Vacuum fluorescent display to clock display

[Ian.M]

@neo,
So does your code drive multiplexed 7 segment LEDs O.K.?  If not, don't bother building a VFD driver circuit!   (Although you've got a 14 segment 'starburst' display it can be wired as 7 segment for initial experimentation by paralleling the two center horizontal segments and grounding the rest of the radial segments, reducing the number of HV pin drivers you need to build by a third.)

@BuriedCode:  12V DC to 100V boost circuits capable of providing up to 100mA at up to 100V  (figure on 14 segments + a grid @6ma each) without a tapped inductor or transformer are possible.  A simple boost converter with a beefy external chopper transistor^* feeding a Cockcroft–Walton voltage multiplier in place of the normal rectifier will do it - use a doubler for 32 to 66V and a tripler for 66 to 100V.  Efficiency wont be great, but as long as it doesn't overheat, who cares for a hobby project?  Unfortunately that means you'll need a separate AC supply for the filament, but you can use a H-bridge + PWM for that, alternating the 'direction' on each PWM cycle to maintain a net DC balance while letting you vary the filament current under software control.   The whole lot can be done in the same MCU as the display control if it has two PWM modules (one full-bridge), a comparator that can trip off the PWM output for cycle-by-cycle current limiting, and an ADC.

* Beefy & external,  because it has to withstand the charging current spike of the caps in the multiplier every time it turns on.

[neo]

i think i have the power supply figured out, nothing stops me from using multiple transformers of which i have plenty, i also made my own chip out of transistors (schematic/explanation below)
 the arduino code that makes my plan work for those into that kinda thing (tested, not on the vfd yet but ive yet to find any faults in my design or plan)
 <snip>
thank you freddy for that part number, but overall what do you think of my solution?

If you're going for discrete transistor solution, then you'll need help from others on this forum as I'm not an EE, more programmer !

my experience of VFD's is from this project which I reproduced

I can't see a schematic so it's not really possible to put the code you posted into context

it isnt just you i forgot to put it there somehow also as for the code all it does is switch individual transistors (or in the case of my schematic layers of transistors) on and off very fast

[neo]

schematic revision

[Ian.M]

Your level shifting and digit and segment drive circuits don't make any sense.

Draw ONE segment and ONE digit's circuit in full with all transistor symbols, all resistors, all component values and part no.s, and named input and outputs e.g, from the same segment in other digits, from the 4026 chip for the segment signal and from the multiplexing pulse generator for the digit select signal and to one each of the VFD's digit grid and segment anode pins. 

My best guess is, depending on the transistor arrray types, your current schematic will either do nothing or blow the 4026 chips and multiplexing pulse generator immediately on switch-on! 

[Ian.M]

You'll need something like this to actually drive the grids and segment anodes effectively without frying your logic:

It can do 250ns 10%..90% rise and fall times when swinging 100V into a 10mA 100pF load.  It pulls low to 0.5V of ground, and depending on the load current, high to within 3V of +HT.   For operation up to 100V, R1 should be rated 1/2W.

You also need to maintain a positive bias on the cathode (filament) which is commonly done by putting a Zener between the CT of the filament transformer secondary and ground

[neo]

Your level shifting and digit and segment drive circuits don't make any sense.

Draw ONE segment and ONE digit's circuit in full with all transistor symbols, all resistors, all component values and part no.s, and named input and outputs e.g, from the same segment in other digits, from the 4026 chip for the segment signal and from the multiplexing pulse generator for the digit select signal and to one each of the VFD's digit grid and segment anode pins. 

My best guess is, depending on the transistor arrray types, your current schematic will either do nothing or blow the 4026 chips and multiplexing pulse generator immediately on switch-on! 

All transistors are MPSA42 and i do plan to use current limiting resistors on the transistor bases i just didnt draw them in

[neo]

why 100 volts? where does the signal enter in that schematic and yes my code was able to drive a multiplexed led

[helius]

Vbb, the voltage between the anodes and the filament (cathode), needs to be high enough to accelerate the electrons over a fairly short distance (~5mm). Typically it's between 45 and 90 V. Using 100V to run the driver allows for some losses in the switching transistors.

