Same applies
for 4-digit: replace every 6 with 4) The effective Current is then
the same!
I asked a few 'Experts' about this, but none has Experience with
tubes. Does this rule of thumb not apply to them????

Odd. My experience shows that you don't need anywhere near 6 times the
current to achieve a usable display; on the contrary, 6 times the current
would cause the tubes to act funny.

There is a brain phenomenon called persistence of vision that comes into
play here, which makes the bightness-vs-duty-cycle curve nonlinear.

I don't believe that LEDs need anywhere near N times the current either.
For example, a typical 8-digit LED calculator certainly does NOT require
100 milliamps per segment - if it did, then the batteries would last only
15 minutes! I believe that they typically use about 2-3 times the
non-multiplexed current in calculators.

The best way to find out for yourself how much current is really needed is
to take a Nixie tube and run it at a fractional duty cycle, then vary the
current with a variable resistor and look at the tube. You may be surprised
at what you find.


--David Forbes, Tucson, AZ
http://www.cathodecorner.com/

Odd. My experience shows that you don't need anywhere near 6 times the
current to achieve a usable display; on the contrary, 6 times the current
would cause the tubes to act funny.

There is a brain phenomenon called persistence of vision that comes into
play here, which makes the bightness-vs-duty-cycle curve nonlinear.

I don't believe that LEDs need anywhere near N times the current either.
For example, a typical 8-digit LED calculator certainly does NOT require
100 milliamps per segment - if it did, then the batteries would last only
15 minutes! I believe that they typically use about 2-3 times the
non-multiplexed current in calculators.

The best way to find out for yourself how much current is really needed is
to take a Nixie tube and run it at a fractional duty cycle, then vary the
current with a variable resistor and look at the tube. You may be surprised
at what you find.

This is good -- it begs a side-by-side A-B test, both of nixies and LEDs. One of the devices is run at a DC current and the other is subject to a variable duty cycle and variable current. Adjust them for identical apparent intensities, and do so at different current levels for the DC drive one as the "standard". Who wants to do it?

It should be done both in a dark room and in a bright room, as the eye will likely exhibit different time-decay persistence phenomenon depending on the amount of visual purple and whether the receptors are predominantly rods or cones.

Ray

--- In NEONIXIE-L@y..., David Forbes <dforbes@a...> wrote:

Odd. My experience shows that you don't need anywhere near 6 times

the

current to achieve a usable display; on the contrary, 6 times the

current

would cause the tubes to act funny.

There is a brain phenomenon called persistence of vision that comes

into

play here, which makes the bightness-vs-duty-cycle curve nonlinear.

I don't believe that LEDs need anywhere near N times the current

either.

For example, a typical 8-digit LED calculator certainly does NOT

require

100 milliamps per segment - if it did, then the batteries would

last only

15 minutes! I believe that they typically use about 2-3 times the
non-multiplexed current in calculators.

The best way to find out for yourself how much current is really

needed is

to take a Nixie tube and run it at a fractional duty cycle, then

vary the

current with a variable resistor and look at the tube. You may be

surprised

at what you find.


--David Forbes, Tucson, AZ
http://www.cathodecorner.com/

Yes and no! I meant the impulse current is a few times higher!
for example: in IR Remotes the LED is driven with impulse current up
to 1.5 Amps, while Datasheet says Imax 100ma const. Pulses are below
1 percent, so effective current is only 15 milliamps, way below imax!

My idea was to keep the active current below the Imax (10ma for my
tubes), with 1/6 duty cycle (hh:mm:ss), leads to effective 1.6ma,
which is below the recommended maintaining current of 2-2.5ma. the
overall current is full-time 10ma, as it would be with 6 steady-on-
tubes at 1.6ma. Where is my fault of thought???

--- In NEONIXIE-L@y..., Raymond Weisling <fire.mountain@p...> wrote:

Odd. My experience shows that you don't need anywhere near 6 times

the

current to achieve a usable display; on the contrary, 6 times the

current

would cause the tubes to act funny.

