Nixie Tube Clock

[TheEPROM9]

Since David got sent some Nixie tubes, I think he should make a clock, counter, multimeter display mod or something display related. He could focus on high voltage design and how to get it right while making something for a bit of fun. 

Everyone likes Nixie's 

[electronics man]

He could do it wile talking about high voltage safety.

[TheEPROM9]

He could do it wile talking about high voltage safety.

Indeed, could be part of the safety video, shock his self for science >=-)

[Napalm2002]

My vote is in

[VK5RC]

Designing and building a nixie clock from scratch is a huge undertaking,  voltage multiplier,  nixie drivers and microprocessor program including anti burn in,  daylight savings,  blanking (extend life)  and gps linking are all needed.  The kit from Peter Virada has this coveted including pretty led base lighting,  I built 2 (work and home)
www. pvelectronics.co.uk
(i have no vested interest)

[GK]

microprocessor program including anti burn in,  daylight savings,  blanking (extend life)  and gps linking are all needed. 

They are?

Everyone likes Nixie's 

That's probably why there are about 2 million nixie tube clock projects on the interwebs already. 

[linux-works]

count me in for a nixie clock (or generic display) thread or project.  I'm now interested in them and planning on doing a board of some kind for my own design.

and yes, its a lot of work if you want to do it on your own.  there are so many features that can be added and so it turns into more of a software project than a hardware one.  I could easily spend months doing the firmware (part time) and adding this or that feature to it.

my first nixie project was to add a dB volume level display to an amplifier I built:







I took that to an audio show/demo and I was the only one there with a nixie digital display; everyone else is still stuck in the modern age with lcd's

quite a few people stopped by to comment on the nixies, so I know there is a newfound interest in these old things.

[kc0ngu]

Designing and building a nixie clock from scratch is a huge undertaking,  voltage multiplier,  nixie drivers and microprocessor program including anti burn in,  daylight savings,  blanking (extend life)  and gps linking are all needed.  The kit from Peter Virada has this coveted including pretty led base lighting,  I built 2 (work and home)
www. pvelectronics.co.uk
(i have no vested interest)

Wow, how did I design/build a nixie clock without considering any of these suggestions??

[linux-works]

I'm staring on a new one, a proper clock, this time.

early steps (not even a single tube is fully wired yet, but all the key parts are placed):



using an arduino nano this time around.

[nathanpc]

Designing and building a nixie clock from scratch is a huge undertaking,  voltage multiplier,  nixie drivers and microprocessor program including anti burn in,  daylight savings,  blanking (extend life)  and gps linking are all needed.  The kit from Peter Virada has this coveted including pretty led base lighting,  I built 2 (work and home)
www. pvelectronics.co.uk
(i have no vested interest)

I have to disagree with the statement that it's a "huge undertaking". Designing a HV power supply for nixies isn't difficult, you can even use flash circuits if you're not in the mood of designing your own. It's extremely simple to write a clock program for a microcontroller, the daylight savings feature could be left out or implemented later, so the user will have to manually do it just like 100% of the digital clocks (not connected devices) we have today. Why would you require GPS linking?

my first nixie project was to add a dB volume level display to an amplifier I built:

Not a single picture of the rest of that awesome circuit?

[linux-works]

the reason I plan to add both a gps input and an ntp (via a small linux board system) input is to entirely avoid having to SET the clock, at all   sure, you get nice accuracy but its always a PITA to set clocks, so an ascii input to a serial port that comes from gps or a pc would really make things so much easier to use.

I'm hoping to make more progress on mine this weekend and if I do, I'll certainly post some more pics and the schematic I used, along with the arduino code that I end up writing for it.

[linux-works]

testing the dc/dc converter as well as dry-fitting the tubes (which are round pin-out) on a square grid:



since dc/dc was mentioned, here's my zero-effort Just Buy It(tm) solution:

http://www.ebay.com/itm/140964710736

or

http://www.ebay.com/itm/151314847910

zero effort. give it 5v and set the output with a single resistor.  done.  I saw no need to design one when you can just buy it and its already a proven design and implementation.

in my previous photo, the black module is mounted upside down at the top/left corner.  what I didn't like is that its not .1" hole friendly (grrr!) and so I had to hack it with ugly stupid wires, sigh.  I asked the vendor why he didn't go with .1 spacing and he had no valid answer, unfortunately.  some makers just don't get it..

[ctz]

the reason I plan to add both a gps input and an ntp (via a small linux board system) input is to entirely avoid having to SET the clock, at all

I would have thought that a WWVB, DCF77 or MSF receiver is cheaper, more reliable and actually quite easy to build from scratch. But you can buy modules too.