[Ian.M]

V1 provides a 5V pulse simulating a logic signal.  100V is the maximum voltage you are ever likely to need.  Some VFDs can work with anode to cathode (filament) voltage differences (Va-k) much lower than that, as low as 30V for some types of multiplexed display.  However the off-state grids and anodes need to be pulled far enough negative with respect to the cathode to ensure they are fully cut-off even at the negative going peak of the filament drive waveform.  This means the mean cathode voltage probably needs to be somewhere around +7V to +12V, which adds to the +HT voltage required.  Somewhere in the 40V to 50V range for HT is common for planar VFDs.  Russian tubular VFDs generally run at higher voltages for pulsed (multiplexed) operation - up to 80V Va-k in some cases, hence the design requirement for a 100V capable driver.

You *REALLY* don't want to over-drive the anodes or grids as excess Va-k or Vg-k can cause sputtering and rapid degradation of the phosphor, the high emissivity filament surface treatment and even darkening of the glass window.  Long term under-driving is also bad - it causes cathode poisoning and loss of emission.   Over-running the filament obviously shortens its life and Noritake specifically caution against under-running the filament as a method of brightness control.   

Must read: Noritake's A Guide to Fundamental VFD Operation

Establishing the correct filament voltage without a datasheet can be difficult - as a rough rule of thumb you shouldn't be able to see it glow in a darkened room.   It would be easier if enthusiasts with VFDs they have datasheets for or OEM displays in equipment in good working order would measure the cold resistance of their VFD filaments, and post that with either the typical filament voltage and current from the datasheet or measured in original application circuit with a true RMS meter that can handle the frequency of the filament drive, so we can calculate the ratio of cold and hot resistances as this should be more or less constant for all tugsten filaments running at the same temperature.

I haven't yet found datasheets for the VFDs I've got on hand, but the cold filament resistances and physical description are as follows:

Make, Model, Dimensions (ex. pins and 'pip'), Digit/Char height, Cold Resistance
Description

NEC, FIP13C8C, 112mm x 26mm x 8mm, 8mm  20 ohms
'Bathtub' cover glass and flat glass back with visible interior metallisation traceable to the filament and each segment and grid.
12 digits + M, minus and left arrow at leftmost position,  10 digit anodes (7 segments, dot/comma on right and apostrophe on left of each digit. The M - and left arrow have separate anodes.), 3 parallel filaments, 28 pins bonded to one edge of the back glass, 0.1" pitch arranged in two groups of 14 with a 5 pin gap centered on the evacuation 'pip' as follows in one line from left end:


1 2  3  4  5  6  7  8  9  10 11 12 13 14
F Am A- A' A< Gs Af Ag Ae G1 G2 G3 G4 G5

5 pin gap (pip)

15 16 17 18 19  20  21  22 23 24 25 26 27 28
G6 G7 G8 G9 G10 G11 G12 A, Ad A. Ac Ab Aa F

All Aa-Af anodes according to usual seven segment notation.

Futaba, 16-SD-01G, 97mm x 20mm x 6mm, 5mm, 56 ohms
Planar with visible multilayer metallisation not traceable to pins.  Pins fused through frit seal. pip on end side, bottom right
16 char 5x7 dot matrix,  presumably 35 dot anodes. 16 grids, 2 parallel filaments, 64 pins, 31 along top edge and 33 along bottom edge.  Filament is end pins of 33 pin edge.  Grids also appear to be on this edge.

[neo]

i just noticed on the example in the simplified schematic i forgot to connect the segments to the display but it is just the example

[Ian.M]

UUUUGGGHHHHH!!!

Iv'e traced ONE segment and digit from your schematic and redrawn it as I asked you to:

Edit: Q1 should be marked Q85-92

It cant work.  Its incapable of delivering more than 3.8V to any VFD electrode, and will do so to the selected grid and *ALL* the anodes!

You obviously don't 'grok' bipolar transistors! 

I believe you are using various segment outputs of your 4026 counters/7 seg. display drivers for your clock logic, so you cant simply gate the 4026 chip's display enable pins with your multiplex digit select signals and diode OR their outputs together to feed the segment anode drivers.   Ideally you'd rip out all the 4026 chips and the associated logic and rebuild with 3x 4518 dual BCD counters, feeding 3x 4508 dual 4 bit latches with tristate outputs (with their STROBE and OUTPUT DISABLE pins tied together and driven by an active low 1 of N digit select signal) to multiplex the BCD into a single 4055 7 segment decoder with its DF pin tied high for active low segment outputs.  Get the digit select lines from a 4017 and some 4049 inverters.  The active low segment and digit signals can directly drive an instance of the VFD level shifting driver circuit I posted earlier.

