Nixie Clock

[DotComBoy]

Hey guys, this is my first real project.
Since i started watching Dave´s blog, it inspired me to try to build something my self.

I got some in-12a nixie tubes with 74141 controllers and i thought about using them to build a clock with RTC and some nice enclosure.
I´m not very good in Eagle but i´m trying

Below is how the schematic look so far. Any sugestions?

[David_AVD]

How do you set the time?

[DotComBoy]

I still haven´t decide about using resistor ladder DAC or just simple switches. i think simple switches it will be..

[DotComBoy]

There we go.. 

[David_AVD]

I gather the power supply for the logic is not shown?  Nixie tubes need heater voltage too don't they?

[GK]

I'm afraid that using IC's, let alone a micro controller for a project like this is cheating.

[GK]

Nixie tubes need heater voltage too don't they?

No, they are just neons.

[David_AVD]

Ah, just a HV source then (apart from the logic supply).

[DotComBoy]

I'm afraid that using IC's, let alone a micro controller for a project like this is cheating.

I choose not to multiplex the tubes because i think it may shorten it´s lifetime.

I gather the power supply for the logic is not shown?  Nixie tubes need heater voltage too don't they?

Power supply is not shown, but probably it will be 7805 + caps for the Atmega and a HV module that i have for the tubes.

[GK]

I choose not to multiplex the tubes because i think it may shorten it´s lifetime.

Nnnooooooooo.........I mean IC's altogether.......

I designed an alarm clock as an apprentice tech without IC's, full schematic prints out onto about 20 A4 pages, 800 odd transistors. Now some 12 years later I am finally getting around to building it:

http://www.users.on.net/~glenk/ddac/ddac.htm

[GK]

BTW, clocking from the mains frequency in your design will have better long term accuracy than the RTC.

[mariush]

For such a simple project, I think you could drop the external oscillator and capacitors and save about half a dollar.  The internal oscillator will have enough precision for a watch, it will only update the display several times a second at best so you don't need much precision.

Add decoupling capacitors between  Vcc and GND .. 0.1 - 0.47 will be fine. The DS1307 would like one as well.

[GK]

The RTC makes for a cheap and easilly implemented battery backup. Nothing more annoying than having to reset a clock after a brief power interruption.

Crystal oscillators do have to be very accurate for long term accuracy in clocks. An frequency error of 100ppm will work out to about 1 minute of error per week; approximately half an hour in six months. I wouldn't rely on the untrimmed uC oscillator for this application.

[mariush]

It seems to me he's reading the time from the RTC, which has its own independent oscillator. I was talking about the external 16 Mhz oscillator attached to the Atmega.

That Atmega as far as I can tell from datasheets at a summary glance  works at 1-4 Mhz using the internal RC oscillator. I think he doesn't need more than 1 Mhz to loop several times a second reading the time from the RTC and flipping some bits on the output pins.

If he reads the time from RTC ten times a second, in the worst case scenario where the atmega slowly drifts the updates on the nixie screen  within 50 ms from the desired refresh moment.

[DotComBoy]

It seems to me he's reading the time from the RTC, which has its own independent oscillator. I was talking about the external 16 Mhz oscillator attached to the Atmega.

That Atmega as far as I can tell from datasheets at a summary glance  works at 1-4 Mhz using the internal RC oscillator. I think he doesn't need more than 1 Mhz to loop several times a second reading the time from the RTC and flipping some bits on the output pins.

If he reads the time from RTC ten times a second, in the worst case scenario where the atmega slowly drifts the updates on the nixie screen  within 50 ms from the desired refresh moment.

For such a simple project, I think you could drop the external oscillator and capacitors and save about half a dollar.  The internal oscillator will have enough precision for a watch, it will only update the display several times a second at best so you don't need much precision.

Add decoupling capacitors between  Vcc and GND .. 0.1 - 0.47 will be fine. The DS1307 would like one as well.

Since i´ll be using Arduino plataform for development, seams easier to me just to use the crystal, and because this won´t go in mass production, $1 or $2 won´t make a lot of diference, but is something to think about for the future, thanks.

Will add the decoup caps, thanks

I choose not to multiplex the tubes because i think it may shorten it´s lifetime.

Nnnooooooooo.........I mean IC's altogether.......

