Arduino-based FLW

[Matthew]

I'm one of the seemingly many folks that got stung by Zetalink, having paid for two FLW kits a few years ago and never receiving them. Life's too short to cry over spilt milk... but have finally decided to do something with the tubes I bought at the time - 4 B7971s and 4 ZM1350s - so have decided to embark on making my own Arduino-based FLW.

Although I'm fairly new to electronics, having really just built a few clocks based on the TubeHobby kit, I am a programmer by day so I'm reasonably confident with the software side of it. So I plan to start building the prototype based on some common anode LED starburst displays which I have - I figured they are at least logically the same as doing it with the tubes.

With that in mind:

1. Once the code is done and working on LEDs (it may be quite some time, given the trials of life...), is anyone interested in helping with circuits to drive the high voltage tubes? Am hoping to come up with a fairly easy to build design that can be open sourced along with the code. I have the schematics for Ray Ws design if it's any help.

2. I'm planning to code the LED version as closely as possible to the final tube version, so will be multiplexing them. Looking at Ray Ws design, it looks like the digits are multiplexed, with only one driven at a time... could/should/shouldn't I multiplex the individual segments too? Not sure if the tubes would be bright enough?

Feel free to infer as much cluelessness in me as seems appropriate :) But any tips, advice etc much appreciated...

[David Forbes]

On 5/15/2012 3:21 PM, Matthew wrote:
>
> 2. I'm planning to code the LED version as closely as possible to the final
> tube version, so will be multiplexing them. Looking at Ray Ws design, it
> looks like the digits are multiplexed, with only one driven at a time...
> could/should/shouldn't I multiplex the individual segments too? Not sure if
> the tubes would be bright enough?
>

Matthew,

I have an early FLW board, which was not multiplexed. I think it would be a
better choice to do than multiplexing, for both tube life and code simplicity.

I'd recommend that you use common DIP-packaged parts for the board. The 74HC595
8-bit serial shift register and SN75468 100V 7-way high voltage driver are
low-cost, readily available parts that would be easy to use and program.

I'd also recommend that you go with John Taylor's tiny power supply module
unless you care to get into power supply building, in which case Nick DeSmith's
design is an excellent choice.

Using the above parts will ensure that you spend most of your time working on
software rather than hardware.

--
David Forbes, Tucson, AZ

[Adam Jacobs]

Think about the requirements for your project before you start working on a design. :)

Is power consumption a concern? 4 B7971's will draw a lot of current if direct-driven.

Do you want the tubes as bright as they can be? direct-drive gives you lots of brightness.

Is tube-life a concern? This is trickier, but it seems to be fairly well agreed that multiplexing extends tube life (25% duty cycle means 4x more effective lifetime) UNLESS you overdrive them to make them brighter (closer to the brightness of direct driven) in which case, you shave life according to how much you overdrive the tubes.

I think that your plan to develop the software against the common anode 16-segment LED's is a great one. Once you have the code working, you can come back and start figuring out how to plug your arduino into some nixie driving IC's instead of LED driving IC's. If you really think ahead, those IC's that David mentions below will probably drive LED's as well as nixies (I've never used those chips.. check your datasheets).

As far as multiplexing each segment in the nixie... Why? I think that this is needless complexity without any clear value.

-Adam

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.

To post to this group, send an email to neoni...@googlegroups.com.
To unsubscribe from this group, send email to neonixie-l+unsubscribe@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/neonixie-l?hl=en-GB.

[Matthew]

Many thanks David and Adam for your hugely helpful replies - I'm at the needing encouragement stage on this and yours is much appreciated :)

David, thanks for pointing out the PSU - for the money, it's a no-brainer. And for the IC ideas - I'm actually ahead of you on one of them, I have 74HC595s already which I plan to use rather than using all those pins - I figured I could chain 2 to make  a 16 bit register.

I saw the SN75468 used in Ray's plans and didn't quite get it at first, but now I see that (at least in neophyte terms) it's a bunch of transistors on a neat IC. Haven't yet figured out if I need something like that for my initial LED solution, or if I can drive it directly from the shift registers.

Adam, good point about the multiplexing - I picked up on the idea of multiplexing the segments somewhere, but I guess it would affect the brightness and it seems simpler to just fire all segments of each digit at a time. As you suggest, I figured it will extend tube life - I don't mind the extra coding for it, but David, you are suggesting direct drive would be better? Is there some penalty from the constant switching?

