VFD multiplexing.

[neo]

I have an idea, it probably sucks but hear me out, I have lots (a dozen or so) of old muxed VFDs that i want to use but have yet to really think of a reasonable way yet since i am awful at programming.
So i got an idea, just for learning or one offs not practical on a mass scale, the max 7219 readily available with 8*8 LED matrix modules and it is common cathode so ^{in theory} the only thing i would have to do is add the circuit _{(pictured below)} and at most make a few tweaks to the program.
If i can make if work and interface the LED, which i should be able to since there are videos on that, then i can add the transistors and interface a VFD up to 8 digits with 7 segments and a dp to each digit. Also yes i know it is common cathode so it uses negative to activate each led digit but that is a problem easily solved with a logic inverter and then transistors.
http://www.ebay.com/itm/MAX7219-8-Digital-Segment-Digital-LED-Display-Tube-Arduino-51-AVR-STM32-/272678447709?hash=item3f7ce6ea5d:g:oFMAAOSwurZZHqvW

So my idea, in summary, is to take an led module like that, which is available cheap, learn to use it as is and then take that knowledge and begin learning to interface it to a VFD. VFD multiplexing is after all the same concept as LED multiplexing, only with a higher voltage level needed and my transistor circuit, pictured below, does work. I also happen to have a few hundred high voltage transistors, both PNP and NPN, laying around.
Any obvious problems i missed?

[floobydust]

There are different types of VFD - direct drive and muxed which has control grids. Using one-digit VFD's I don't see a control grid on them.
Which are you thinking about driving?

The Ice Tube Clock used MAX6921 to drive 8-digit VFD.

I think it's an alright idea to add HV buffers to LED driver IC's.
The 5k output resistor is too small, it can be much larger 100k it only has to deal with leakage current.

https://www.noritake-elec.com/technology/general-technical-information/vfd-operation

[neo]

There are different types of VFD - direct drive and muxed which has control grids. Using one-digit VFD's I don't see a control grid on them.
Which are you thinking about driving?

The Ice Tube Clock used MAX6921 to drive 8-digit VFD.

I think it's an alright idea to add HV buffers to LED driver IC's.
The 5k output resistor is too small, it can be much larger 100k it only has to deal with leakage current.

https://www.noritake-elec.com/technology/general-technical-information/vfd-operation

I am talking about the types of VFDs like what you would see in microwaves, stereos, dvd palyers and a lot of other stuff.
Just to sum up what i have on hand, i have everything from simple 4 digit up to a 12 digit display from a calculator and even some fancy ones from a stereo.
Yes the russian tubes also are on that list, pretty much any VFD with more than a digit is multiplexed and those are the types for which i need a way to drive them.
What i find the most appealing aspect of my idea is that the driver itself is as common as it is WITH information readily available.

EDIT:
My only problem with the 6921 is the lack of videos directly about it, between that, the lack of a library for it and the cost of it it just doesn't seem like a good place to start learning SPI. As i implied i am about as good at writing my own code as a frog, i even have trouble making libraries work sometimes.
Alot of my problems with making things work might not even be problems if not for my love of obsolete/weird technology.

[Ian.M]

Reasons to use a SPI display chip that does its own multiplexing:

• My controller is running an OS that does not guarantee realtime SPI transfers so the display will flicker if I do the multiplexing on the controller.
• My SPI peripheral is tied up for longer than 1/(frame_rate*N_digits_per_display) doing bulk transfers I cannot interrupt.
• My controller is too slow to be capable of updating all digits in the time 1/(frame_rate*N_digits_per_display).
• I need extra features like individual character or digit brightness control.
• The display must freeze with the last data still displayed if the controller is reset.