I designed an alarm clock as an apprentice tech without IC's, full schematic prints out onto about 20 A4 pages, 800 odd transistors. Now some 12 years later I am finally getting around to building it:

http://www.users.on.net/~glenk/ddac/ddac.htm

it was my first idea, but because my idea is to learn more about uC, i choose the atmega..
Besides that, with the uC the project will be ready earlier ( i hope )

[DotComBoy]

Hey guys, long time without messing with electronics.
Since a few components arrived i built in the breadboard this HV power supply for the display: http://slacklist.olek.waw.pl/smietnik/nixie2/zasilanie/nixie_supply.pdf and worked like a charm.

To improve my learning, i draw the schematic in eagle(bellow), routed the board and assembled.
the problem is that after turning on, the display turn on for aprox. 30seconds and then the inductor blows out.
Is there any mistake i did that i´m not seeing?

[cwalex]

are you using a power inductor or one of those ones that looks like a through hole resister but fatter? I think the latter will blow up in that circuit.

#disclaimer: I'm a beginner and take my advice with that in mind

[DotComBoy]

i´ve tried the one that looks like resistor and another in a plastic case, both worked in the breadboard but blew in the pcb after a few seconds.

[cwalex]

i´ve tried the one that looks like resistor and another in a plastic case, both worked in the breadboard but blew in the pcb after a few seconds.

That's weird! have you triple checked your PCB layout is correct?

[A Hellene]

This is not an entry-level project! Apparently, you will have to do some serious homework in order to become able to create designs that work as expected. Fortunately, every little detail of that black art is out there, awaiting to be accessed, learned, combined and internalised. Do not be afraid to look up the terms that you think you understand; there is always something more to learn about anything, and the little details are always the ones that can really make a difference.

Anyway, this is not only an excellent tutorial on the Nixie Clock project you are building; it is an extremely detailed teaching course, written by a brilliant author, on the use of high-voltage DC-DC converters, inductors and drivers, RTCs and of all the other project subsystems as well as the use of the microcontroller itself, with schematics, source code and references most of the textbooks would not even dream of touching!
And, yes, a single "anaemic" m48 microcontroller can easily run everything!

As a side note, do never think of the mains power supply as an accurate frequency reference. It is not!


-George

[c4757p]

Can I offer some friendly criticism?

• BIG pet peeve: don't use text labels as links unless you really must. It's much easier to read your schematic if you draw every wire. Some of them are pretty bad - pin 3 on the 555 shouldn't be labelled "out" for example, it suggests that it's the circuit output. Something like "drive" would be better. (It is a good idea on more complicated schematics to keep the labels and the wires - use them to mark wires, not make connections.)
• Straighten up the funny corners like above the MOSFET.
• General layout orientation is left-right=input-output, top-bottom=positive-negative. Basically, a horizontal flip of the whole schematic (the 555 is on the "input"), and move the MOSFET so ground/source are at the bottom.
• More general layout: If it's desirable that things be physically close together (like C2 and L1), put them together on the schematic. Keep short wires short (no need to run all the way around the 555 to connect 6 to 2), and it makes more sense visually to connect R5 (for example) to the VCC port on the 555 rather than the one on the inductor. Electrically the same, logically not.
• I've built this circuit before. One important fundamental problem: if the trimmer's wiper comes off the track, the control loop is disconnected and it flies up towards the maximum output voltage. Not good, it's capable of quite a bit depending on the component values. I have used a similar circuit to supply 20W at 350V! You should include a second voltage control system (the two resistors, potentiometer, and BC547), but with a fixed voltage (no trimmer) a bit higher than the highest you will use this at. (Work out the resistor values with Ohm's law, keeping the total resistance above around 200k, so that you get around 0.65-0.7V on the transistor's base at the desired output voltage). You want enough margin that you don't need a trimmer. Test it - not just by disconnecting the trimmer and actually causing a fault, since depending on the voltage it could blow up the capacitor or something else awful, but by changing the values on the trimmer circuit so you can adjust the voltage too high, start out low, and gently bring it up, making sure it hits a ceiling at the right spot.
• I'm a bit safety obsessed when I design things like this - this might be a bit unnecessary - but instead of that, I might include a latching overvoltage shutdown above a certain voltage. A similar circuit to the feedback one, switching a latch (you can build one with two transistors, Google "two transistor latch"), pulling down the reset pin through a resistor. I'd even consider having the latch blow a fuse so it stays permanently shut down until inspected.
• Are you really using IRF512, or is that just for convenience out of the components library? It's only rated 100V! I haven't used Eagle but I bet you can change the component value so it's clear.
• Hint: T1 (BC547) can be pretty much any general-purpose NPN signal transistor, so don't go out and buy a bunch of BC547s if you don't already have some.
• I understand the "DIY" appeal of using just a 555, but you'll get much better characteristics with a part for the purpose - I've used UC3843 with good results, for example, and there are many others. Alternatively you could drive the thing with the microcontroller's PWM output, though you have to be a bit more careful about duty cycle since you don't have such a nice, easy analog feedback loop. Though this circuit does work for the purpose, if you like it.
• The feedback loop will have a good amount of dependency on the input voltage, though it might not be obvious at first. I've seen 50V variation on the output over 9-16V in. I'd power at least R6 and the 555 through a regulator to avoid that (or redesign the feedback loop not to depend on it so much). Do not power the inductor through the regulator, it's a big waste.
• Connectors: You probably shouldn't have your HV output sharing a connector with anything, much less the VCC that probably goes off to other circuit boards that might have human interface stuff on them! Pay special attention in general to keeping the HV away from humans, even if components fail. Safety standards are generally better these days than in the Nixie tube era, and this is one place where the appeal of doing things the old-fashioned way might not be worth it.