Your best way forward keeping your existing 4026 based clock circuit and using only 4000 series CMOS, would be to collect your separate counter's segment outputs using a 4051 8:1 analog switch for each multiplexed segment, selected  by a 3 bit binary counter (low bits of a 4018), and inverted with 4049s to drive the VFD level shifter for the anodes.  The digit select signals to the grid drivers need a 3 to 8 line decoder with active low outputs - use a 4556 dual 2 to 4 line decoder + one inverter from a 4049 to invert the /E signal to one of them, driving the enables from the MSB of the 3 bit digit select.

[neo]

so why cant it deliver more than 3.8v?

[neo]

forgive me if im a bit dense but i think it should work as transistor 1-42 turns the logic level clock out to a higher voltage and then the transistors after that allows for the micro controller to switch it on and off, such as to say when the clock says 12:00:00 for instance then it would turn on the right transistors and then between those and the vfd are the ones switched by the microcontroller. and on the grid transistors they are linked to their corresponding set that goes to that grid.

[Ian.M]

The base-emitter junction of a bipolar transitor in normal operation is a forward biassed diode.  For a NPN, if the emitter ever rises above the base voltage, the diode becomes reverse biassed and the transistor turns off.  If you ever exceed about 5-7V reverse bias the B-E junction tends to Zener which usually permanently damages the transistor.

Referring to my redrawn version of your schematic, IOn are 5V logic signals so Q93-100 emitters can never be driven above 4.4V (normal emitter follower behaviour).  Therefore the emitters of Q85-92 driving the grids and Q43-84 driving the anodes can never rise above 3.8V (assuming minimum Vbe of 0.6V).  Additionally it doesn't matter if the segment lines to Q1-42 are high or low, or even if Q1-42 are present or shorted out as when they are off, the B-E diode junction of Q43-84 lets their base current pull their emitters up to the same voltage, no matter what's happening at their collectors. 

[neo]

so how DO i make it work? i got an idea what if i used relays?  Thank you very much ian by the way i know im slow but to my credit i jumped straight off the shallow end and into the ocean. this is my first circuit anywhere near this complex i guess its a bit of a trial by fire.

[neo]

relays would work speed wise, and i could be operated at the voltage i need and give me up to 100v out at the amperage id need, so what would be wrong with this idea? besides the cost that is

[neo]

my plan to cope with high voltage unless theres a better/ easier way

[Ian.M]

Relays wouldn't work (well maybe mercury wetted reed relays would).   You are in the USA so have 60Hz mains.  You need AC for the filament so for simplicity will be using a low voltage transformer, with a series resistor to adjust the filament current, and a pair of fairly low value resistors across the filament to fake a center tap and keep it balanced about its nominal Vk.   Something like:

Caution: verify the actual filament voltage and current to get its hot resistance using a DC bench supply and check the transformer output voltage before choosing R1.  It may need a series dropper resistor to keep R1 dissipation within the range of presets you can afford, or you my even be able to use a fixed resistor.  Getting it wrong at best will burn up R1 and at worst will blow your VFD filament

If you'd read the Noritake link I gave (section 4) you'd know that with a 60Hz filament supply the recommended scan rate is >100Hz.  You have 8 digits to scan, so each digit must be on for no more than 12.5ms.  You would need relays that can switch sufficiently faster than that to allow for inter-digit blanking to prevent ghosting, so 3ms is a reasonable target.  That's 6ms to open and close again which is equivalent to 167Hz, about two or three times faster than you can push a normal small relay.  If you find any that can handle it, they'll probably only be rated for a life of 50M operations, and at 100Hz scan rate, you'll wear them out in a bit under 6 days!

Back in reply #35 I suggested 4000 series logic chips and  two alternative general strategies to provide appropriately multiplexed signals to control discrete transistor VFD drivers (like I suggested in reply #29), or you could simply start over with an Arduino and a couple of the Maxim VFD driver/controller chips suggested earlier.

