Neon lamps, flashy LEDs, and soldering along the Nixie trail

Over the last couple days, I really sat down and sorted a couple of things that were unknowns. Either I had forgotten why I designed it that way on the schematic or had only roughly figured it out in the first place. One was the flashing LEDs from the last post. I had subsequently tried other resistor values for different effects, but in the end stuck with the 1.5kΩ one. I did like the blanking and hopefully it will be a nice effect once i have it all assembled. I can always change the resistor later if I wish.


Neon lights

Aside from the title of a very classic Kraftwerk tune, I needed a couple of small neon lamps for my clock, to form the flashing colon. I had waffled about the idea of not including them but I really couldn’t see the design without them.

Previously, I had a couple of them, but can’t find them now. I think I may have fried them. I have a vage memory of popping a couple in experiments I was doing a couple of years ago. I bought a couple more during an excursion down to College St. yesterday and refined the circuit I had previously on the schematic with something a little more workable.

For some unknown reason, I had 10kΩ resistors as current limiters, which seemed far too low. Indeed they are the same technology as the Nixie tubes so I would think that would pull too much current.

I decided to run some tests, of course, by whacking it up on the breadboard and taking some current measurements. As I suspected, the current draw using the 10kΩ resistor was ~8mA, a little high. It was nice and bright, but I was worried about burning them out too quickly and honestly I didn’t want that end of the circuit drawing so much power. I remember the switchmode high voltage supply was rather limited and I didn’t want to overload it. Quizzically, the written material that went with the source of that part of the schematic didn’t indicate a maximum current but just that there wouldn’t be much voltage swing with a 0-10mA load. So I had taken that as my maximum.

Well that’s a problem… fortunately easily solved

I hooked it up using 33kΩ for the neons and the nixies, wiring them all up effectively how they will be in the finished product (minus the switching transistors of course). The first thing I noticed was rather heartbreaking. The Neons were nice and bright but the nixies were rather pale and the numerals not all lighting up. I took a measurement and found that it was drawing ~9.5mA and the voltage had plummeted to ~143V. Interesting. The switchmode seems current limiting by design (as one would figure) and was dropping the voltage to keep the draw under 10mA. Fortunately, as the one thing I really didn’t want to do was melt my supply or worse – have to find a new supply design with a higher current output.

Fortunately, I knew an easy solution – simply lower the current limit on the neons, they were too bright relative to the Nixies anyway. I tried 330kΩ, 1MΩ, and finally settled on 100kΩ as the best balance. I found the circuit could regulate just fine and everything looked more or less a consistent brightness. Perfect!. I took some measurements and found that the Nixies and neons were pulling 6.62mA and the voltage was stable at 170V after a small trimpot adjustment. I couldn’t ask for better really.

Love that glow. For anyone building something like this, mind the loose wires, I made sparkies earlier.

Love that glow. For anyone building something like this, mind the loose wires, I made sparkies earlier.

The numbers

Can’t have hobby electronics without some numbers now, can we?

I had previously measured the four Nixies combined draw at 4.67mA. Had I left the neon droppers to 10kΩ as per my original schematic, they would have drawn a combined 8mA, making a total of almost 13mA, over the maximum that the supply will be able to output the needed 170V. As it was, I tested it with 33kΩ resistors and the neons sucked the life out of the supply, dimming the Nixies. I could have adjusted the pot to get the voltage back up to 170V (if that was even possible under such a load) but that ran a very dangerous risk of overloading the power supply by causing it to draw more than 10mA.

That seems a really tiny limit – 10mA – and it is, but keep in mind the supply voltage is 170V which magnifies how much power that’s dissipating. It’s always a tradeoff between voltage and current. That maximum would make for a power dissipation of 1.7W.

Even if it worked for a bit on my bench, who knows how long it would take for the stress on that circuit to start burning out components? It’s going to be a continuously running clock after all, some extra margin for component longevity is more than prudent.

The general rule of thumb is to always provide 20% margin of error. Since apparently 10mA is the ceiling for the high voltage power supply, any draw under 8mA is just fine. Since I’m ringing in at 6.62mA, I’m more than happy with that. That gives us a total power dissipation for that part of the circuit, supply, Nixies and neons – of 1.125W. So what apparently is a tiny change in current made a huge difference in power.

Now to make them flash

Incidentally, I had hooked the neons themselves up to the same MPSA32 transistor switcher I use for the Nixies and tried pulsing them to see how they looked and worked when they flashed. Just fine as expected. I was a bit concerned about using the 1Hz signal line as that is obviously used by the clock itself. I didn’t want to weaken the line with two outputs, or have to shore them up with opamp buffers. I did discover something neat. The colons on a digital clock that we’re all used to, don’t flash at 1Hz. I realized that they flash at 0.5Hz or once every 2 seconds. This makes perfect sense, as the faster those colons flash, the more irritating and distracting it will be. So that was an easy problem to solve as the frequency divider has an 0.5Hz output already there for me to use.

In other news… on the same topic

Having pretty much finished all of the legwork and testing for the next big part of the project, it was time to move away from playing with the bright shiny bits and do some soldering. I actually find this soothing and fun, especially since my soldering doesn’t royally suck anymore.

Amazing. I'm almost not-shit at soldering.

Amazing. I’m almost not-shit at soldering.

It’s quite neat and fast once you get a feel for it. Also, as is often said – can never have enough flux.

I’ve also sorted all the routing and finalized what’s going to be on the first board, and how and how much I’m going to link to the second. I’ve sorted out that I need a 6-pin pcb connector that will transfer (in order) 170V, GND, 5V, Carry Out from the 10s seconds, 0.5Hz for the neons, and 4Hz for setting the hours and minutes (more on that later). I will also need an additional 3 pin connector to go to and from the LEDs which I had almost forgotten about, but it works out anyway since an 8-pin and 3-pin pair of connectors is more common than a single 9-pin connector.

With all that in mind, I soldered up the sockets for the two 4017 decade counters that will handle counting the seconds, as well as soldered down a bunch of loose connections as well as that 6-pin connector I mentioned.

Getting there. You can see the two sockets for the 4017s and the 6 pin connector. The yellow wires are all clock signals from the frequency divider.

Getting there. You can see the two sockets for the 4017s and the 6-pin connector. The yellow wires are all clock signals from the frequency divider.

I only have to solder in some connections and passives around the 4017s, as well as the LED driver circuit and its 3-pin connector, and that’s the first board done!

Wow. Nice to see progress and good progress at that. After several year hiatus I actually feel like this project will be finished someday soon.

I still need to find a suitable case for the thing… and figure out how to mount the Nixies.

Comments are closed.