BTW am also planning to add this 'Clock with Tics' mode to it - kind of a reverse FLW... http://leblanc.co.cc/?p=701

[David Forbes]

On 5/16/2012 3:10 PM, Matthew wrote:
> Adam, good point about the multiplexing - I picked up on the idea of
> multiplexing the segments somewhere, but I guess it would affect the
> brightness and it seems simpler to just fire all segments of each digit at
> a time. As you suggest, I figured it will extend tube life - I don't mind
> the extra coding for it, but David, you are suggesting direct drive would
> be better? Is there some penalty from the constant switching?
>

The advantage of direct drive is that you avoid having to deal with timing
issues. Nixie tubes are not very fast, so you have to pay attention to how many
microseconds elapse between when you issue the command to turn off the tube via
its anode switch, and when the gas stops ionizing, before you may change the
cathode signals to that for the next tube.

If you get this wrong, then you start to see ghosts of different characters in
the tubes. It can be rather frustrating to eliminate these ghosts, particularly
if you don't have an oscilloscope to see what's going on with the signals.

Simply put, your task is simpler if you don't multiplex the tubes.

[Jeff Thomas]

Besides the points already noted, audible ringing can be an annoying problem when multiplexing.

Regards, Jeff

[Matthew]

Many thanks again for the help on this. Somewhat improbably, I now have a working LED-based prototype on the breadboard - it includes the modes of the original and a few new ones too. I have multiplexed it for now - to save space and bits at least - I started out the hard way but then discovered the timer interrupt of the Arduino chip which makes it much easier. But my display is fairly well abstracted from the rest of the code so if it turns out I want to direct drive, it's not rocket science to change it.

So my circuit is fairly simple - In each multiplex cycle, the Arduino writes out to 3 74HC595 shift registers:

* 1 controls the anodes of the 14 segment displays (just one at a time in this case) - these go through 2N3904 transistors, since the total draw of one character is more than the register can handle

* The other two control the segments; for now I'm using resistors such that the 74HC595 can sink the current so they are connected directly to the cathodes.

Now thinking about moving up to the tubes. Very simplistically, I'm seeing the eg B7971 as a bigger version of the common-anode LEDs I'm using - just requiring a higher voltage and current, so needing to be driven through appropriate transistors. I'm sure it can't be *that* simple though - can anyone broadly summarise what other things I need to consider?

[Adam Jacobs]

Another stepping stone you might consider (although it is completely
backwards) would be a 16-segment VFD based FLW.
I know that I personally would be very interested in such a device,
given the significantly higher VFD availability than B7971. VFD's are
common cathode and are voltage regulated rather than current regulated
as nixies&LED's are.

As for how to drive Nixies versus common anode LED's.. They're pretty
similar. Just as you say: Higher voltage & higher current. In both cases
you are sinking current through a voltage limiting resistor. Your
current limiting resistor calculations will be different, your supply
voltage will be different, very likely your driving IC (or transistors)
will be different. They easy NPN transistor that is commonly used is the
MPSA42. Other than that, there are tons of different options for driving
the nixies ranging from the antiquated 74141 to the supertex parts to
all kinds of solutions.

-Adam

> --
> You received this message because you are subscribed to the Google
> Groups "neonixie-l" group.

> To view this discussion on the web, visit
> https://groups.google.com/d/msg/neonixie-l/-/iCfUhWC0mzgJ.

> To post to this group, send an email to neoni...@googlegroups.com.
> To unsubscribe from this group, send email to

> neonixie-l+...@googlegroups.com.

[Matthew]

Adam, thanks for the reply, much appreciated.

As to driving the nixies, I feel I'm missing something - David suggested the SN75468 transistor array, which I have seen mention of elsewhere in reference to Nixies - but its rated at 100v, less than that required by the tubes. Implying that the 'off' cathodes need to be held at eg 80-100v somehow, rather than 0v? If so, is that just an issue of this particular driver; or the 'right' way to drive them whichever transistors are used?

Interesting you mentioned the VFD tubes - I'm now getting ahead of myself and thinking of making a few of these (having seen the variable vulgarity and bonus 8-letter expletives, everyone wants one...), so am looking for other display alternatives and have just ordered some IV-4s, which I see a couple of other folks have also made FLWs from (although am as yet clueless as to how to generate the required voltage).

Any other ideas for readily-available display solutions? I'd love to make a split-flap version, but they certainly don't seem readily-available :|

[Adam Jacobs]

Hi Matthew,
You've actually already got the answer: Off is 80-100v. Unlike LED's,
Nixies change state from a very high impedance device (below the
ignition/striking voltage) until the ignition voltage is achieved, where
they become very low impedance. The nixie will stay very low impedance
while "lit", until the voltage is dropped to below the "extinguishing"
voltage. Once the extinguishing voltage is achieved, the nixie goes back
to being a very high impedance device. Therefore, we do not need to sink
the entire 180v to turn the nixie off.
With clever application, there are tons of IC's that can be press-ganged
into driving nixie tubes. In fact, the venerable 74141 sinks only
something like 80v (I think).