There is a brain phenomenon called persistence of vision that

comes into

play here, which makes the bightness-vs-duty-cycle curve nonlinear.

I don't believe that LEDs need anywhere near N times the current

either.

For example, a typical 8-digit LED calculator certainly does NOT

require

100 milliamps per segment - if it did, then the batteries would

last only

15 minutes! I believe that they typically use about 2-3 times the
non-multiplexed current in calculators.

The best way to find out for yourself how much current is really

needed is

to take a Nixie tube and run it at a fractional duty cycle, then

vary the

current with a variable resistor and look at the tube. You may be

surprised

at what you find.

This is good -- it begs a side-by-side A-B test, both of nixies and
LEDs. One of the devices is run at a DC current and the other is
subject to a variable duty cycle and variable current. Adjust them
for identical apparent intensities, and do so at different current
levels for the DC drive one as the "standard". Who wants to do it?

Not me. (assuming we are looking for which method provides maximum
lifetime) I'm kindof 'busy' for the next 10 years.

Forgive my inaccurate quoting but was it 'if you double the current
you decrease the life by a factor of ten' ?

Makes Davids mention of 1.7 make sense now. Your tube is only on for
1/6th of its life whereas the direct drive is always on.(assuming you
never turn of the display)

Who, in the direct drive camp, run at maximum current anyway ? Isnt
it the nature to put a few more Kohms on the anode resistor to make
the tubes last longer at the expense of running a bit darker. What
standard are we talking about when we assume a multiplexed display is
dimmer, dont we mean 'cant ever be as bright as the brightest of
direct drive displays'. As David mentioned, you dont need six times
the current either, and isnt it interesting how a tube in a
multiplexed display has a *lot* more brightness than a mathematical
1/6th of the brightness of a direct drive display.

Hey, it is to me anyway !

At 8:12 AM -0700 4/2/02, ptsearcher2001 wrote:

--- In NEONIXIE-L@y..., David Forbes <dforbes@a...> wrote:

I believe that they typically use about 2-3 times the
non-multiplexed current in calculators.

--David Forbes, Tucson, AZ

Yes and no! I meant the impulse current is a few times higher!
for example: in IR Remotes the LED is driven with impulse current up
to 1.5 Amps, while Datasheet says Imax 100ma const. Pulses are below
1 percent, so effective current is only 15 milliamps, way below imax!

My idea was to keep the active current below the Imax (10ma for my
tubes), with 1/6 duty cycle (hh:mm:ss), leads to effective 1.6ma,
which is below the recommended maintaining current of 2-2.5ma. the
overall current is full-time 10ma, as it would be with 6 steady-on-
tubes at 1.6ma. Where is my fault of thought???

The problem is that the tube's life decreases by much more than a factor of
6 when you raise the current by a factor of 6... it decreases by a factor
of 6^3.3, which is 370. Still, it will be on only 1/6 the time, so the
lifetime decrease would be effectively 370/6 = 60, or an average lifetime
of about 1000 hours. That's how long a regular house light bulb lasts,
which is way too short to be practical in a clock.

The other consideration is that you don't need that much current to obtain
a decent display. Nixie tubes are very bright for indoor use. I find a
naked Nixie tube to be too bright to be easily readable.





--David Forbes, Tucson, AZ
http://www.cathodecorner.com/

##### level-1 quote by fixitsan2002 (on) 02.04.02 (at) 15:27 +0000

> ...it begs a side-by-side A-B test, both of nixies and
> LEDs. One of the devices is run at a DC current and the other is

subject to a variable duty cycle and variable current. Adjust them
for identical apparent intensities, and do so at different current
levels for the DC drive one as the "standard". Who wants to do it?

Not me. (assuming we are looking for which method provides maximum
lifetime) I'm kindof 'busy' for the next 10 years.

I am not talking about lifetime. I was referring to the peak current or on-state current required at a duty cycle of X percent on, to achieve the same apparent intensity. This is solely psycho-optical and can be completed for a given subject in a very short time. This was to prove or refute statements made about (for example) 1/6 duty requiring 6x current to achieve the same apparent intensity, either in a nixie or in a LED (they may not be the same, and in a LED it might even vary with the colour).

