Multiplexing Minitron incandescent displays?

[dexters_lab]

I have some Minitron 3015F 7 segment displays i want to use in a clock and will want to multiplex probably 4 or 5 of them together.

http://www.decadecounter.com/vta/articleview.php?item=218

given they are just tungsten filaments is there any issue on longevity when you multiplex or use pwm on them?

[Ian.M]

They aren't so simple to multiplex.  You need a diode in series with each segment's filament of each digit to prevent reverse conduction through all the other digits illuminating the filaments that are meant to be off of the current digit.   Apart from that, to maintain a usable brightness you will need to maintain the same RMS voltage in spite of the 20% or 25% duty cycle, which will need approximately double their rated supply voltage.  That's not a problem until your multiplexing fails which would result in rapid burnout.   In view of the complexity of the diodes, high side switching for the digits, missing pulse detector and inhibit to protect against multiplexing failure and high current segment drivers,  wouldn't it be simpler to just use one 74HC595 shift register per digit and drive them direct with DC?

[dexters_lab]

They aren't so simple to multiplex.  You need a diode in series with each segment's filament of each digit to prevent reverse conduction through all the other digits illuminating the filaments that are meant to be off of the current digit.   Apart from that, to maintain a usable brightness you will need to maintain the same RMS voltage in spite of the 20% or 25% duty cycle, which will need approximately double their rated supply voltage.  That's not a problem until your multiplexing fails which would result in rapid burnout.   In view of the complexity of the diodes, high side switching for the digits, missing pulse detector and inhibit to protect against multiplexing failure and high current segment drivers,  wouldn't it be simpler to just use one 74HC595 shift register per digit and drive them direct with DC?

aye, yea quite probably! Thanks for the nudge in the right direction 

[Macbeth]

...and use a CPLD to implement all that 74 series logic too!

I'm beginning FPGA stuff with a £5 Altera MAX EPM240 board from ebay. Using another dirty cheap USB Blaster and the free Quartus II. Great fun! 

PS: 1000 posts, welcome to the EEVBlog Supporters club / Area 51   

[Ian.M]

Thinking about it, a 74LS595 may cope better than the 74HC595 with the transient overcurrent when switching on a cold filament.  Some experimentation required . . . .

Incidentally, PWM /OE for brightness control.   

[dexters_lab]

i was working out the cold current this morning which is what made me start thinking about the possibility of multiplex/pwm maybe shortening the life, there is no mention in the 2 page datasheet about it

Cold filament resistance is about 98ohms, i tried a 100ohm resistor in series with it which should make the cold on current about 25ma with a supply of 5v, dropping to just under 7ma when it's fully on, and doesn't affect the brightness much.

PS: 1000 posts, welcome to the EEVBlog Supporters club / Area 51   

  yea i noticed i tipped over the 1k posts, pretty slow going considering i was only the 83rd person to join the forum!

[Pjotr]

Don't multiplex those ones as mentioned before. An IspMach CPLD will do the decoding trick for a few bucks. Development/programming software is free and those are ISP programmable through a simple parallel port interface. With a 32 macrco cell one you build a decoder with a SPI interface or whatever you want.

Keep in mind that those displays need a voltage drive. The filaments itself act as a current source and it is tricky to drive them with standard current drive LED drivers (run away --> blow out) if not voltage limited. Also diving them at 10% lower voltage than rated, multiplies lifetime by a factor of 10 or so.

[SeanB]

Use CD4050 buffers to do the digit drive. That way you can have a separate filament supply that you can vary from 3-6V to dim the display, and the buffer is able to drive the display properly with no problem about current, and the slow turn on from the buffer resistance will prolong the display life. The 4049/4050 are a lot more capable drive wise than most other CMOS or TTL. Only drawback is you need 1 1/6 devices per digit, or 1 1/3 if you want a decimal point as well.

Moving map display, which used IIRC 40 4050 buffers on a board right behind the display, which used socketed filament displays that had 10cm long PTFE leads you soldered to the board.There was a conversion kit to red LED diplays, which was 2 PCB in a cordwood sandwich and over 200 330R 1/8W resistors in it. Was not nice to install, plus was too bright, so I did a mod using 2 1N5406 diodes to lose 1V2 from the drive. The incandescent lamp dimming was by varying the buffer power supply from 2V to 5V using a toasty power transistor supplied from the 28V supply. Not as bad as the main 150W lamp dimmer, which used PWM, but which would still get hot enough to unsolder the BUX20 switch transistor. The 2 10A solder in fuses never stood a chance......