MC34063 nixie-tube SMPS

[hempel]

Hi,

maybe Dave remembers when I asked him about the MC34063 SMPS Controller in one of his live-show test runs. I wanted to know if it is possible to build a high-voltage power supply to fire up a nixie-tube. Here is the result:
I found this schematic http://www.bader-frankfurt.de/nix/netztnixie3.GIF, and after changing a few components (the FET and the diode where not available) I get up to 180V out of 12V. The output voltage is stable up to at least 12mA, which is enough for my 6 nixie-tubes (thats why I did not test further). I dont have as much multimeters as Dave, so I also left out the test for efficiency.
The bad thing is that the power supply badly interferes with the DCF77-time-signal which you can receive here in and around Germany. But maybe that will be better after the power supply is on a PCB without that much parasitic antennas :-)

bye,
hempel

[alm]

Cool that you got it to work.

Not sure if I'd trust those crappy alligator jumper leads with 180V, even though it's only a low current. The ones I've seen use the cheapest construction and materials you can imagine. A real PCB might certainly help by making the loop area smaller, and will decrease parasitics by at least an order of magnitude.

[hempel]

I don't trust them (they were really cheap). I always stay away with my fingers when there is high voltage :-)
Maybe I can adjust the switching frequency, so it will not interfere that much. It would be really nice to have an oscilloscope for doing that, but I will find one.

[Bored@Work]

I won't trust the way P1 is connected. If the wiper gets lose the feedback loop will be open, and the MC34063 will ramp up the voltage to very high values. Until something gives in. Maybe the FET, maybe the MC34063, maybe the breadboard, worst case maybe the user coming close to the output.

This is one of those circuits where you better use the pot as an adjustable resistor (wiper terminal permanently connected to one of the other terminals), instead of a resistive divider. And put the adjustable resistor on the high side. Which means it should be an isolated one. That way, if the wiper becomes lose the pot doesn't open the feedback loop. Instead it goes to its maximum value. If on the high side of the feedback loop this will in turn drive the output voltage low.

You have the same risk with the breadboard. Breadboard connections are notoriously bad. If the feedback loop gets open because of a bad breadboard connection you can get very high output voltages. Further, breadboards aren't rated for such high voltages. Be prepared that the breadboard goes up in flames or other such nastinesses.

Regarding the test leads. Even the quality-brand, pre confectioned,  4 mm jack,  1 mm^2 test leads I have here are just rated for 60 V, 16 A. This is because of the single isolation and blank metal parts.

You should really look for a safer prototyping construction than you currently have. A breadboard and those test leads is not a good choice.

[hempel]

Thank you BoredAtWork,

I already asked myself what will happen when the contact in the pot is bad (which isn't that unlikely), but I did not want to test it by opening the loop with a fire extinguisher next to the breadboard :-)
Anyway, I did not think about changing the design of the feedback loop until now, but I am glad I made the post so I can realize the loop in a much safer way now. Thanks again!
The PSU has moved to a perfboard now (can you say perfboard?) to test the nixie-tube driver IC together with the microcontroller and it is working well. The decoding of the DCF time signal also is working.
I think the next step is to design the PCB. I once made a PCB with eagle, but I am not shure if I like the software. Which programs do you use to make your own PCBs? I have a book with a test version of OrCAD Layout. What do you think about this program?     

[mobbarley]

when I build my nixie clock I used the processors PWM with a mosfet to buzz a coil, this rectified with a fast diode could run 6 tubes quite well. To make things a bit nicer I used the processor's adc to monitor the HT voltage and use proportion control to adjust the PWM duty cycle.

[TheWelly888]

I have some nixie tubes - is it possible to multiplex them like we do with 7 segment LEDs?

[hempel]

It is possible to multiplex the nixie-tubes, but for my clock I don't want to take a picture or a movie of it and only see one digit 

[Bored@Work]

I think the next step is to design the PCB.

Why? If it works on the perfboard, why not leave it on perfboard? Unless this is supposed to become a polished showcase. Granted, I do get strange looks because I like to leave stuff on perfboards or stripboards, or even build according to Linear Technologie's AN-47. One of the greatest application notes ever written by Jim Williams. His others are great, too.

Which programs do you use to make your own PCBs?

At home I change between Eagel, KiCad and gEDA PCB. I don't like any of them very much. At work we have people to do it :-)

I have a book with a test version of OrCAD Layout. What do you think about this program?

I have never worked with it. But before you invest time in it check if it is still available and affordable. There is little point in investing time in a test version and later finding out that the program is no longer maintained or not affordable.

[DJPhil]

It is possible to multiplex the nixie-tubes, but for my clock I don't want to take a picture or a movie of it and only see one digit 

I have some nixie tubes - is it possible to multiplex them like we do with 7 segment LEDs?

Yes, this was done much like VFD and LED displays, just at a different voltage.
See the 74141 driver chip, see here for a driver tester project with a bunch more info and a cross reference, and peek here for a huge list of links to various info and projects.

[hempel]

Why? If it works on the perfboard, why not leave it on perfboard?

It is only to gain some practical experience. I want to know that I can do it if I have to 

See the 74141 driver chip

I guess turning off the tubes can also be done with the other driver chips by switching off the chip's VCC. But maybe it will damage the chip. If I have one left I will check.

[hempel]

This is one of those circuits where you better use the pot as an adjustable resistor (wiper terminal permanently connected to one of the other terminals), instead of a resistive divider. And put the adjustable resistor on the high side. Which means it should be an isolated one. That way, if the wiper becomes lose the pot doesn't open the feedback loop. Instead it goes to its maximum value. If on the high side of the feedback loop this will in turn drive the output voltage low.

I thought about it and maybe I get it wrong, but with the pot on the high side I get the max. output voltage for its max. resistance...?

[Bored@Work]

Yes, I got that wrong. A larger resistor on the low side reduces the output.