Ray

--- In NEONIXIE-L@y..., Raymond Weisling <fire.mountain@p...> wrote:

##### level-1 quote by fixitsan2002 (on) 02.04.02 (at) 15:27 +0000

> ...it begs a side-by-side A-B test, both of nixies and
> LEDs. One of the devices is run at a DC current and the other

is

subject to a variable duty cycle and variable current. Adjust

them

for identical apparent intensities, and do so at different

current

levels for the DC drive one as the "standard". Who wants to do

it?

Not me. (assuming we are looking for which method provides maximum
lifetime) I'm kindof 'busy' for the next 10 years.

I am not talking about lifetime. I was referring to the peak

current

or on-state current required at a duty cycle of X percent on, to
achieve the same apparent intensity. This is solely psycho-optical
and can be completed for a given subject in a very short time. This
was to prove or refute statements made about (for example) 1/6 duty
requiring 6x current to achieve the same apparent intensity, either
in a nixie or in a LED (they may not be the same, and in a LED it
might even vary with the colour).

Ray

Oh alright, I misunderstood, I was going by the message title and
should know it changes quickly here. True now, it would be
interesting.

Nevertheless, I'm still busy !

In my first clock (non-multiplexed, discreet CMOS) I used much larger
resistors than recommended, the tubes are B-5870ST's IIRC running at
around 300v with 100k resistors. I tried using 47k resistors and the
tubes got slightly warm but weren't much brighter.

--- In NEONIXIE-L@y..., "fixitsan2002" <fixitsan@a...> wrote:

--- In NEONIXIE-L@y..., Raymond Weisling <fire.mountain@p...> wrote:

Odd. My experience shows that you don't need anywhere near 6

times

the

current to achieve a usable display; on the contrary, 6 times

the

current

would cause the tubes to act funny.

There is a brain phenomenon called persistence of vision that

comes into

play here, which makes the bightness-vs-duty-cycle curve

nonlinear.

I don't believe that LEDs need anywhere near N times the current

either.

For example, a typical 8-digit LED calculator certainly does NOT

require

100 milliamps per segment - if it did, then the batteries would

last only

15 minutes! I believe that they typically use about 2-3 times the
non-multiplexed current in calculators.

The best way to find out for yourself how much current is really

needed is

to take a Nixie tube and run it at a fractional duty cycle, then

vary the

current with a variable resistor and look at the tube. You may

be

surprised

at what you find.

This is good -- it begs a side-by-side A-B test, both of nixies

and

LEDs. One of the devices is run at a DC current and the other is
subject to a variable duty cycle and variable current. Adjust

them

for identical apparent intensities, and do so at different

current

levels for the DC drive one as the "standard". Who wants to do it?

Not me. (assuming we are looking for which method provides maximum
lifetime) I'm kindof 'busy' for the next 10 years.

Forgive my inaccurate quoting but was it 'if you double the current
you decrease the life by a factor of ten' ?

Makes Davids mention of 1.7 make sense now. Your tube is only on

for

1/6th of its life whereas the direct drive is always on.(assuming

you

never turn of the display)

Who, in the direct drive camp, run at maximum current anyway ? Isnt
it the nature to put a few more Kohms on the anode resistor to make
the tubes last longer at the expense of running a bit darker. What
standard are we talking about when we assume a multiplexed display

is

dimmer, dont we mean 'cant ever be as bright as the brightest of
direct drive displays'. As David mentioned, you dont need six times
the current either, and isnt it interesting how a tube in a
multiplexed display has a *lot* more brightness than a mathematical
1/6th of the brightness of a direct drive display.

Hey, it is to me anyway !

Fine bit of work Nigel.

I'm more intereted in muxing frequency myself.

