Pic driven nixie power supply.

[Pommie]

I am in the process of building a nixie clock and started with the power supply. The tubes I'm using are IN14s (NH14) which need 170V @ 2.5mA. As I didn't have a suitable switching chip I used an 8 pin pic chip with software pwm and used the comparator to stop at the right voltage. The results are pretty impressive. The comparator on the pic keeps the voltage very steady and fully loaded it can supply 25mA - enough for 10 nixie tubes. If anyone is interested I can draw the schematic and post the code. The component selection is a bit random as I used what I had and only bought an IRF840 for the high voltage bit. Now I just need to build the rest of the clock - I'm thinking of using three 18F2620 to drive two tubes each via serial with ksp44 transistors.

Mike.

Edit, miscalculated mA.

 

[killivolt]

I am in the process of building a nixie clock and started with the power supply. The tubes I'm using are IN14s (NH14) which need 170V @ 2.5mA. As I didn't have a suitable switching chip I used an 8 pin pic chip with software pwm and used the comparator to stop at the right voltage. The results are pretty impressive. The comparator on the pic keeps the voltage very steady and fully loaded it can supply 25mA - enough for 10 nixie tubes. If anyone is interested I can draw the schematic and post the code. The component selection is a bit random as I used what I had and only bought an IRF840 for the high voltage bit. Now I just need to build the rest of the clock - I'm thinking of using three 18F2620 to drive two tubes each via serial with ksp44 transistors.

Mike.View attachment 88763
Edit, miscalculated mA.

I was telling someone awhile back it would be interesting to build a binary clock out of nixies. But this is cool; will it be coded in c? If so, I would be interested.

Thanks, Pommie.

kv

 

[ronsimpson]

Very interesting.
I would like to see more information.
thank you

 

[killivolt]

I think it would be a cool Blog

 

[BobW]

Yes, very interesting. Please keep us posted.

Here is another option for driving the Nixies. A few years ago, I built a Nixie frequency counter, and used Allegro A6818 high voltage drivers. These are serial input chips with 32 high voltage parallel outputs, which eliminate the need for additional high voltage transistors, and also reduce the number of PIC I/O lines to just 3: data, clock and strobe. One A6818 is enough to drive a 4 digit clock (hour & minutes only) and can be cascaded for more digits. Unfortunately, the A6818 is out of production, but a similar chip is available from Supertex, the HV5812 which is also a serial in, parallel out, but with 20 discrete HV outputs instead of 32. So, you would need one HV5812 for every two digits—three for a six digit clock. These used to be somewhat pricey, but have dropped in price recently and may work out to be less expensive and certainly fewer parts than other decoder/drivers combined with HV transistors.

 

[Pommie]

I like the look of the HV5812 from Microchip. As I've only done the supply at the moment I might use those for the actual tubes. Edit, not suitable as can only handle 80V.

For anyone interested, here's the schematic for the supply I'm building.

The first fet is a 2n7000 to do the level shift for the IRF840. The LED lights when it's up to voltage - or rather goes out if it's overloaded. The inductor is the most important thing - mine is 100uH 3A - RF chokes won't work. The diode needs to be a fast recovery type, I used a uf4004 and the cap is 4.7uF 400V - bigger just makes it more dangerous and isn't needed. The trim pot varies the output from about 120V to 400V.

The code is really simple,

#include "p12F683.inc"

; CONFIG
__CONFIG _FOSC_INTOSCIO & _WDTE_ON & _PWRTE_OFF & _MCLRE_ON & _CP_OFF & _CPD_OFF & _BOREN_ON & _IESO_ON & _FCMEN_ON

        cblock  0x20
temp
        endc
        org    0
        bsf     STATUS,RP0      ;bank 1
        movlw   0x70
        movwf   OSCCON          ;8MHz
        movlw   b'11101011'     ;GP2 = LED GP4 = 2N7000
        movwf   TRISIO
        movlw   b'10101001'     ;09/24*5V
        movwf   VRCON
        bcf     STATUS,RP0      ;bank 0
        movlw   b'00000011'
        movwf   CMCON0
loop    btfss   CMCON0,COUT     ;if clear then wait
        goto    loop            ;as at required voltage
        clrwdt                  ;kick the dog
        bcf     GPIO,4          ;turn on for 30uS = 60 cycles
        movlw   18
        call    delay
        bsf     GPIO,4          ;and off for 6uS+
        movlw   3
        call    delay
        goto    loop

delay   movwf   temp        ;1      delay = 3*w+5
dloop   decfsz  temp,f      ;1/2    including call
        goto    dloop       ;2      and loading W
        return              ;2

        end

Have fun,

Mike.

 

[Pommie]

And here's a quick and dirty board.

Mike.

 

[BobW]

I like the look of the HV5812 from Microchip. As I've only done the supply at the moment I might use those for the actual tubes. Edit, not suitable as can only handle 80V.

FYI: The old TTL SN7441 nixie driver chips were only rated for 70-80 volts. That's all that's required, since the tube itself drops most of the HV supply, so the driver doesn't see much voltage.

Thanks for the schematic.

 

[Pommie]

FYI: The old TTL SN7441 nixie driver chips were only rated for 70-80 volts. That's all that's required, since the tube itself drops most of the HV supply, so the driver doesn't see much voltage.

Thanks for the schematic.

That's true when the digit is lit but when it isn't then the 170V must be across the chip or did I miss something?

Mike.

 

[BobW]

Very good point. I questioned this myself when I built mine. So, I did some measurements before I committed myself to the HV driver chips. I connected a HV supply to the nixie anode (through a 27k current limiting resistor) and with the cathodes open circuited, I measured the cathode voltage with a high impedance (several hundred megohms) multimeter. I never measured more than about 70 volts on any cathode. Interestingly, I found that if one cathode was grounded it would change the open circuit voltage on adjacent cathodes a bit, sometimes higher and sometimes lower, but not enough to cause problems. FYI, the A6818 chip (see attachment) that I used were only rated for 60 volts, but I was using IN-16 nixies which work quite well with a 150 volt supply. So, I kept the supply limited to about 150-170 volts and the open circuit voltage on the cathodes never went above the chip's rated voltage. I've been using this thing for several years now, and haven't had any problems.

My conclusion is that it takes less than a microamp of leakage through the switching device to drop the voltage to safe levels, and that low current level isn't enough to have any effect on the driver.

BTW, I used an 8-pin PIC12F629 for the smarts.