[linux-works]

we don't have some of those freq's in the US.  and the longwave and shortwave things are too finicky to be reliable for instant time-setting.

otoh, its rare to NOT have internet and ethernet access these days.  this clock will go on my desk, at work, and I'll have wired ethernet in abundance.  a cheap rasp pi box OR even my desktop workstation with a serial port (usb/serial or usb/ttl serial) is all I need to feed it ascii time.  even if the local workstation is not perfectly accurate, it will be close enough for desktop clock use and it will allow instant setting of the clock.  also, if you let the pc source time, it will take care of DST changes for you.

for when I don't have NTP time access, I should have gps access.

but I'm allowing anything to be a time source.  as long as there is a module that can receive time, somehow, and send some serial report to the clock, that's fine.  the main point is to avoid having to have debounced switches in the back and force the user to have to enter time manually.   I just hate that since I'm lazy as can be

[Rigby]

WWVB is in the US and it works well, and it is accurate.  It isn't "instant", true, but GPS isn't either.

Use a Spark (spark.io); they can access the internet, are programmed like an Arduino, except in a browser, and are almost as small as the Arduino Nano.

[linux-works]

with the adafruit gps module I'm using, it often locks in under 30 seconds.

the linux board will take that long to boot, but then it gets NTP data pretty much instantly after that.

I don't consider wwv and other rf methods to be all that reliable, in my experience.  NTP over wired or wireless really is the more reliable method since we have IP pretty much everywhere, these days   how you get NTP time, that matters less to me.  the small linux boards seem natural for that and I know their IP stack is 'real' enough, plus flexible enough if I had to go thru firewalls or use proxies or whatever.  and if NTP is being blocked, I could http out and get time, that way.

if I'm not near some kind of IP-reachable network, its not likely I'll care about seeing time on my nixie clock.  if the network is down, I'm going home

[VK5RC]

@linux works, How sensitive is that adafruit gps module, work inside a building reliably?
Rob

[linux-works]

I took it to work and it didn't work from inside a 2nd floor building (san jose).  but, my android phone didn't work either until I left the building.

at home on my 2nd floor (cupertino) it works almost instantly.  and the battery backup keeps enough setting so the gps is not cold-starting (that's why its faster on successive restarts).

[linux-works]

this photo has the adafruit module shown:



its just tack-wired to an arduino module (the lcd backpack) for testing.  the adafruit gps does have an antenna socket so you could use the module as the antennna+gps OR connect a real antenna to this and have the local patch antenna auto-switch off.  I use the module as the antenna and have not found the external one to be all that much better.  at home, I get at least 5 or 6 sats locked, often going over 11 on good days.

this photo, though, does have the ext antenna connected, but these days I just use the module, solo.

[VK5RC]

I have used the GlobalSat EM408 (probably now outdated- bought a couple some years ago) for a couple of my GPS based clocks and have found them good but expensive (were 70-80AUD), one obtained lock inside a steel filing cabinet inside a single storey building. I was impressed.

[Rigby]

The best thing about the adafruit module is the 10Hz update rate.

[linux-works]

I just found a good use-case for using a linux system to get network time.

suppose you are inside a firewall (corporate) and they block NTP and other nice things.  what if you could get the time over port 80 (www), which is pretty much *always* allowed from inside corp networks?

here's what I came up with, as a quickie hack

% curl -s 'http://viv.ebay.com/ws/eBayISAPI.dll?EbayTime' | grep -A 1 'The official eBay Time is now' | tail -1 | awk '{ print $5 }'

try it, if you have a unix system.  you can see what it does: it just fetches ebay's 'time page' and looks for the magic line and gets the hh:mm:ss from that.  it works.

its not fast, but at least it gets you close and it DOES work inside my company's network.

can't do this with an arduino or pure hardware!  again, having a software solution wins over pure hardware or 'standard' protocols that may not work everywhere.

[Rigby]

Your company should have several time servers running.  Even windows uses NTP when it needs the current time.  Active Directory needs all domain joined computers to be on the same time, so there is a time server somewhere.

[linux-works]

there is definitely a timeserver on our LANs.  agreed.

but I don't know them (yet).  I could probably get time from a default router (etc) but since port 80 'always works', this seemed like a trouble free way to get time, at least within a few seconds.

I like having several fallbacks.  using ntp or ntpdate would be a first try; if that times out, then try other methods, which could include hitting a webserver that is known to keep time.  ebay is a great example since they NEED to be synced and they are big enough to be trusted to keep proper time

there is also something called 'nist-80' which does not look like a standard (yet) but is supported by some vendor code (non-free, I think).  that gave me the idea to try to get time over port-80.

I don't mind 'wasting' a rasperry pi for this.  its cheap, its fanless and it can do any of these tricks.

the rasp-pi box will have a usb connection to my arduino nano and maybe every half hour or so, it will push updated time across the serial connection and the arduino will just pick out the ascii HH:MM:SS and set the clock from that, directly.

I'll post a follow-up once its implemented, in case anyone cares.

[Rigby]

The Windows Registry key HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\services\W32Time\Parameters will contain the time server you're syncing to.

You can also turn on a time server on your Windows machine in a nearby key.  This may interest you if your HTTP method doesn't get you close enough.

You can add accuracy to the HTTP method if you can work out the latency in the network and the serving, downloading & parsing of the page ebay is sending you.  If you know the time you've pulled out is 0.75 seconds old, you can just add 0.75s to the time you've parsed out of the page to get the proper time.  It will depend on how close you care about getting.