Vacuum fluorescent display to clock display

[Ian.M]

You don't *HAVE* to use an Arduino bootloader as the Arduino can be used with an ISP sketch to load your code into another AVR, but it makes development *MUCH* easier to simply flash it with the bootloader and use a *GENUINE* FTDI USB<=>Logic level serial cable or other high quality equivalent to make it appear as just another flavour of Arduino to the IDE. 

The 16MHz crystal is probably fine - get a spare ATmega328P chip, flash it with the normal Arduino bootloader and see if it works with the crystal and what caps are needed for it to get it to its nominal frequency.  (There are other ways of checking the crystal - build a pierce oscillator round it or simply use a 74HC4060, but the AVR option also gets you a useful chip and practice at moving from Arduino to direct AVR on breadboard without having to learn anything new on the software side or patch the IDE)

Incidentally, your breadboard is pretty fugly - its *EXTREMELY* limiting not having power and ground busses between the breadboard sections, and I'm sure that's not helping with the reliability of  your circuits so I strongly recommend getting some solderless breadboard bus strips (like these) and sticking a pair down to your baseplate either side of each breadboard section.  I make do with the single bus strips found on my BIMBOARD, but frequently wish I had more.  If you've got the space and the money, more is better.

[neo]

If you are too cheap to get a USB<=>logic level serial cable, once youv'e stuffed a bootloader in there, you could even use a Uno with RESET tied low and tap its RX and TX pins to the equivalent pins on the  PDIP, then pulse its RESET low manually to trigger the bootloader.

there is a difference between cheap and practically broke, ill look into the bus strips the breadboard itself is a very old one i got from the same place i got the crystal

[Ian.M]

If you are too cheap to get a USB<=>logic level serial cable, once youv'e stuffed a bootloader in there, you could even use a Uno with RESET tied low and tap its RX and TX pins to the equivalent pins on the  PDIP, then pulse its RESET low manually to trigger the bootloader.

there is a difference between cheap and practically broke

CH340G breakout boards can be found for as little as $2, or hack a really old cellphone data cable.  If its cheaper than a couple of bottles of pop, drink only water for a week to pay for it.. . . .  At this point I should note that my ancestry comes from two groups with reputations for being the tightest with money in Western Europe so min-maxing every purchasing decision is nearly instinctive for me.

However, knowing how to use an Arduino as a logic level USB UART is valuable to anyone who needs one in an emergency.   If you have a real UNO with a socketed '328P, you can also pull the chip, and then hook up the reset line to your breadboard which makes life even easier.

[neo]

my uno has a surface mount chip aka no socket

[neo]

so just get a grasp on what were talking about (just confirming what i think is going on at this point), i need the ch340g and a ATmega1284P-PU chip so i can build my own arduino? also as for the crystal im meant to figure out what i need to get it hummin at the right frequency, i have a leftover 4060 from the supplies i got from my clock

[neo]

i do apologize if im a bit dense at times, i learn by doing more so than reading, dont get me wrong i can and am learning from this it just takes a bit of effort on my part to do it thank you for your patience

[neo]

how can the 7407 output 30volts i have yet to find anything on google to tell me that

[Ian.M]

The easiest way to understand what you need to get an Arduino on a breadboard is to look at the Arduino Pro Mini schematic as that shows the required logic level serial connections on header JP1 and you can substitute any USB <=> 5V logic level serial adapter for the recommended FTDI one (including the CH340G one I mentioned)  or even use a MAX232 chip to convert the levels from a real serial port.  The absolute minimum serial hookup to support downloading sketches, is TXD, RXD + a manual reset button.  Add the DTR signal (or RTS for 3rd party USB<=>serial adapters) and the 0.1uF cap from that to Reset and AVRDUDE will reset it for you when it starts a download.

For the TXD + RXD + manual reset you can use your existing Arduino with its reset tied low to disable the surface mount  ATmega328P, as a USB<=>logic level serial adapter.  You have to hit the reset button on your breadboard when you see the end of compilation message just before AVRDUDE runs.  If you don't mind modding your  Arduino board, you could cut the link in the middle of the RESET EN solder jumper (easily reversible with a dab of solder) and attach a wire to the capacitor side of the jumper so you can connect it to the breadboard ATmega reset pin. 

