EEVblog #948 - Nixie Tube Display Project - Part 1

[FxDev]

This is my Nixie Tube Clock.
First I want to use USB to get power and communication.

I used STM32F072 48pin series. It has internal RTC and yes, crystalless USB!
I wrote PID algoritm for 5V to 170V converter and bingo, i did it with classical boost converter and around %98 duyt cycle!
For more information please read my entry about that, unfortunately it's Turkish: http://www.firatdeveci.com/gecmise-donus-nixie-clock/

Dave you have to tell about cathode poisoning.

Then I wrote 00-99 counter demo! God, Nixies response times are so fast! Look at it!



Then after that I did my first clock with using IN12A and IN12B. Also RGB background led display and buzzer for alarm.
Now i am designing very small nixie tube clock it will be around 4.5x4.5cm and will work with only one vertical nixie.
Here is the video of that and some photos.

[Macbeth]

The timing of this is pretty good, I've just recently built a 4 nixie board that can be cascaded together.

It uses a K155ID1 for driving the digits and a transistor for the decimal point (since the K155ID1 only has 10 outputs). The transistor is the MMBTA42LT1.

This is great, pretty similar to what I have been procrastinating, including the transistor for the decimal point. You are using a K155ID1 per 4 nixies too. Where did you get the pins to mount the nixies to? The official sockets are kind of bulky and soldering nixies to the board is bad.

It's the first PCB I've ever had made and it probably shows I can upload the eagle files if anyone is interested.

It looks far better than what I can muster with my ferric chloride and bubble tank setup. Please do publish, but at least in PDF format 

[Dubbie]

Very pleased to see a board level design video again.

[NivagSwerdna]

There are plenty of designs using devices rated at 80V so I think that voltage drop video is very misleading.

If you apply 0V to the pin the segment will glow, if on the other hand you apply 60V to the pin then it will not and there will be negligible current.  That's because 170V-60V isn't enough to ionize the segment.

[EEVblog]

There are plenty of designs using devices rated at 80V so I think that voltage drop video is very misleading.

Why? it's what I measured.

[EEVblog]

In some cases I have seen zener diodes to clamp the voltage.
That way you can use a 80V shift register.

You can use jelly bean ULN2003's with a zener clamp on the COM pin.
Tossing up if I'll do that or go with a Microchip part.

[NivagSwerdna]

There are plenty of designs using devices rated at 80V so I think that voltage drop video is very misleading.

Why? it's what I measured.

http://www.avrfreaks.net/forum/tut-hard-soft-how-build-nixie-clock?page=all has an explanation of their approach which is done by someone far more capable than myself.  In this configuration they drive the HV5812 with 60V, perhaps using a resistor divider or resistor/zener on the HV supply.

A much simplier solution is probably the HV5622 which ticks all the boxes.

[Towger]

I wonder what the pink insulator beads between the digits are made from?   I have a number of the same nixies without the dp and they have white insulators, except for one which has a pink one mixed between them.

May not be wise to breathe in the dust if one got crushed...

BTW the numbers on the side 9010 is the date of manufacture.

[JoseLog]

Some time ago i found this image of a Panaplex display schematic.
The original site does not have the image available anymore: http://www.glowbug.nl/neon/HowToDriveNixies.html

What's interesting, i think, are the high voltage rails and the clamping diode. I'm not entirely sure about what's going on there, but to me it seems like the 80V rail is acting as a voltage limiter for both upper and lower transistors, so that they never exceed 110V and 80V respectively, neither turned on nor off. Something of that kind may work when using a ULN... type shift register.

Now, in the previos link also explains why there's no need for such a high voltage driver... it sounds reasonable, but has someone implemented it and tested it?

[boffin]

I recently was having a little think about how best to drive IN12Bs for a little project I have on the go, and was considering making a little board with the shift and drivers on it (old school 141s) until I came across this guy's boards: (and he deserves a plug because this is nice stuff) http://www.tayloredge.com/storefront/SmartNixie/

He's got nice tiny boards with on-board drivers and a PIC that you drive with i2c, all the high voltage drivers, and they're cheap cheap cheap.  His socketed IN12 board is $8.45

Even the schematics are available (he uses BSS131 as the HV driver transistor)

Also a nice small/cheap 170 V power supply for $14.  Wasn't worth even considering doing it myself.

[FxDev]

I've just finished my one nixie board.
It can also work with USB. It has RGB led, buzzer and IN14 socket.
I will share all files after i comleted my work!

Sizes in mm.

[robinb]

I've used Tayloredge PSUs and his version of smartsockets for B7971 to make this beastie :


Details here: https://hackaday.io/project/9155-nixiebot

It communicates with the outside world via twitter :  https://twitter.com/nixiebot   so if you have an account you can make it display something by sending the right commands.

Commands are listed on it's tumblr page: http://nixiebot.tumblr.com/ref but basically you just tweet anything with the hashtag  #NixiebotShowMe then a word and it'll display it and send you back a picture by replying to your tweet. If the word won't fit on the tubes it'll scroll and make a movie instead, some characters can't be displayed so to get a scrolling phrase use something like #NixiebotShowMe this:is:a:long:phrase:but:the:parser:thinks:it:is:a:word:because:no:spaces:and:will:scroll:it:for:you   Though there's a 100 frame limit on the movies due to maximum file upload limits. 