On many a data sheet, e.g. Z568, the max current for pulsed operation
is 25mA. But it doesnt mention pulse duration sadly. However of the
ones that do, 100-200uS seems to be average. I was at one time
running at 100uS but some bad electronics made me slow things down to
ten times that figure, 1mS. As far as brightness goes there wasnt a
great deal of difference. Remembering that as you increase the ontime
you decrease the refresh rate. But before anyone asks I have no idea
what the best frequency is, there must be an ideal one. At 150uS and
25mA theres a slight ringing from Z568 but you would expect that. It
was about equivalent to the brightness of 6mA of constant DC. So
thats a factor of 4. Interestingly now, if I run at 25mA with a 1mS
ontime, Its as bright as about 8mA DC, a factor of about 3.

What I'm saying is we havent got the objectiveness of this sorted out
enough have we ?

Does anyone have any old design sheets which might give a clue of
any 'ball park' muxing specs, if ever there were any. If we could
establish something steadfast and solid we'll have a definitive
answer to all of this.

Until then.....






--- In NEONIXIE-L@y..., "Nigel" <nigel@n...> wrote:

I decided to do some quick tests in response to Ray's suggestion. I

had the beginnings of my first Nixie clock with 6 ZM1040's wired up
as a multiplexed arrangement driven off a PIC with transistor anode
and cathode drivers. By disconnecting the anode on one of the Nixies
and feeding it from a resistor straight from the HV supply I was able
to do a direct comparison between that Nixie and the remaining Nixies
which were still multiplexed.

I adjusted the series resistor until the non-multiplexed Nixie gave

the same apparent brightness as the multiplexed Nixies and these are
the results.

HV supply = 220V

Multiplexed Nixies:
Anode resistor = 15K
V drop across resistor = 76V
Anode current = 5mA

Non-multiplexed Nixie:

Anode resistor = 133K
V drop across resistor = 95V
Anode current = 0.7mA

So the current required on the multiplexed Nixies to achieve the

same apparent brightness is about 7 times which allowing for
experimental error probably equates to the 6 times one might have
thought.

One thing I did notice was that, at the same brightness level, the

non-multiplexed Nixie appeared to give a 'thicker' or even slightly
fuzzy display compared to the multiplexed Nixie. I initially thought
it was a variation between devices so I swapped over the digits to
make sure it wasn't that.

I must stress that these tests were done quite quickly, I would

welcome hearing other people's observations.

I'm off to Las Vegas now for 2 weeks (working not playing!) so

probably won't be able to reply to any posts but I will endeavour to
keep looking at this group while I'm out there.

Nigel.


----- Original Message -----
From: Raymond Weisling
To: NEONIXIE-L@y...
Sent: Tuesday, April 02, 2002 3:24 PM
Subject: Re: [NEONIXIE-L] Re: Nixie lifetime and multiplexing


>Odd. My experience shows that you don't need anywhere near 6

times the

>current to achieve a usable display; on the contrary, 6 times

the current

>would cause the tubes to act funny.
>
>There is a brain phenomenon called persistence of vision that

comes into

>play here, which makes the bightness-vs-duty-cycle curve

nonlinear.

>
>I don't believe that LEDs need anywhere near N times the current

either.

>For example, a typical 8-digit LED calculator certainly does NOT

require

>100 milliamps per segment - if it did, then the batteries would

last only

>15 minutes! I believe that they typically use about 2-3 times the
>non-multiplexed current in calculators.
>
>The best way to find out for yourself how much current is really

needed is

>to take a Nixie tube and run it at a fractional duty cycle, then

vary the

>current with a variable resistor and look at the tube. You may

be surprised

>at what you find.

This is good -- it begs a side-by-side A-B test, both of nixies

and

LEDs. One of the devices is run at a DC current and the other is
subject to a variable duty cycle and variable current. Adjust

them

for identical apparent intensities, and do so at different

current

levels for the DC drive one as the "standard". Who wants to do it?

It should be done both in a dark room and in a bright room, as

the

eye will likely exhibit different time-decay persistence

phenomenon

depending on the amount of visual purple and whether the

receptors

are predominantly rods or cones.

