The other thread on converting a 6-segment bar graph to a 7-segment LED got me thinking about an old project idea. I'd like to convert a regular alarm clock to Nixie tubes.

The major problem for me is converting the normal 7-segment LED output to what would be the equivalent of a 10-dot bar graph. Any thoughts on this?

I know a few people have already tried this sort of thing but I recall thinking that they were approaching it all wrong at the time, so I didn't pay attention. ops:

 

There is no IC for converting 7-segment data to Hex, Binary or any other format. But you can implement it using PAL/GAL or any other suitable PLD. There is a sample PAL program for converting 7-seg. data to hex in "The Art of Electronics" by Paul Horowitz. Check it out if you have it.
Then decoding hex to decimal (10-outputs) is preety simple.

Or you can do it using logic gates also. It is quite complicated but no doubt solving k-maps is fun

__________________
"There is no way to peace, peace is the way!"

 

ugh 10, 7-variable k-maps.. :shock: no thank you. :lol:

i think kinjalgp has the right idea

 

I'm thinking that if I can't get a clock that has a serial output (fat chance) then I'm either going to use a PIC for the job, or probably make make the whole clock from scratch using a PIC.

 

First, please re-check the 6-led to 7-segment thread for some background on using partial decoding with sparsely populated truth tables. The same idea applies here, except now the 7 segment lines are the table inputs.

As kinjalgp noted, you could either decode the lines into a BCD format if you're planning to use a nixie tube driver chip like the 74141, or decode individual lines (0-9) and use peripheral drivers. These usually have 2, 4 or 8 drivers per chip. The high voltage ones are hard to find, as is the 74141.

http://www.camuw.demon.co.uk/web_sit...out/74141.html

The following article had the nice idea of using inexpensive high voltage NPN transistors (2N4409) in place of hard to find IC driver chips.

http://www.old555.com/LPE/unit/1CA46C1798B5F40A.shtml

The only catch I can see in converting a LED clock to nixies is that any commercial clock almost certainly has a multiplexed display. The nixie elements have enough persistence (just like lamp filaments), but you'd have to work around the low duty cycle of the strobes. I'm thinking each 7-seg strobe in a 4 digit clock has around a 25% duty cycle. From what I know the nixie supply is usually around 175-185Vdc, with a 15K-45K current limiting resistor in series with the anode. I don't know enough about nixie tubes to suggest raising the anode voltage, though I'd play with that idea first. Maybe each nixie needs latched inputs instead?

I've posted schematics below for both BCD and individual decodes. The only way to get the BCD version down to 3 standard logic chips is to use XNOR gates (CD4077 or 74ls266), which aren't very common. The BCD version could use a 74xx08, 74xx86, and 74ls266 (or if you'd rather, two XOR packages, one AND, and a spare inverter).

The individually decoded version needs 4 chips. (ex; two 74xx08s, a 74xx02 and a 74xx86) There's one AND gate left over to power the rest of your clock...

There are other ways to partially decode the inputs and get the same results, but you'll end up with approximately the same gate count.

I've seen some free-ware minimizers which can do Quine-McCluskey logic reductions. These will give the minimum gate logic needed, but unless you're implementing with programmable logic, it may not result in the minimum package count. This example and the previous 6-led thread show that XOR gates can be quite useful for decoding sparsely populated truth tables, as long as the "don't care" states are taken advantage of.

- CAL


I threw out a large box of tubes about 15 years ago. Now I wish I'd kept the nixies, they would make a nice "retro" looking clock.

Attached Images

	7seg_to_BCD.gif (11.3 KB, 983 views)
	7seg_to_10led.gif (14.5 KB, 985 views)