There's a twitter imposed rate limit on how fast a bot can tweet ( no more than 25 times every 15 mins ) so if you all jump on it at once you can expect a delay in reply but they do all get queued up. The queue size is mentioned on the biography bit of the twitter page ( turns out that updating that doesn't count towards the tweet rate limit so it's a useful back channel ).

Next phase is to make a proper case for it, I'm probably going for lasercut perspex pyramid stack with internal lighting, openscad wrangling is going on right now

"Big old pile o' python in need of sorting out but it works" code is on github here: https://github.com/Zedsquared/NixieBot 

Have fun!

Cheers,
        Robin.

[bills]

Dave I don't mean to nit pick your video I loved it.
BUT, L1 is on the side not up-rite.
Layout not that bad.
BTW Mike (threeneurons) is a friend and has many nixie tube kits on his web site. (think pile of poo)
FYI.
bill

[bills]

To all this was in reference to the power supply board assembly in the video.
 

[EEVblog]

Dave I don't mean to nit pick your video I loved it.
BUT, L1 is on the side not up-rite.
Layout not that bad.

Silkscreen shows both vertical (circle) and horizontal mount.

[EEVblog]

[NivagSwerdna]

Enjoyed Part 2.

It showed the Totem Pole connected to the HV... if you look at some of the designs they instead connect the VPP of the HV drivers to 60V (ish) providing the pre-bias voltage and removing the glow etc from the non-driven elements.  The 60V being provided by either a resistor divider from HV (since the current is small) or alternatively a resistor/zener combination.

When you compare the HV driver with 60V VPP from a zener you get a picture that looks very like the DaveCAD schematic.

The use of the jelly bean is a nice touch and actually 7 drivers is quite nice if you are driving tubes with decimals, for some I have there are left and right decimals needing 12 lines... need additional current limiting on the decimals but it might work, 7x2=14; only 2 wasted per tube when direct driving.

[mikeselectricstuff]

With the totem-pole driver, I wonder if simly leaving the HV pin unconnected would work.

[EEVblog]

With the totem-pole driver, I wonder if simly leaving the HV pin unconnected would work.

I was thinking that. Don't know for sure, would have to try it.
Even if it worked, unless you understand the process of why, not something you'd in a production unit. Fine for a one-off.

[NivagSwerdna]

With the totem-pole driver, I wonder if simly leaving the HV pin unconnected would work.

Isn't the intention of the bias to bring the non-lit elements to a potential where they will not glow (60v giving 170-60=110v, not enough for a glow), if left to flap in the breeze then they adopt a potential which reflects their nearest element?

Here's an interesting graph... http://www.decodesystems.com/re-how-figure-5.jpg

And the full treatment of how much glow is objectionable etc! http://www.decodesystems.com/re-how-nixies-work.html

[exile]

An alternative is to use standard high voltage BJT and Charlieplexing.
Then you would get by with about 10 I/O from UC, 10 resistor and 80st BJT.
Maybe not so practical but fun little odd.

[b_force]

In some cases I have seen zener diodes to clamp the voltage.
That way you can use a 80V shift register.

You can use jelly bean ULN2003's with a zener clamp on the COM pin.
Tossing up if I'll do that or go with a Microchip part.

Yes, that was what I wanted to refer to actually, lol
Although, I'm still not a huge fan of these ULN things.
They are just simple 'crude' driver units, that act as an extra buffer.

This means that you still need heaps of I/Os OR a shift register.
In the end you don't win much at all compared to just a few transistors.
(which you can get in dual or quad arrays for just a few pennies).
Maybe just a tiny bit PCB space, but also not that much (also because you can't drive 10 channels.

I wonder what will happen if you connect the high side of the totem pole just to ground?

[Twoflower]

In the Video Dave says it several times it does not matter which digit is on/you're measure. Is this correct? I think it does matter which digit is active and which you're measure. The reasons why I think it does:

• Each digit has a different surface (I think some Nixies use for the one two parallel wires to increase the visibility).
• The distance to the anode might have an influence as well (including the other cathodes).
• It might matter if a off segment you're going to measure is in the path of the electrons to the active one or behind.

But in general your results indicate that there pitfalls. Even if Dave might haven't found the worst case. Still the 120-odd volts might be close I would say. And a few volts more or less doesn't matter as you need to handle with them in a appropriate way.

Keep it going!

[BU508A]

Probably you'll find some inspiration on this website (it is in German and English):

http://www.nixiekits.eu/

The author of this website has the schematics of his Nixie clocks online and imho they are quite nice.

[robinb]

There's also a forum for nixie clock builders here, it's been a bit quiet of late but has recently undergone a change of admin so might perk up.

http://www.tubeclockdb.com/forum/main-forum.html

Some very knowledgeable people hang out in the neonixie google group too:

https://groups.google.com/forum/#!forum/neonixie-l

Cheers,
       Robin.