(1) applies to driving it from something like a Raspberry PI or bitbanged from a PC printer port.
(2) may apply if you are doing data-logging to a SD card on Arduino.
(3) would be VERY rare - its possible when driving a large number of cheap LED displays, but the prohibitive cost of VFDs and their support circuits make it vanishingly unlikely that you'd have enough to run into the limit even if your controller is a PIC at 1MHz.
(4) is a valid reason but will push you towards expensive high end display chips, and (5) is a show-stopper that forces you to use a multiplexing display chip.

A simple SPI shift register + level translating driver chip is almost always cheaper so when none of the above apply using a fancy display multiplexed display chip only because you cant be arsed to learn to code your own multiplexing is just intellectual lazyness . . . . (or pure desperation with only a couple of days left to the project deadline!)

[neo]

Following my understanding, it can multiplex 8 digits with 8 characters per digit and all i have to do is tell it what to light up in the program. For the program i have libraries, the chip is cheap and i tried using an spi shift register before i couldn't make it do anything. At all.

[Ian.M]

Following my understanding, it can multiplex 8 digits with 8 characters per digit and all i have to do is tell it what to light up in the program. For the program i have libraries, the chip is cheap and i tried using an spi shift register before i couldn't make it do anything. At all.

If you couldn't at least toggle the outputs of a SPI shift register, your chances of doing anything other than curse at a MAX7219 are minimal.  Its got five control registeres that have to be configured to get it to do what you want which is a lot more complex than sending N bits for the required number of pins to a SPI shift register then pulsing its Load pin high.

[neo]

Following my understanding, it can multiplex 8 digits with 8 characters per digit and all i have to do is tell it what to light up in the program. For the program i have libraries, the chip is cheap and i tried using an spi shift register before i couldn't make it do anything. At all.

If you couldn't at least toggle the outputs of a SPI shift register, your chances of doing anything other than curse at a MAX7219 are minimal.

But there are libraries for the MAX7219. What am i meant to do? Give up and smash the displays just because i am near useless with micro-controllers?

[Ian.M]

IIRC we've discussed driving multiplexed VFDs off pure logic before now.  No SPI involved at all, just a job lot of latches with tristate outputs, a sequencer to step through them enabling one to provide the data for one digit at a time, a decoder for the character or digit patterns and a job lot of HV drivers..  I've also posted Arduino code that multiplexes up to 8 seven segment digits via two 74HC595 SPI-compatible shift registers.

You probably bit off more than you can chew.  Simply trying to flash a single LED connected to one of the shift register output pins would be a good start.   Yes you could try jumping straight to the MAX7219 using libraries to drive it, but if (more likely 'when') it doesn't work, you will be absolutely stuck.  That's the problem of using other people's packaged libraries you don't understand.  Your code is then all 'monkey see monkey do' with the worst elements of 'cargo cult' programming, and doesn't actually teach you anything. 

Its remotely possible you are one of the small proportion of the population that has a negative aptitude for casual logic and thus for procedural programming, but most people with that sort of problem don't do too well with the complexities of using the internet so are unlikely to be asking for help here.  Everyone else (not suffering from a diagnosed cognitive deficit) can learn to code, and whether or not they become adequate depends on the effort they put in building their way up from the absolute basics.  OTOH its generally reckoned that less than 1% have the aptitude to become a great programmer - of the level that founds a technology company or launches a game-changing open source project so don't beat yourself up because you are not the world's next Linus Torvalds

[neo]

IIRC we've discussed driving multiplexed VFDs off pure logic before now.  No SPI involved at all, just a job lot of latches with tristate outputs, a sequencer to step through them enabling one to provide the data for one digit at a time, a decoder for the character or digit patterns and a job lot of HV drivers..  I've also posted Arduino code that multiplexes up to 8 seven segment digits via two 74HC595 SPI-compatible shift registers.

You probably bit off more than you can chew.  Simply trying to flash a single LED connected to one of the shift register output pins would be a good start.   Yes you could try jumping straight to the MAX7219 using libraries to drive it, but if (more likely 'when') it doesn't work, you will be absolutely stuck.  That's the problem of using other people's packaged libraries you don't understand.  Your code is then all 'monkey see monkey do' with the worst elements of 'cargo cult' programming, and doesn't actually teach you anything. 