Ray

Yahoo! Groups Sponsor
ADVERTISEMENT




To unsubscribe from this group, 2x click here:
<mailto:NEONIXIE-L-unsubscribe@y...>



Your use of Yahoo! Groups is subject to the Yahoo! Terms of

Service.

--- In NEONIXIE-L@y..., "fixitsan2002" <fixitsan@a...> wrote:

Fine bit of work Nigel.

I'm more intereted in muxing frequency myself.

On many a data sheet, e.g. Z568, the max current for pulsed

operation

is 25mA. But it doesnt mention pulse duration sadly.

I meant to say, there is a max current spec for pulsed operation, and
the Z568 is 25mA

##### level-1 quote by fixitsan2002 (on) 03.04.02 (at) 16:26 +0000

On many a data sheet, e.g. Z568, the max current for pulsed operation
is 25mA. But it doesnt mention pulse duration sadly. However of the
ones that do, 100-200uS seems to be average. I was at one time
running at 100uS but some bad electronics made me slow things down to
ten times that figure, 1mS. As far as brightness goes there wasnt a
great deal of difference. Remembering that as you increase the ontime
you decrease the refresh rate. But before anyone asks I have no idea
what the best frequency is, there must be an ideal one. At 150uS and
25mA theres a slight ringing from Z568 but you would expect that. It
was about equivalent to the brightness of 6mA of constant DC. So
thats a factor of 4. Interestingly now, if I run at 25mA with a 1mS
ontime, Its as bright as about 8mA DC, a factor of about 3.

Note that even though the duty cycle may be 1/6 (15% or so, allowing for some all-off time to kill ghosting), the pulse duration is not freely scalable. If data sheets list 100-200 us, then it may be safe at that range, but not with that current at 1 ms. Obviously one second in and five off would damage the tubes quite quickly, so the relationship is not linear.

I find Nigel's information about crispness (focus) interesting. I noticed this as well when I was working on the built-in segment test mode for the Four Letter Word (B-7971 tubes). The test turns on only one segment at a time, one tube at a time, so with 15 segments there are 60 phases. At first the software look was very fast and they looked steady, but quite dim (and very uniform all along the segment). Even at such low intensity, the line was crisp and uniform. That is 1/60 duty. (In the end the test is slowed way down so you can watch the sequence, but there are several speeds selectable.)

Ray

--- In NEONIXIE-L@y..., Raymond Weisling <fire.mountain@p...> wrote:

I find Nigel's information about crispness (focus) interesting. I
noticed this as well when I was working on the built-in segment

test

mode for the Four Letter Word (B-7971 tubes). The test turns on

only

one segment at a time, one tube at a time, so with 15 segments

there

are 60 phases. At first the software look was very fast and they
looked steady, but quite dim (and very uniform all along the
segment). Even at such low intensity, the line was crisp and

uniform.

That is 1/60 duty. (In the end the test is slowed way down so you

can

watch the sequence, but there are several speeds selectable.)

Ray

You're saying that you noticed it was not as crisp or that it was
crisper ?
I've found some frequency/current combinations which make the neon
look like it is almost being sprayed away from the digit, sometimes
at lower current but higher frequency than when the digit is crisp. I
keep trying to make the digits go in and out of focus with a bit of
success. Not useful to man or beast right now and I'm from the
minimalist camp as far as controls go, but I know a few folk who
would enjoy the idea of a clock with a focus control. All sober
people too !

##### level-1 quote by fixitsan2002 (on) 03.04.02 (at) 17:09 +0000

You're saying that you noticed it was not as crisp or that it was
crisper ?

Far more crisp at the pulsed mode than when steady. Both were with rather conservative currents to the anode (of course the same resistors -- it is just a self-test mode that is selected by some of the six user-preference DIP switches; with so many segments and components involved -- 60 MPSA42 transistors and base resistors, 8 shift registers -- such a test mode is useful to locate problems).

I find your de-focused, spray-effect intriguing. I wonder why it occurs. It is almost suggesting that certain frequencies of on-off ionisation, or thermal effects of same, have some resonating effect with the atoms or electrons. Hmmm. Another thing to investigate, eh?

Ray