Get a bit fancier and fit a 2 pin header wired to the jumper pads and mounted with a dab of epoxy and you can enable/disable the auto-reset with a simple plug in jumper and when its off, use a M-F jumper wire to get the DTR reset pulse to your breadboard.

Personally I'd hack the Arduino and start saving up for a better adapter.

Start with a 'vanilla' ATmega328P in PDIP, with a 16MHz crystal for a breadboard Arduino that runs unmodified Uno sketches, then upgrade to the ATmega1284P which requires you to patch the IDE and maybe some of the libraries to handle the extra pins.

A normal CD4060 or equivalent wont run a 16MHz crystal reliably if at all because its not fast enough even at 15V Vdd.  You'd need a 74HC4060, or 74AC4060.  For the crystal, start with a 33pF cap to ground either end of it + whatever chip you choose to make it oscillate.  If its running slow, go down to 22pF.  If its running fast, go up to 47pF.  The caps are non-critical if best frequency accuracy isn't a requirement.  You can mix and match adjacent values to get it closer or even replace one of them with a trimmer or add a gimmick cap in parallel to get it really close.  However as 30PPM error is only 2.6 seconds/day, you'll have to run your clock for a very long time making a very small tweak each day to adjust its rate by trimming the crystal.   Its much easier to simply note the daily error, calculate the actual crystal frequency from the error and trim it in software.

Read http://ww1.microchip.com/downloads/en/AppNotes/00826a.pdf for an introduction to crystals that tries to keep it relevant to digital designers.

The 7407 cant output 30V, but its quite happy switching 30V provided via an external pullup resistor >=1K to its output.  Input at logic '1' output transistor switches off and pullup takes it up to 30V.  Input at logic '0', output transistor switches on and pulls it down to about 0.7V.  As 30V is the Abs. Max. limit, you'd want to keep the HT rail at or under 27V above ground to have a little safety margin.

[neo]

now im confused you say transistor but a transistor cannot output more than its base, also im not hacking my arduino it works and i like it that way, knowing my luck if i touched it that fact would soon change

[Ian.M]

now im confused you say transistor but a transistor cannot output more than its base, also im not hacking my arduino it works and i like it that way, knowing my luck if i touched it that fact would soon change

Err. . . . I think you are confusing it with an Emitter Follower circuit where the Emitter cant go above Vb - Vbe unless the load pulls it up.  For a NPN transistor that isn't fully saturated, (i.e its in its linear region or Off) the collector voltage is almost always above the base voltage.  If its On and fully saturated at a moderate fraction of Ic_max, Vce may be a few hundred mV lower than Vbe.  Start a topic about 'transistor theory for beginners' to get the forum consensus if you are still confused as I learnt out of the 'Mullard Reference Manual of Transistor Circuits' 2nd (1961) edition, with a srew-cup breadboard and PNP Germanium transistors, a *LONG* time ago and I don't know which affordable newer books or online tutorials are any good for beginners as there are too many crap 'monkey see, monkey do' blogs, instructables and Youtube videos by posters who don't really understand transistors in any depth themselves.

I agree its not optimal to hack the Arduino.  However if you are only plugging wires in its headers, grounding its RESET, and using a manual reset button for the ATmega on the breadboard, its no more risky than any other low voltage Arduino + breadboard experiment.  The real risk comes if you take a soldering iron or Xacto knife to it., and considering the state of your Iron, I think that decision is very wise.