Its remotely possible you are one of the small proportion of the population that has a negative aptitude for casual logic and thus for procedural programming, but most people with that sort of problem don't do too well with the complexities of using the internet so are unlikely to be asking for help here.  Everyone else (not suffering from a diagnosed cognitive deficit) can learn to code, and whether or not they become adequate depends on the effort they put in building their way up from the absolute basics.  OTOH its generally reckoned that less than 1% have the aptitude to become a great programmer - of the level that founds a technology company or launches a game-changing open source project so don't beat yourself up because you are not the world's next Linus Torvalds

The thing is, i don't care about programming. All i want is to make a display work, programming comes along as a side effect of that.
Basic logic i am OK at, never could make 7400 series work the inputs being floating high and all. 4000 series i am pretty good at as evidenced by my VFD clock.
Also for what it is worth i do have a couple mental deficits, not worth elaborating on.

[Ian.M]

Fair enough. Its probably not worth pushing ahead with *ANYTHING* SPI based that requires a MCU to control it if programming is, to you,  at best a detested chore and at worst a snakepit that impacts your quality of life till you get out of it.  The extra costs of building everything in pure hardware may be acceptable.   

Alternatively there is no shame in commissioning someone else to write the code for you, (we cant all play 1st violin in the orchestra)  but its essential to agree a firm specification, agree on deliverables and deadline, and the price and payment schedule for the project as specified, rates for any specification changes or further work and discount or penalties for late completion before work starts. 

Also, if you've got someone local who can (and is willing to) help you hands on with an Arduino + MAX7219, that's a good third alternative, but you'll owe them major favour points up to and including a full cook-out with all beverages included for them and their family, and its advisable to look at some of their own projects to check their skills before committing to too much time or more than a beer or two.


If I haven't totally put you off the idea, I *ASSUME* you are still thinking of going down the Arduino based road with this - If so you'd do well to buy an off the shelf MAX7219 board with seven segment LED displays on it from a reputable supplier, and get demo code from the same supplier so you have got known working hardware to test your code on and known good code to start from for the 'bare chip' MAX7219 =>VFD board you are proposing to build.

I always found 7400 series easier than CMOS as its not generally fussy about ESD when prototyping and unless you were doing high speed stuff you could get away with letting unused gate inputs float high for the first proof of concept then add a proper pullup tied to all the unused inputs on that package before moving to the next stage.  OTOH  CMOS requires you to tie *EVERY* gate input to a valid logic level *EVERY* time  . . . .

[neo]

Call me stubborn, but i am going to get a VFD working even if i have to use 4000 series logic to do it.

[neo]

Fair enough. Its probably not worth pushing ahead with *ANYTHING* SPI based that requires a MCU to control it if programming is, to you,  at best a detested chore and at worst a snakepit that impacts your quality of life till you get out of it.  The extra costs of building everything in pure hardware may be acceptable.   

Alternatively there is no shame in commissioning someone else to write the code for you, (we cant all play 1st violin in the orchestra)  but its essential to agree a firm specification, agree on deliverables and deadline, and the price and payment schedule for the project as specified, rates for any specification changes or further work and discount or penalties for late completion before work starts. 

Also, if you've got someone local who can (and is willing to) help you hands on with an Arduino + MAX7219, that's a good third alternative, but you'll owe them major favour points up to and including a full cook-out with all beverages included for them and their family, and its advisable to look at some of their own projects to check their skills before committing to too much time or more than a beer or two.


If I haven't totally put you off the idea, I *ASSUME* you are still thinking of going down the Arduino based road with this - If so you'd do well to buy an off the shelf MAX7219 board with seven segment LED displays on it from a reputable supplier, and get demo code from the same supplier so you have got known working hardware to test your code on and known good code to start from for the 'bare chip' MAX7219 =>VFD board you are proposing to build.