VFD's aren't too hard to drive. If you look back through the list
archive, you will find lots of discussion on how to drive VFD's. The
easiest way for discrete tubes like the IV-4 is to run the filament on
DC off of a 5v supply (a beefy one, I like the simple-switcher series
because they can attach a heatsink). Run the segment voltage from a
small boost mode switcher, any nixie supply can probably be modified to
lower the output voltage into the correct range.

-Adam W7QI

> --
> You received this message because you are subscribed to the Google
> Groups "neonixie-l" group.
> To view this discussion on the web, visit

> https://groups.google.com/d/msg/neonixie-l/-/pWXn3qySAW8J.

[Matthew]

Finally getting back to this after more boring bits of life, and finishing a couple of nixie clocks I was working on. Back to FLW = Back to asking questions - this time about resistor values.

* As before, I'm experimenting with cathode drivers - and came across this blog post which explains it (via some help from this group). I also tried it with the 50v ULN2003A and a 47v Zener diode and it works - this is good news since they are cheaper and more readily available. I'm not quite sure how to calculate the pullup resistor, but it doesn't seem too critical.

* However, given that I know from the spec sheets what current should be going through each cathode, I can't quite get my head around what voltage differential I should be using to calculate the cathode resistors...?

* Similarly, I just discovered that unlike LEDs, nixie circuits seem to use an anode resistor as well as cathode resistors - why is this? The spec sheets of the tubes show the anode current - but as above, I'm not sure what voltage difference I should be using to calculate this.

* Also been experimenting with anode drivers; I was going to use an optoisolator, but I'm trying to make a design which will drive both the nixies and LEDs (with appropriate changes in supply voltage and resistors), and it turns out I can't get one to handle the current of the LEDs. So I'm probably back to the common MPSA92/42 combination. I see lots of schematics including these, but I'm not quite getting the logic of how the resistors are calculated...

As ever, any insight hugely appreciated...

[David Forbes]

On 8/22/2012 12:47 PM, Matthew wrote:
> Finally getting back to this after more boring bits of life, and finishing
> a couple of nixie clocks I was working on. Back to FLW = Back to asking
> questions - this time about resistor values.
>
> * As before, I'm experimenting with cathode drivers - and came across this

> blog post<http://lucsmall.com/2011/07/19/using-the-sn75468-as-a-nixie-tube-driver/>which explains it (via some help from this group). I also tried it with the

> 50v ULN2003A and a 47v Zener diode and it works - this is good news since
> they are cheaper and more readily available. I'm not quite sure how to
> calculate the pullup resistor, but it doesn't seem too critical.
>
> * However, given that I know from the spec sheets what current should be
> going through each cathode, I can't quite get my head around what voltage
> differential I should be using to calculate the cathode resistors...?
>
> * Similarly, I just discovered that unlike LEDs, nixie circuits seem to use
> an anode resistor as well as cathode resistors - why is this? The spec
> sheets of the tubes show the anode current - but as above, I'm not sure
> what voltage difference I should be using to calculate this.
>
> * Also been experimenting with anode drivers; I was going to use an
> optoisolator, but I'm trying to make a design which will drive both the
> nixies and LEDs (with appropriate changes in supply voltage and resistors),
> and it turns out I can't get one to handle the current of the LEDs. So I'm
> probably back to the common MPSA92/42 combination. I see lots of schematics
> including these, but I'm not quite getting the logic of how the resistors
> are calculated...
>
> As ever, any insight hugely appreciated...
>

First, remember that the B7971 is not a stacked-digit Nixie tube, rather it's a
planar segmented display. So it's more like an LED character display in terms of
its circuit topology.

You don't need an anode resistor if you use cathode resistors.

The resistor value is Rcath = (Vsupply - Vtube)/Icath

Where Vtube is the voltage across the tube from anode to cathode when operating
at the desired current. This number is in the data sheet as an approximate value.

I recommend that you hook up a variable resistor to the tube's cathode and
measure the voltage across the tube and across the resistor at different
resistor settings. You'll get a feel for how the tube works.