Sorry, I know it's a lot and some is a bit nit-picky, but it's always good to improve 

Attached is how I would draw it, though it could possibly be tidied up a bit more.

[kizzap]

i´ve tried the one that looks like resistor and another in a plastic case, both worked in the breadboard but blew in the pcb after a few seconds.

This may sound like a silly question, but were you loading the circuit, or just measuring the voltage output?

[jnd]

I choose not to multiplex the tubes because i think it may shorten it´s lifetime.

Nnnooooooooo.........I mean IC's altogether.......

I designed an alarm clock as an apprentice tech without IC's, full schematic prints out onto about 20 A4 pages, 800 odd transistors. Now some 12 years later I am finally getting around to building it:

http://www.users.on.net/~glenk/ddac/ddac.htm

Oh come on, you're using brand new components? That's cheating. You should use NOS, at least 20 years old components, preferably metal and ceramic packages, no cheap TO-92s! You can use ICs that way too but only from that era, like 8048 mcu.

[svofski]

I choose not to multiplex the tubes because i think it may shorten it´s lifetime.

I do not know, why do you think so? This is the first time I hear of this. If anything, it might prolong the life because the current through the nixies will be lower.

Generic thoughts — you seem to like to have many parts in your circuit. It is really easier to take a higher pin count uC than to use 2 external shift registers. It is also easier to multiplex the top side and use only one decoder at the low side: that way you can stay with the same pin count and without 2 extra registers.

Not sure if this was already said before, but you need to carefully select the MOSFET for the boost converter. There are many very good MOSFETs that would not open from +5V. I had good success with IRF630N, from 5V TTL level they open just enough and they're fairly robust. Using a driver is still not a bad idea imo.

How are you going to regulate the duty cycle of your 555-driven converter? You have a micro that can provide rock stable signal with adjustable frequency and duty cycle entirely in hardware, without any CPU interaction. It also has ADC which can be used for feedback. Why not use what you have?

My nixieclocks: Patashnik (non multiplexed, Mega8535), Satashnik (multiplexed, Mega8). The latter is probably as minimalistic as it gets.

[c4757p]

I choose not to multiplex the tubes because i think it may shorten it´s lifetime.

I do not know, why do you think so? This is the first time I hear of this. If anything, it might prolong the life because the current through the nixies will be lower.

I've heard this a few times, though I haven't actually seen any evidence. People seem to think that the breakdown process that strikes the tube in the first place is damaging to it, so it's better to only strike it when changing digits than to flicker it many times a second. My hunch is that this is bullshit, considering neon lamps happily operate on AC all the time as pilot lights. I don't have that much experience with neon lamps, but in my limited experience, I've yet to see one burnt out.

How are you going to regulate the duty cycle of your 555-driven converter?

The feedback circuit connected to the CV pin does this. It's not true PWM - the frequency changes as well - but I've used a very similar circuit a few times and it does a decent job of regulation.