I always found 7400 series easier than CMOS as its not generally fussy about ESD when prototyping and unless you were doing high speed stuff you could get away with letting unused gate inputs float high for the first proof of concept then add a proper pullup tied to all the unused inputs on that package before moving to the next stage.  OTOH  CMOS requires you to tie *EVERY* gate input to a valid logic level *EVERY* time  . . . .

When it comes to cmos having to tie them high doesn't both me in my clock i actually used every single gate in all the chips.
Also you may want to beat me with a stick but i used a library and got the 74HC95 working on arduino.

[Ian.M]

GOOD!      I'll save the  beatdown for anyone that is pushing you too far outside your comfort zone on the coding side of things, though stretching your skills a little bit each time is a *GOOD* *THING* 

Assuming you want parallel data input in BCD or HEX for a seven segment display, you'll need a 4 bit latch for each digit, (or 8 bit for an ASCII 14 or 16 segment display) with tristateable outputs, a clocked counter with N decoded outputs (where N is the number of digits) to select one digit and one latch chip's outputs at a time, a segment pattern generator - either a BCD to 7 segment decoder, or a 4 to 16 line decoder driving a diode matrix for HEX, (or a PROM for ASCII - you'll need to get help with that)  and parallel in parallel out HV drivers, one for each digit and each segment (unless you group segments in parallel to run a 14 or 16 segment display as a 7 segment one).  We've already discussed the HV drivers - Microchip still make them or you can improvise with  a >30V rated open drain or open collector buffer + a pullup resistor, a NPN transistor and a diode to boost the high state current capability.

I suggest you put together a block diagram and a proposed list of ICs and post it for comment.  Just ignore anyone who suggests 'SPI' or 'Arduino' etc.  for the core of the project^*.    

* You can always use the 74HC595 and Arduino  as an easy way of generating the test data for the inputs of the logic board.  You can get help here putting together a sketch that accepts 8 digit input from the Arduino terminal window and outputs 32 digital lines from 4 74HC595 chips.accordingly.   Heck, for test purposes only, someone will probably even write it for you!

[neo]

GOOD!   

Assuming you want parallel data input in BCD or HEX for a seven segment display, you'll need (1) a 4 bit latch for each digit, (or 8 bit for an ASCII 14 or 16 segment display) with tristateable outputs, (2) a clocked counter with N decoded outputs (where N is the number of digits) to select one digit and one latch chip's outputs at a time, (3) a segment pattern generator - either a BCD to 7 segment decoder, or a 4 to 16 line decoder driving a diode matrix for HEX, (or a PROM for ASCII - you'll need to get help with that)  and parallel in parallel out HV drivers, one for each digit and each segment (unless you group segments in parallel to run a 14 or 16 segment display as a 7 segment one).  We've already discussed the HV drivers - Microchip still make them or you can improvise with  a >30V rated open drain or open collector buffer + a pullup resistor, a NPN transistor and a diode to boost the high state current capability.

I suggest you put together a block diagram and a proposed list of ICs and post it for comment.  Just ignore anyone who suggests 'SPI' or 'Arduino' etc.

I won't lie, that confuses me. 

(1) CD4508
(2) CD4017
(3) CD4056

I have no clue how to make a block diagram for this but this is the list of ICs you asked for i do believe.
As for HV driver that answer is simple i figured out how to make my own 5V to ^{(up to)} 300v.

The only part of that i am definitely sure i understand is,

a clocked counter with N decoded outputs (where N is the number of digits) to select one digit and one latch chip's outputs at a time.

and the 4017 is a perfect chip for that.

EDIT: If you were to give me part numbers for 7400 series equivelants i may have them, i have a huge lot of 7400 chips. But as i said before, i think in 4000 series. I just don't happen to have a massive surplus on 4000 series like i do 7400 series.