[Michel]

It sounds like you like to try something new rather than using the old
MPSA transistors. Tell you what, why don't you use an opto triac? For
example the FOD4208, for sure it can support the required LED current.
The only thing that will be a (small) challenge is to switch it off as
it won't do that for as long as the current flows to the triac, which
means it will stay on until you pull the plug..... You will need to
disable all segment drivers so that the current will become 0mA, and
then the triac will turn off and you can then go to the next tube. You
need a blanking time anyways, so it's not really a disadvantage to do
it this way.

There's probably people who have tried this already, so you may ask
them if that works fine, but I don't see any reason why it shouldn't
work.

Michel

On Aug 23, 5:47 am, Matthew <why...@gmail.com> wrote:
> Finally getting back to this after more boring bits of life, and finishing
> a couple of nixie clocks I was working on. Back to FLW = Back to asking
> questions - this time about resistor values.
>
> * As before, I'm experimenting with cathode drivers - and came across this

> blog post<http://lucsmall.com/2011/07/19/using-the-sn75468-as-a-nixie-tube-driver/>which explains it (via some help from this group). I also tried it with the

[Matthew]

First, remember that the B7971 is not a stacked-digit Nixie tube, rather it's a
planar segmented display. So it's more like an LED character display in terms of
its circuit topology.

You don't need an anode resistor if you use cathode resistors.

The resistor value is Rcath = (Vsupply - Vtube)/Icath

David, many thanks - this is actually kind of what I figured - very analogous to the LED displays I'm working with, where I wouldn't use an anode resistor either, just cathode ones. I was just a bit thrown by seeing a number of B7971 circuits which use anode resistors, and the spec sheet for it specifying an anode current and a suggested resistor value - is this just overengineering?

Thanks also for confirming where the voltage difference is - I should have twigged, again just the same as with an LED version. I guess the voltage and the resistor are interlinked, but using a variable resistor to figure it out sounds like a good plan. The one thing I couldn't figure out was that if I'm sinking the current into a 50V ULN2003A clamped to 47v, would that affect the eg 170v of Vsupply?

It sounds like you like to try something new rather than using the old 
MPSA transistors. Tell you what, why don't you use an opto triac? For 
example the FOD4208, for sure it can support the required LED current. 

Michel - thanks also - I didn't even know what a triac was, but now I do :) Definitely an option. I don't have anything against the MPSA transistors, it's just that my understanding of the circuit is lacking - after much research online, I still don't quite get why we don't just use one to switch the tubes, as opposed to the linked PNP and NPN arrangement.

(I did, however, figure out that on the same PCB layout, I could use a a TLP627-4 for the lower-current nixie version, and regular individual transistors for the LED one - might go with that even though it's a bit of a hack...)

[fixitsan]

Any other ideas for readily-available display solutions?

 

My Smartsocket design makes life easy when yoiu want to integrate B7971 or IV-17 vfd's into a design.

 

The latest IV-17 version has a built in 4Play function (like four letter word but with added word ladder feature)

 

If you search Youtube for 'Smartsocket', 'B7971 Smartsocket' or 'IV-17 Smartsocket' you'll find plenty of videos of them working

 

I am sharing the code and the designs for free at the Yahoogroup http://groups.yahoo.com/group/smartsockets/

 

Chris Barron

[Matthew]

On Friday, August 24, 2012 12:45:37 PM UTC+1, fixitsan wrote:

Any other ideas for readily-available display solutions?

 

My Smartsocket design makes life easy when yoiu want to integrate B7971 or IV-17 vfd's into a design.

 

The latest IV-17 version has a built in 4Play function (like four letter word but with added word ladder feature)

Chris - thanks for this - these look great! I'd be ordering them now if I wasn't having such fun (relatively, it's a bit of a steep learning curve for me) how do do the logic for this stuff.

What I am keen to find, though, is another display solution - I have the B7971s for me, but I want to make a few more of these things and the large nixies obviously aren't sustainable. LEDs are the next best choice, but I'm keeping my eyes open for other ideas.

Anyway, thanks to all the help from everyone here, I now have a working FLW! Still various glitches to iron out, and features to add to the code, but the basic circuit is there and multiplexing them happily. I'll post a video if I get round to doing one... 

[fixitsan]

On Tuesday, August 28, 2012 10:31:50 PM UTC+1, Matthew wrote:

What I am keen to find, though, is another display solution - I have the B7971s for me, but I want to make a few more of these things and the large nixies obviously aren't sustainable. LEDs are the next best choice, but I'm keeping my eyes open for other ideas.

I also have an LED version of the Smartsocket using a 18F46k22. In between the pic and the LED displays are just 17 resistors and 4 transistors. I haven't updated it with the latest IV-17 software but I guess I that ought to.