Hello,

I made a nixie power supply according to this pdf: http://www.ledsales.com.au/kits/nixie_supply.pdf
I also ordered 100uH 3A inductors (yellow-white core) from ebay: https://www.ebay.com/itm/Toroid-Cor...hash=item3ae868ad43:m:mfDzHpVRdotUD3p8NDfPf1g
Nixies: http://www.tube-tester.com/sites/nixie/dat_arch/Burroughs_Bulletin_1104D.pdf
The only difference to the schematic is that I use UF4007 diode. The input voltage is given from a wall adapter of a router(12V, 500mA).
The problem is that the inductor gets really hot (~80-90°C) after a minute, which I think is abnormal, but every value of the components are ok.
Loaded with 3 nixies: Vin=11,7V Iin=0,32A Vout=175V Iout=5mA f=~30kHz
I also tried salvaged inductors, but those also get hot after a few minutes ( sometimes the FET also). The only inductor which only gets warm is a hand wounded one, on a ferrite core (d6*30mm) with ~80 turn in 2 layers.

The question is why the toroid inductor get hot or how can I evade the warming?

 

The UF4007 is fine.
Did you build this on the kit PCB or is this your own construction?
Did you get any output voltage at all?
Can you give us some good pictures that will allow us to trace the connections and read resistor values? We might see something wrong.

 

Hello,

I made a nixie power supply according to this pdf: http://www.ledsales.com.au/kits/nixie_supply.pdf
I also ordered 100uH 3A inductors (yellow-white core) from ebay: https://www.ebay.com/itm/Toroid-Cor...hash=item3ae868ad43:m:mfDzHpVRdotUD3p8NDfPf1g
Nixies: http://www.tube-tester.com/sites/nixie/dat_arch/Burroughs_Bulletin_1104D.pdf
The only difference to the schematic is that I use UF4007 diode. The input voltage is given from a wall adapter of a router(12V, 500mA).
The problem is that the inductor gets really hot (~80-90°C) after a minute, which I think is abnormal, but every value of the components are ok.
Loaded with 3 nixies: Vin=11,7V Iin=0,32A Vout=175V Iout=5mA f=~30kHz
I also tried salvaged inductors, but those also get hot after a few minutes ( sometimes the FET also). The only inductor which only gets warm is a hand wounded one, on a ferrite core (d6*30mm) with ~80 turn in 2 layers.

The question is why the toroid inductor get hot or how can I evade the warming?

If the UF4007 isn't hot then it is fast enough for the job.
Most yellow toroids are for RF suppression, they suppress the RF by turning it into heat (being lossy), these aren't suitable for your switch mode power supply. Use radial open core coils, like this one http://uk.farnell.com/coilcraft/rfb0807-101l/inductor-power-100uh-1a-10/dp/2457723 and like the one in the finished product. The advantage of these cores is that they don't saturate easily. The only advantage of toroid is that they radiate less magnetic field.

 

The UF4007 is fine.
Did you build this on the kit PCB or is this your own construction?
Did you get any output voltage at all?
Can you give us some good pictures that will allow us to trace the connections and read resistor values? We might see something wrong.

I wrote in my opening post: "Loaded with 3 nixies: Vin=11,7V Iin=0,32A Vout=175V Iout=5mA f=~30kHz"
Here are the images of the circuit assembled on breadboard:
I have a blue-gray core, but with that the FET gets hot also. With my hand wounded one it works quite well, the coil is just warm, but thats a bit big. With this the operating parameters are:
Vin=11,87V Iin=0,163A Vout=175V Iout=5,2mA f= ~30,6kHz

I'm trying to find a radial type inductor in the local shops and try that out.

 

I wrote in my opening post: "Loaded with 3 nixies: Vin=11,7V Iin=0,32A Vout=175V Iout=5mA f=~30kHz"
I have a blue-gray core, but with that the FET gets hot also. With my hand wounded one it works quite well, the coil is just warm, but thats a bit big. With this the operating parameters are:
Vin=11,87V Iin=0,163A Vout=175V Iout=5,2mA f= ~30,6kHz

I'm trying to find a radial type inductor in the local shops and try that out.

I would also say that it is the coil that is still the problem. However, the efficiency after switching to the hand-wound coil is still below 50%.

Do you have access to an oscilloscope? - because the coil current waveform is a really important when troubleshooting.

 

I managed to use an oscilloscope, I used an 1ohm shunt resistor to measure the coil current (1V->1A).
Here are the waveforms of the toroid from ebay:

And the hand wounded one:


I made the circuit model in multisim also, there I calculated ~63% efficiency with 4 nixies (used 4 resistor to get the correct output current)

 

Fine to get some curves to illustrate what happens.

I have a few comments to the circuit: I assume that the ground of the oscilloscope is connected to + 12V, the upper curve is the junction between the shunt resistance and the coil, and the bottom curve is the MOS-FET Gate signal.

If above is true, it seems that the ring core coil charge reaches 1.5 Amps and starts to "round off" due to the resistance of the coil.
The rod shaped core coil charge looks quite linear and only reaches 1.0 Amps.
Both coils discharge to the high-voltage output, looks immediately very fast - I would expect at least a quarter division time.