[Ian.M]

The CD4508 contains two 4 bit latches with separate output disable pins.  When they are high the latch outputs are tristate. 

For each input digit use 1/2 4508.  Wire the outputs of all the 4508s together, all Q0 in parallel, then separately all Q1 in parallel etc. to form the BCD data bus.   Its crucially important (to avoid smoke!) that you only turn on ONE half 4508's outputs at a time so you need a sequential  1 of N active low signal to drive each the OUTPUT DISABLE pins.  If you use a 4017 its outputs are active high so you'd need an inverter for each 1/2 4508.

The combined Q0 - Q3 signals then go to the BCD to 7 segment decoder chip which in turn drives your HV segment anode drivers.  A CD4055 would be a better choice here because you don't have to drive a latch pin.  Its Disp. Freq pin can be tied high or low for inverted or non-inverted output respectievly which may make you HV driver interfacing easier.

Each HV digit grid driver needs to be turned on one at a time by the active 4017 (or other decoded counter) output so the digit is lit at (almost) the same time its segment data is put on the bus and decoded by the 4055.

Ideally, to avoid ghosting, you'll have a master blanking signal to the HV drivers for either the grids or the segments (or use a BCD to 7 segment chip that supports blanking) so your clock logic can turn the current digit off fractionally before it increments the 4017 count and turn it back on for the next digit after the 4055's outputs have had time to settle.  That means you need two monostables, one for the blanking pulse and the other to delay the 4017 clock input so it occurs in the middle of the blanking pulse.  A CMOS 555 is plenty good enough for the display scan clock source.

See https://www.noritake-elec.com/technology/general-technical-information/vfd-operation, section 4 for timing details.

Put that all together, and to the rest of your project, the input side of all the  4508 latches is just like driving individual BCD to 7 segment decoders each wired to a single LED digit.  Its worth noting that the 4508 doesn't need a strobe, just tie it high and the current input data goes straight to its tri-state output buffers under the control of the DISABLE signal.

Hopefully that gives you enough of an idea to sketch a block diagram - don't worry about the individual data or segment lines yet, just get the control signals right for two digits and their grids and annotate it that you have N-2 more digits the same (assuming your display has 10 or less digits.  If its got more, *DON'T* use a 4017 because they are total absolute pigs to cascade for sequential outputs without glitching).   

[neo]

If its got more, *DON'T* use a 4017 because they are total absolute pigs to cascade for sequential outputs without glitching).

what would you recommend instead in that case?

[Ian.M]

Binary output counter and separate 4 to 16 line decoder/demultipexer (which can be built from two 3 to 8 line decoders with opposite polarity enable inputs.  Connect one /EN and one EN pin from the two decoders to the MSB from the counter)

[neo]

Binary output counter and separate 4 to 16 line decoder/demultipexer (which can be built from two 3 to 8 line decoders with opposite polarity enable inputs.  Connect one /EN and one EN pin from the two decoders to the MSB from the counter)

i'll try to avoid it then.
Below is my attempt at what you described in schematic form for eagle.
Question, do the latches serve any other function other than just a digital switch?

[floobydust]

A common trick is to bias the VFD's filaments (cathode) -ve at say -6V.
This allows a single supply and if you are using 4000 (15V) CMOS, you get the extra 6V so 21V drive for lots of brightness.

The Russian IV-11 tubes need -3V for cutoff on the grid.

[Ian.M]

Can you post that as a lossless image file (PNG, GIF or compressed TIFF)  or a PDF as a lot of us don't have Eagle 8 installed?

The latches serve three purposes :-

• Capturing data all at the same time if you wire their STROBE pins in parallel and apply a brief low going pulse or digit by digit, possibly off a shared input bus if you wire their STROBE lines individually.
  .
• Freezing rapidly changing input data - you could build a stopwatch by simply adding a crystal controlled counter chain and use the STROBE line to freeze the display to take laps.
  .
• gating one digit at a time onto the shared 4 wire BCD data bus to the BCD to 7 segment decoder.

