Thanks! That's alot of positive information - it seems that numitrons are 
indeed sturdy devices and they do not need that much protection as I 
What is the frequency of the dimming PWM? I wonder how high it has to 
achieve steady temperature due to thermal inertia of filaments. 
PWM dimming would end up much cheaper than LED drivers, I was afraid that 
it could affect the filaments life length, but now that I know that the 
idea is well tested... I'll still think about a way to change the 
brightness without directly PWMing the filaments, but it is really tempting.

I'll definitely use direct driving, for same reasons as you mentioned - 
ability to use any old phone charger as power supply, to keep things 
cheaper and more portable.

By the way - I stumbled upon this graph taken from DA-2300 datasheet:
Taken from this site:
Note the logarithmic scale on tube life!

W dniu piątek, 23 lutego 2018 15:10:24 UTC+1 użytkownik Mark Moulding 
> I've built quite a few Numitron-based clocks, as well as other displays 
> using the technology.  Personally, I love them: they're easy to drive, last 
> a *long* time, and are, to my eyes anyway, very attractive.  They provide a 
> wonderful glow, and because they're broad-spectrum, they can be filtered to 
> any color.  However I usually leave them "naked", and the non-monochromatic 
> light they emit seems to give a richer, warmer look than LEDs, or even 
> Nixies, to some extent.  (Of course, they don't have the fully-formed 
> character set of the Nixies.)
> I've used a variety of schemes to power the devices.  At first, I used a 
> dedicated microprocessor (an AT89C2051 - about $1 each) for each tube, with 
> the segments direct-driven from the processor outputs.  I implemented a 
> serial communication scheme that allowed the displays to be daisy-chained 
> to any length, and controlled with a single 2400-baud serial signal.  These 
> were driven using the RCA-recommended strategy of keeping the filaments 
> just below a visible glow whenever they were "off" for maximum lifespan, I 
> did this using PWM, and implemented brightness controls also, since I 
> already had the code to do the PWM.  These worked very well, and the only 
> failure I've had, even in extremely rough service (a mutant vehicle roaming 
> the desert at Burning Man) was due to mechanical damage to the tube.  
> (Amazingly, it even kept working for a few minutes without any glass 
> envelope at all!)
> A scheme that I've never used, but which is easy and should work well, is 
> to use a single fixed resistor from each segment to the supply (or ground, 
> of course, depending upon your circuit topology) to provide the keep-warm 
> current for the filaments
> For most of my clocks, I've just ignored the keep-warm function, and have 
> seen no ill effects as a result.  I've been using 74HCT595 serial-in 
> parallel-out latching shift registers to direct-drive the tubes; their 
> outputs are rated to sink 35 mA (74LS595 are even higher, at 130mA), and I 
> PWM the enable input to provide dimming.  I've had clocks running for a 
> decade now with no tube or other part failures (except for wall-wart power 
> supplies - they're often notoriously poor in quality).  The highest stress 
> is probably on the colon lamps, which are two 3.5-volt grain-of-wheat bulbs 
> that I'm running in series to match the luminosity of the Numitrons; even 
> though they've been blinking on and off every second for years, I haven't 
> had any failures there either.  Note that since my clocks are often placed 
> in bedrooms, they're usually run at less than full brightness - PWM ratio 
> of around 50% or so; however, even the one I gave to my 94-year-old mother, 
> which runs at full power, has also had no failures.
> A note about using the Numitron clocks (or Nixies, for that matter) in 
> bedrooms: they're excellent for this application!  A reasonable amount of 
> research shows that ambient light containing a significant blue-spectrum 
> component can disturb sleep, reducing the amount of REM sleep and overall 
> sleep time.  This is true even when the eyes are closed!  The warm glow of 
> Numitrons, especially when dimmed, contains almost no blue light, and so is 
> the ideal bedroom clock display.  I'm lucky enough to have acquired most of 
> the remaining supply of DTF104B tubes (1" digit height, end-view), and 
> clocks made with these can be seen easily across a room, even if the viewer 
> normally wears glasses.
> I always use direct-drive in my designs, because multiplexing pretty much 
> obviates the big advantage of Numitrons, which is the ability to drive them 
> directly from the +5-volt logic supply - there isn't sufficient brightness 
> without a higher-voltage supply.  I did use multiplexing once, to retrofit 
> an existing 3-digit LED display on a musical keyboard I rebuilt into a 
> steampunk enclosure.  In this case, I simply replaced the LEDs outright 
> with the Numitrons, and removed the current-limiting resistors.  This 
> actually worked better than I thought it would; although the displays were 
> a bit on the dim side, they were still completely legible in normal room 
> lighting.
> Other comments about the lifespan of Numitrons are very accurate - at 
> reduced voltages, the tubes will last basically forever.  RCA data sheets 
> show a mean life expectancy at full voltage of 100,000 hours (longer than 
> LEDs are specified for!), and their lifespan curves are basically 
> asymptotic to infinity at 70% voltage.  Also of note is the specified 
> ruggedness of these devices: rated impact acceleration (while operating!) 
> of 200G, vibration 5-200 Hz 20G, etc.  The Russian tubes are not specified 
> as highly (15,000 hrs operational life - I have no idea about vibration), 
> but I use them all the time, and again have never had a failure.
> ~~
> Mark Moulding
> On Wednesday, February 21, 2018 at 5:26:39 AM UTC-8, Tomasz Kowalczyk 
> wrote:
>> Hi, 
>> I'm recently getting more and more interested in making an IV-16 clock. 
>> I'm planning to make 6 tubes one, so burning filament can be an issue - 
>> while the datasheets can state 5000+hrs, I'd like to be extra careful.
>> One idea is to use a LED driver with programmable current, for 
>> example STP16CP05XTTR - current is set by an external resistor, so it would 
>> be possible to introduce dimming the display without PWM. There are 
>> numerous chips which allow current sinking, however I am concerned with 
>> power dissipation on these - that's why my choice is a package with thermal 
>> pad. As far as I know, most critical moment in incandescent bulbs is turn 
>> on due to inrush current - with a CCS this problem would simply not exist.
>> However I've noticed a small mention of numitron driving in this CD4511 
>> datasheet:
>> <>
>> On last page there is a small schematic. It shows another method of 
>> dealing with inrush current - all filaments are prebiased with small 
>> current, which causes them to be already on edge of glowing - kind of how 
>> VFD filaments are biased.
>> What are your experiences with numitrons? What methods of prolonging 
>> their life did you use? Or did you just ignore the problem, as IV-9 and 
>> IV-16s are still quite cheap?
>> Please share your experiences and thoughts.

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 post to this group, send an email to
To view this discussion on the web, visit
For more options, visit

Reply via email to