And finally, some comments on the measurements: Shunt resistance is big, I would have chosen no more than 0.1 Ohm. I see that the short ground for one of the probes is not connected, and it may give a little extra ringing on the oscilloscope.

 

Fine to get some curves to illustrate what happens.

I have a few comments to the circuit: I assume that the ground of the oscilloscope is connected to + 12V, the upper curve is the junction between the shunt resistance and the coil, and the bottom curve is the MOS-FET Gate signal.

If above is true, it seems that the ring core coil charge reaches 1.5 Amps and starts to "round off" due to the resistance of the coil.
The rod shaped core coil charge looks quite linear and only reaches 1.0 Amps.
Both coils discharge to the high-voltage output, looks immediately very fast - I would expect at least a quarter division time.

And finally, some comments on the measurements: Shunt resistance is big, I would have chosen no more than 0.1 Ohm. I see that the short ground for one of the probes is not connected, and it may give a little extra ringing on the oscilloscope.

Yeah, top curve is the current of the coil, bottom curve is the Gate signal.

I connected that to the ground on the circuit. I also tried the connection of the oscilloscope ground to the +12V, everything was the same, except that the coil current was a bit lower, with around 0,1-0,2Amps. That ringing only showed with the hand wounded rod core only. I tested two other scavanged toroids, those also didn't have that ringing, but warmed also.

I've only got 1ohm resistance with me, but I think It worked well.

The two coils have quite similar wire diameters and the toroid one was bought as a 100uH 3A coil. While I made the two picture it got really hot and that was within a minute.

I'm trying to get a radial type coil from local shops and repeat the measurement with that also.

I don't want to start the logic part of the clock until I don't have a correct PSU for the nixies.

 

Hello,

I made a nixie power supply according to this pdf: http://www.ledsales.com.au/kits/nixie_supply.pdf
I also ordered 100uH 3A inductors (yellow-white core) from ebay: https://www.ebay.com/itm/Toroid-Cor...hash=item3ae868ad43:m:mfDzHpVRdotUD3p8NDfPf1g
Nixies: http://www.tube-tester.com/sites/nixie/dat_arch/Burroughs_Bulletin_1104D.pdf
The only difference to the schematic is that I use UF4007 diode. The input voltage is given from a wall adapter of a router(12V, 500mA).
The problem is that the inductor gets really hot (~80-90°C) after a minute, which I think is abnormal, but every value of the components are ok.
Loaded with 3 nixies: Vin=11,7V Iin=0,32A Vout=175V Iout=5mA f=~30kHz
I also tried salvaged inductors, but those also get hot after a few minutes ( sometimes the FET also). The only inductor which only gets warm is a hand wounded one, on a ferrite core (d6*30mm) with ~80 turn in 2 layers.

The question is why the toroid inductor get hot or how can I evade the warming?

Apparently there's a new thing called; "Lixie displays" - stacks of engraved transparent slides edge illuminated by multi colour LEDs.

One manufacturer dismantled a scrap Nixie tube and scanned the number electrodes to get the original Nixie font.

The ones I heard about have onboard controllers with serial interface that can be daisy chained, but I'd be surprised if there were no bare LED versions.

 

I managed to buy some radial type inductors, and I also tested the circuit:
The inductor, and the mosfet, gets warm, can be touched. Efficiency is the same around 50-60 %.
I want to assemble it on PCB and test it, I made my own arrengement for it:

 

On your PCB, it is a good idea to keep the tracks from the coil, switching transistor and main filter cap as short as you can. Not long as it is shown there.
Also, it would be worth while running the switch transistor track back to the supply cap on its own track, with nothing else connected to it. Use a seperate track for the other bits. The idea is to keep the high switching currents away from the control and sensing tracks and that improves the stability of the circuit.
There are some good design notes on line for switchmode supplies. And BIG tracks for low resistance and inductance help.
Have you used any software to calculate the coil inductance required?

 

Nope, I used the values from the pdf.

 

On your PCB, it is a good idea to keep the tracks from the coil, switching transistor and main filter cap as short as you can. Not long as it is shown there.

Yes, the pulse current loop should the smallest area possible. On this board that loop goes all the way round the outside of the board - the maximum area possible.

 

Is it better? The one before was similar with the one in the pdf. Trace widths are 1,5mm and 0,5mm.

 

That looks a lot better
If you can get the track from Q2 to go back to C2 and not the bottom of Q1 it may be better. The idea is to keep the high and low currents running in different tracks.
And add a 100nF cap on the 555 between pins 1 and 4 as power supply bypassing.

 

If you can get the track from Q2 to go back to C2 and not the bottom of Q1 it may be better. The idea is to keep the high and low currents running in different tracks.

Maybe that track could go clockwise round the board back to C4?

 

Hello,

As for the diode, the following project uses the BAV21, UF4004, UF4007, MUR140, or MUR160 diodes:
https://threeneurons.wordpress.com/nixie-power-supply/

bertus