[3] could be handled by just about any other buffer or latch with tristate outputs, but as you are working in BCD, there are very few 4000 series to choose from that handle 4 bits at a time and even fewer that give you two such units in the same chip.   Also you probably want to avoid clocked (D type) latches in favour of strobed (transparent) latches as otherwise you'll always have to generate a clock pulse to load input data.

One thing that would be fun to build would be data input from an old mechanical phone dial.  Your circuit would need to zero a counter when the dial left its home position, and debounce and count pulses to present a BCD number to all the 4508 inputs paralleled.  There would be another decoded counter with output enable (not tristate) that would pulse the current digit position's 1/2 4508 LATCH input high to capture the dialed digit.  This digit select counter would be clocked by the dial leaving home so it advances for the next digit.  Either preload it all 1 or don't use its decoded output for 0 or you wont get anything in the first digit.   You'd also need a clear button that jams '1111' on the 4058 inputs (for blank from the 7 segment decoder), pulses all their LATCH pins to clear the display and resets the digit select counter.

Doing something similar with a 3x4 phone keypad without using a MCU would be a *lot* harder . . . .

On the subject of tricks to drive a VFD on the cheap, another common one was to run the VFD with grid on or segment anode on at +5V, and around -25V dc bias on the cathode.  Each grid or anode would be driven by a PNP transistor with its emitter to the 5V rail and a base resistor driven by the 5V logic (active low), and would also need a high value pulldown resistor down to -30V to turn it off properly.  Because the grid or anode only sink current (ignoring secondary emission, current only flows when they are being hit by negatively charged electrons), the pulldown resistors can be very high resistance - hundreds of K is fine.   Many old VCRs had this arrangement with the output transistors integrated in the same chip as the clock/timer logic.

[neo]

Sorry for my lack of foresight in only uploading eagle file.

[Ian.M]

Everything below the 4508s looks good, but you only want/need ONE 4055 with *ALL* the 4508 D0 pins connected to its I0 pin and so on for the rest of the BCD bus up to all the 4058 D3 pins connected to its I3 pin   The BCD bus on the output of the 4508s is where all the digits values must appear in sequence one at a time for the multiplexing.  You cant (easily) combine the outputs of multiple 4055 chips as they aren't open drain, dont have blanking, and cant be tri-stated.

Edit: corrected typo of 4053 for 4055

[neo]

Everything below the 4508s looks good, but you only want/need ONE 4053 with *ALL* the 4508 D0 pins connected to its I0 pin and so on for the rest of the BCD bus up to all the 4058 D3 pins connected to its I3 pin   The BCD bus on the output of the 4508s is where all the digits values must appear in sequence one at a time for the multiplexing.  You cant (easily) combine the outputs of multiple 4053 chips as they aren't open drain, dont have blanking, and cant be tri-stated.

I should of thought of that, thank you. Now the only problem is that 4508s and 4055s are not the most common chips, but at least they are available.

EDIT:
I know i have HD74HC125P laying around could i use it as a prototype or would that cause problems interfacing with CMOS?

[Ian.M]

Keep the supply voltage at 5V and you can mix 4000 series and 74HC (not HCT) with no problems.  However 74HC125 only gives you four tristate buffers, so you'd need as many of them as you have display digits + the input data has to be stable because they don't latch.   

If you just want to play around with combining 4 individual data lines into a single one, a single 74HC125 will do nicely - just remember that only ONE /EN pin should ever be low at a time.  To protect it against mistakes while experimenting put a 150R resistor in series with each Y pin to limit the current if you cause a bus conflict.

[neo]

Well monkeying around with the shift register code i was able to create an animation,  but i don't fully understand how it works and to paraphrase Eintstein. "Understanding is everything."