Do you have the make and model of the Stereo your HNA-08MS16 displays were salvaged from so we can find a schematic and get the voltages needed to run your VFD?  If not, I found a Samsung service manual for a portable stereo containing a Samsung HNA-08SS23 display which uses 2.3VAC Vf  and +5V and -25V giving a Va-k of 30V, so that would be a good place to start.  You can use an adjustable CC/CV DC bench supply for the filament for initial testing and a stack of PP3 9V batteries on top of the adjustable output of another bench supply to the the anode and grid voltages up where you need them.   Once you have it lighting up a few segments of one digit, build one of the two transistor driver I proposed, and pulse it with a 555 to give a 1:10 duty cycle (10% ON time) and see if the HT supply voltage needs adjusting to get acceptable brightness.   If the final HT voltage is under about 40V there are a *LOT* more options for chips that can drive the anodes and cathodes.   

[neo]

lets say i just give up on the original clock schematic i dont care at this point how it has a clock so long as it does and it can drive the displays my original 4026 is apparently getting nowhere so what do i do?

[Ian.M]

Well the *easy* solution for anyone who can code in C is to string together enough MAX6921 20 output VFD drivers to handle your display and use an Arduino and a RTC chip.  Hackaday has featured several projects like that, or use a pair of MAX6920 12 output ones, cascaded to act as one 24 bit register.  That gives you 24 output lines - perfect for your 25 pin display as 2 pins are the filament.   Digikey has stock of MAX6920 in the USA.

Alternatively, you can build up 23x the discrete driver I posted (or 15x if you parallel anodes to use it as seven segment only) and drive it from anything that can produce the required active low  multiplexed segment patterns and digit pulses.  I'd use a 74HC138 for the digit pulses with a 3 line digit address + a PWM intensity and blanking control feeding one of its enable pins.  That gets it down to 4 digital outputs, which means you need 18 total to run the display.  An Arduino Uno has 20 pins that can be configured for digital I/O, so that's a viable option allowing two spare pins, one for a resistor ladder of buttons and the other to drive an alarm buzzer.  Hint: use digital pins 0 and 1 for  74HC138 address pins as if you use them for a lo-Z load you cant send sketches to it anymore!

The display + my discrete drivers would be signal compatible with an 8 digit multiplexed common anode LED display with high side MOSFET or bipolar digit drivers so build one of those first, either in LSI CMOS logic or as an Arduino shield so you don't have high voltage hassles during development.

If you still want to do a LSI CMOS logic clock, its probably slightly easier in 74HC series as that has the best of 4000 series CMOS + all the standard TTL functions as well.  e.g.its got a nice octal transparent latch with tristate outputs that makes it *MUCH* easier when bussing data around as its inputs and outputs are in order on opposite sides of the chip.   However if you want to stick with the 4000 series logic you are already using, simply download the datasheets and check availability of the chips I mentioned in reply#35 and get stuck in.

[neo]

i think im going to try this idea https://hackaday.io/project/7805-vfduino not mine but it would work mine is apparently an aberration of physics to even contemplate

[Ian.M]

That should work for you.  You need two MAX6920 chips daisy-chained (Arduino SPI SDO to DIN of first, DOUT to DIN of second, SPI SCK to CLK of both, an I/O for /CS to LOAD of both and a PWM output to PWM the BLANK pin of both for brightness control) to fully support your display, but you can get 4 digits wired as a 7 segment display working on your display as if it was his  IVL2-7/5 VFD. 

I don't like the way he's running the filament on DC, It will do for experimenting but not for long-term use on a display with more digits.

If you don't want to use a transformer, you can drive the filament off an Arduino PWM output with an 5V H-bridge capacitively coupled to the filament (at both ends).  Use the same fake center tap resistors and Zener as I showed for the transformer drive, and control the effective filament voltage by selecting the coupling capacitors value and tweaking the PWM frequency.

H-bridges commonly come two to a chip - use the other one to drive a small speaker off another PWM for the alarm.

[neo]

i cant get my around around those connections a schematic would be useful

[Ian.M]

See the second image down in the section 'Multiple Slaves' at https://learn.sparkfun.com/tutorials/serial-peripheral-interface-spi#multiple then read the rest of the tutorial - you'll need it!

MAX6920 <> SPI
pin naming
CLK == SCK,
DIN == slave MOSI
DOUT == slave MISO
LOAD == inverted /SS

LOAD latches on the falling edge, but most SPI chip's chip select pins are active low and latch on the rising edge, so SS is normally /SS a low going pulse for the duration of the data transfer.  You can either use an inverter to fix it or handle it in software as the Arduino SPI module doesn't handle SS in hardware.

BLANK is just an active high display blanking pin, drive it from a PWM to control the brightness. 

[neo]

what is sdo?