[neo]

ok we have been over this i know, purely transistorized version is somewhat weird but this is not an average display, every other option i cant quite understand yet this i can (i of course would use a circuit for the p channel mosfet i just connected them directly for simplicity) and with that font i came up with it is tested to work it wont make full use of the 14 segments BUT its good enough for me. Pardon me if this is considered an insult but try to see it from the way i look at it i can spend who knows how much going down the rabbit hole for a computer to drive it or i can use a very simple microcontroller, 75 cents of mosfets/transistors and stuff i have on hand to build this idea.
  The way it works is in fact somewhat simple all the npns do is act like switches the microcontroller pulses them and one grid mosfet and then the next, With the 4026 chips i bought for a second clock i have 42 outputs so 42 npn transistors, after the transistors, all the matching letters are tied together when one cluster is turned on so is its grid and all i need is 20 p channel mosfets (7 for the base letters, 8 for the grids and a couple more to support my slightly changed font) So instead of needing 50 or 60 io pins i would only need 8 in this circuit and and a uno chip could do that which is the standard chip for programming with arduino so lots of tutorials on it.
Post note i realize i didnt draw a 1k resistor pulling each diode to ground but i meant to

[neo]

the reason i bring this transistor idea back up is as it gets more and more advanced (or complex) my thoughts return more so back to this, each time slightly more sophisticated method

[Ian.M]

Can you find an old copy of Excel from before MS Office got the ribbon and went all XML and added the x on the end of all the document extensions?  If so, I'll post the font editor spreadsheet I wrote.   You could try it on the newer Excel, but I suspect it uses stuff that's been depreciated.

Your transistor idea is making a little progress but is still defective.  Apart from not showing the +HT feed to all the MOSFET sources (pin 2), it has major issues with the bipolar transistor drivers.

You are using P channel MOSFETs to drive each segment anode, but you need to pull their gates low to turn them on, and NPN emitter followers can only drive upwards to a maximum of one diode drop below their base voltage.  If you attempt to pull the bases down while the gates are charged to +HT, you'll Zener the b-e junction which does culminative damage to most types of transistor degrading the H_FE.

If you add pulldown resistors on each gate to turn on the MOSFETs, the NPN transistors could drive the gates up, but the base voltage would need to be the same as the MOSFET source voltage to actually turn them off which means you'd need a level shifter between the transistors and your [IO1] control signal.

Also the segment control signals [A]-[F] wont work the way you hope.  If they are above Vb [IO1], the transistor will draw current from them to feed the emitter, but if they are open circuit the emitter will still reach nearly the same voltage, but all the current it needs will flow through the base.   If any of them are below [IO1] by more than approx 0.6V, they will draw an uncontrolled amount of current through the forward biassed b-c junction and pull down [IO1] with a risk of damaging the chip driving it.

Furthermore, even if you could resolve that issue, so that the transistors successfully combine [IO1] and [A]-[F], you'd need to make sure that the level on [IO1] to activate the Grid MOSFET: LOW is the same as the level needed on the transistor bases to turn the segment MOSFETs on, otherwise it wouldn't gate the segments for a particular digit through to the multiplexed display the way you are hoping.

Doing logic with 'bare' transistors is difficult - that's why historically there was a push to integration leading to ICs and the 7400 series logic families, rather than simply using transistor arrays with all b, c and e pins prought out individually like CA3096.  To try to do level shifting from 5V TTL or 15V 4000 series CMOS up to 25v-30V for the VFD in the same device at the same time makes it even trickier.   You are certainly going to need a mess of resistor and probably diodes to make anything workable.   Imagine a two 'digit' two 'segment' per 'digit' VFD (i.e. two grids and two anodes) and try to figure out how to drive that with just discretes. If you can design a circuit that will generate the correct signals to drive that, extending it for 7, 14 or 16 segments and more digits is trivial.

[neo]

but you dont outwardly object and im closing in on the right idea, the only trouble is if i dont know ics barely at all, such is to say i wouldnt know what i was looking for if it was staring me in my face

[neo]

ok then forget the p channel mosfets what if i just went with the basic npns to do the switching and used buck boost instead of mosfets? something like this http://www.digikey.com/product-detail/en/diodes-incorporated/AP3012KTR-G1/AP3012KTR-G1DICT-ND/4505297

[neo]

as i understand it there is nothing wrong with my idea of npns to switch on/off signals its the voltage amplfiication process im a bit hazy on how, and one thought occurs to me regardless when it first came up they didnt have what we have available now and they pulled it, the question is how and is it applicable

[Ian.M]

You may not be able to  gate those AP3012KTR boost chips fast enough for multiplexed VFDs.  Its got a 550us soft start time, then the output cap needs to charge up to 25V and when you turn them off the cap has to discharge most of the way down to the cathode (filament) voltage. If there's any overlap with the previous digit you'll get ghosting.

I found when I reduced the mux rate below 3ms per digit, the flicker started to become objectionable.  That's an 83Hz frame rate on a 4 digit display.  For your 6 digit display you'd only get 2ms/digit.  However it would be a cool way of static driving your individual digit IV-6 VFDs.

Your NPN transistor switching, combining two logic signals is borked.  The *ONLY* chance of getting two signals to combine logically in one transistor without extra diodes etc. is to use the base and emitter as inputs.  You'll need a series base resistor and the b-e reverse breakdown voltage must be greater than the logic supply.

Truth Table
===========
 NPN    PNP
-----  -----
E B C  E B C
0 0 Z  0 0 0*
0 1 0  0 1 Z
1 0 Z  1 0 1
1 1 1* 1 1 Z

Z is Hi-Z, add a pullup resistor (NPN) or pulldown resistor to the collector to convert to a valid logic level.  States marked * are weak  the only drive current is via the base resistor.  If there is an opposing pullup/pulldown that isnt at least an order of magnitude greater, a valid logic level is unlikely.  NPN is more useful for this application than PNP as it could also do the level shifting for the MOSFETS.  There is one NAND combo, that would turn the MOSFET on with a single inverted input: /E & B if you add a pullup for C.  Put the active low segment drive signals on E and the digit drive signals on B via base resistors as mentioned above, with a separate transistor with its emitter grounded to control the grid MOSFET for that digit.  You can wire-or the collectors of the segment transistors for each digit to handle the multiplexing

However you'd be better served by using MSI logic for the multiplexing and ordering those parallel logic input 50V high side drivers we spotted a while back: Microchip MICROCHIP  MIC2981/82YN which were in stock at most major distributors.  Add a 330K pulldown to each output and it will do very well indeed for logic level to VFD electrode driving, with one of those boost chips to generate the +HT rail.   The pulldowns can be done with a single SIP resistor array if you order  Bourns 4610X-101-334LF Resistor Networks & Arrays 10pins 330Kohms Bussed, or equivalent whuch has 9 individual 330K resistors in a SIP package with a common pin at one end - ideal for breadboarding or for prototyping on matrix board or veroboard with the MIC2981.

[neo]

so with http://www.digikey.com/product-detail/en/microchip-technology/MIC2981-82YN/576-1158-ND/771627 all i have to do is input logic level and out comes 30v right (and do the thing with the resistors)? nd it all just kinda neatly wraps up like that no strings? because if what i perceive to be, is then you are a genius and you found my golden goose

[neo]

i believe there was a misunderstanding i have no intention to use the transistor as a 2 input device what i mean was to combine outputs, like below

[Ian.M]

Maybe.  I still don't think the NPNs are a good way of handling the multiplexing.   Use octal buffers with tristate outputs, one or two per digit according to the number of segments.  parallel all their same segment outputs and activate the output enable pin of ONE of them while activating the grid for the same digit.  Drive the digit selects  with a decoded counter clocked at >=480Hz (i.e. clocked by the Q5 output of a 4060 with a 32.768 KHz crystal, thats also generating your timekeeping clock), and its job done: static decoded segments for each digit => multiplexed VFD drive levels. 

You can go about it a little differently - use quad tristate buffers for each digit then build a single decoder using two 74HC238 chips or whatever the 4000 series equivalent is to get one pin out of 16 high at a time, wire those pins to the rows of a piece of stripboard, 16 rows, by 23 holes per row (min.) with a track cut in every row alternately 4 and 5 hooles from one end (zigzagged for strength) then build a diode matrix for the 'font' to directly drive the MIC2981 segment inputs.  mark out a 16x16 block in the middle of the board and it the diodes vertically cathode up, one row at a time to the long side of each row, leaving a couple of holes at the edge end of the row for off-board wiring, soldering a diode in wherever you want a lit segment, and omitting it where you want  dark one, then when all rows are complete link the cathodes with bare wires in columns and run insulated fly-wires from each column in turn to the short end of a row the far side of the track cuts for the segment outputs.   Don't be tempted to sandwich the diodes between two veroboards with tracks at right-angles: If you make a mistake its nearly impossible to get them apart again to fix it without a lot of damage.  I did that once for a 7 segment HEX display and regretted it bitterly.

[neo]

would this work for that octal buffer http://www.ti.com/lit/ds/symlink/sn74hct541.pdf though im not entirely sure what you you mean by decoded counter? perhaps a part number would help

[Ian.M]

i believe there was a misunderstanding i have no intention to use the transistor as a 2 input device what i mean was to combine outputs, like below above (reply#193)

That looks like two signals per transistor to me.  You've got an [An]signal and an [IOn] signal to each transistor then you are paralleling the emitters which cant ever work properly.

Try this:

I've attached the LTSPICE circuit in case you want to extend it or add pulsed sources ground etc. and simulate it

[Ian.M]

would this work for that octal buffer http://www.ti.com/lit/ds/symlink/sn74hct541.pdf though im not entirely sure what you you mean by decoded counter? perhaps a part number would help

Yes.  You need 6 digit select lines so you need a decade counter with decoded outputs.e.g. 74HC4017 Rig the 7th output (Q6) to reset the counter so it counts Q0 Q1 Q2 Q3 Q4 Q5.  As the '4017 Qn outputs and the MIC2981 grid drive inputs are active high and the octal tristate buffer's enable inputs are active low, you'll also need a hex inverter to generate /Q0 /Q1 /Q2 /Q3 /Q4 /Q5  for the buffer /OE gate signals.

However if you want to put the segment 'font' diode matrix ROM between the tristate buffers and the MIC2981, you'd do better to use 74HC241 which gets you two digits worth of 4 bit tristate buffers in one chip.   One is active high OE ahd the other active low, so you also need three screaming fast inverters (74AC series) to avoid overlapping the active low enable signals with the adjacent timeslot active high signals.  It would probably be safer to not use two timeslots, (Q0 and Q4), and group all the active high enables in slots Q1-Q3, and active low in /Q5-Q7 using Q8 for reset.  Then you could use ordinary inverters rather than 74AC ones.  That would also mean you might need to bump up the multiplexing frequency a bit as the frame rate @512Hz clock would only be 64Hz.

On the combined output of all the 74HC2412 chips, you'd use two 74HC238 active high 3 to 8 line decoders to build a 4 to 16 line decoder to drive the diode font matrix.  You'd probably only populate ten rows of the matrix for "0" to "9" unless you wanted to add extra letters/symbols, maybe "A" "L" "M" "P" "R" which would let you display "AM" "PM" and "ALARM"  Its worth keeping the 0xF row empty for blanking.

[neo]

so this matter is conveniently solved until i either hit a brick wall of misunderstanding or get the parts

[neo]

However if you want to put the segment 'font' diode matrix ROM between the tristate buffers and the MIC2981, you'd do better to use 74HC241 which gets you two digits worth of 4 bit tristate buffers in one chip.   One is active high OE ahd the other active low, so you also need three screaming fast inverters (74AC series) to avoid overlapping the active low enable signals with the adjacent timeslot active high signals.  It would probably be safer to not use two timeslots, (Q0 and Q4), and group all the active high enables in slots Q1-Q3, and active low in /Q5-Q7 using Q8 for reset.  Then you could use ordinary inverters rather than 74AC ones.  That would also mean you might need to bump up the multiplexing frequency a bit as the frame rate @512Hz clock would only be 64Hz.

or alternatively i could just accept that my "font" would be best on leds where it isnt already ludicrously complex