10937p-50 chip?

[hydrolisk1792]

How the heck do you use this chip? I have close to thirty of them I pulled off of busted vfds from old snack vending machines.  I would like to use them to drive 14 segment led displays. I figured it was possible because the data sheet says it can be used for both led and vfd.  I can't get it to work at all, I've tried three chips so far. I know the chips are good because I have a test rig setup with a socket on an original working display board...

I'm under the impression that they are powered from 5 volts and I already have data being sent to sck and data via an arduino. I get nothing on the led displays and all the pinsare at 0.0 volts...  I can attach a data sheet of needed but a quick Google search will also bring it up. Any help would be really appreciated. I've been on this now for three days straight...

[daryl_uk]

Would be helpful if you could post your schematic of your circuit design for de-buging.

[hydrolisk1792]

Unfortunately I don't have a schematic for what I have on the breadboard. All I have is 5 volts hooked to vss pin, and the vdd pin at ground. Of course I also have the reset, sck and data lines hooked up to the arduino. Furthermore I have the reset line tied to the arduino via a pnp transistor like it shows in the dataset with some resistors. I have all the grid and anode control pins tied to ground via 100k resistors.

My main question is on how to use this chip with 14 or 16 segment alphanumeric displays like it says can be done in the datasheet?

[Whales]

We can't help you at all if we can't see what you're doing, there are an infinite number of things that can go wrong with any circuit.  Draw a full schematic of what you have made on a piece of paper and post a picture.

[hsn93]

im not sure but i think the arduino clock is faster than what the chip can handle?

[floobydust]

Rockwell 10937 VFD controller, I remember them in zig-zag package decades ago, but there are DIP and PLCC parts too.

Micrel bought the line and has them as MIC10937-40, 10937-50, MIC10957-40 and 10937-50. Datasheets are on the web http://www.datasheetcatalog.com/info_redirect/datasheets/400/27307_DS.pdf.shtml

[edavid]

The Micrel datasheet shows that the IC requires a 15V supply... for example +5V and -10V.  Check the supply voltages on the original board.

[hydrolisk1792]

Okay I got the chip to power up.  I'm using an at power supply at the moment using the 5 volt rail and the negative 12 volt rail going through a negative voltage regulator to bump it back down to -10 volts.

Now where I'm stuck at is to get it to drive 14 or 16 segment ked displays.  Could someone have a gander at the dataset and clarify how to hook up led instead of a vfd glass?

[hydrolisk1792]

Here is the schematic of what I have so far...

[Whales]

Your schem shows nothing but resistors instead of LEDs.  Good -- see what it does to them before going any further; don't worry about LEDs until you first get the chip itself working.

I already have data being sent to sck and data via an arduino. I get nothing on the led displays and all the pinsare at 0.0 volts

Make sure to use a scope (not a DMM) as the chip will be multiplexing those pins at high freq.

Missing from your schem is the Arduino.  (1) Have you joined grounds?  (2) Have you scoped your data and clock to make sure the chip is getting what you think you're getting?


EDIT: You have the power-on reset (POR)  pin controlled differently to what the datasheet shows.  The DS shows it being driven with 0 to 5V, you're doing it -10 to 5V.  That's outside the max spec of this part?  I presume the resistor will current limit it, but it would still be worth checking out.  Actually I might be misreading here.

Oh, and a pox on you for drawing a P-type part upside down   All the holes will fall out.  You're just trying to confuse me whilst I'm sick.

[hydrolisk1792]

Sorry about drawing the transistor upside down lol.  I did that so the pins were on with where I had it in the schematic lol.

I do have the grounds from the schematic also tied to the arduino. I didn't see a point in sawing the arduino in this schematic because that isn't where my issues lie. 

I think what I'm now trying to ask is what type of displays to use (common anode or common cathode). How to hook them to the circuit (directly, with limiting resistors, with mosfets, with mosfets on the grid and anode drivers)?  I've already assumed that I need to use common cathode because the segment drivers say anode in the datasheet...

Right now with that schematic, the chip does work.

[Whales]

I've just been trying to think how you could use the wrong display type (common anode vs common cath) on this chip and get away with it, but my addled brain isn't letting me.

Problem (1): voltage swing.  The ADx and SDx outputs look like the might be pull-up only (relying on external pull-down resistors), so if you have split rails (eg +5V, GND, -5V) then you should be able to be clever and cheat the voltage directions to light your LEDs.

Problem (2): output patterns.  This device will strobe the display, but if it does it inverted then persistance of vision will be ruined (all segments except the one you want will light and show digits).   You'd have to use external inverter chips to fix this. 

Given how digital this device is (serial interface and all) I'm surprised there's no internal option or bit to flip to change between common anode and common cathode.  The only info I found suggesting the common type is the timing diagram (figure 1 on page 5). 

The fact the datasheet does not mention the common type suggests I'm completely missing something.  Maybe it's just expected to be of one type, or maybe there's some clever way of using either common anode or cathode that I can't figure out.

Sorry hydrolisk 

[edavid]

I've just been trying to think how you could use the wrong display type (common anode vs common cath) on this chip and get away with it, but my addled brain isn't letting me.

You can use common cathode displays with either direct segment drive or non-inverting segment drivers (NPN emitter followers).
For common anode displays you would need inverting segment drivers (NPN common emitter).

For common cathode, you also need inverting digit drivers (NPN common emitter).
For common anode, you need non-inverting digit drivers (logic inverter + PNP common emitter?).

They don't bother showing LED drive on the datasheet because it's really intended for VFD drive.

[hydrolisk1792]

I've just been trying to think how you could use the wrong display type (common anode vs common cath) on this chip and get away with it, but my addled brain isn't letting me.

You can use common cathode displays with either direct segment drive or non-inverting segment drivers (NPN emitter followers).
For common anode displays you would need inverting segment drivers (NPN common emitter).

For common cathode, you also need inverting digit drivers (NPN common emitter).
For common anode, you need non-inverting digit drivers (logic inverter + PNP common emitter?).

They don't bother showing LED drive on the datasheet because it's really intended for VFD drive.

So if I'm understanding you correctly, if I need to invert the signals coming from the chip, I'd need to use something like a 74hc14?

For pull up and pull down resistors, I'd pull the anodes low and pull the cathodes up?

For the npn solution I'd hook the base of the transistors via a resistor to the segment and grid driver pins?  Thanks for the information you have given so far, it has helped me quite a bit so far.

P.s. the only spot in the dataset that show's any indication of using led is on the first page. But that is it.  I've scored the Web for a different dataset that might show the led example bit I've gotten no where. That is why I came here.

[edavid]

So if I'm understanding you correctly, if I need to invert the signals coming from the chip, I'd need to use something like a 74hc14?

That would be OK.

For pull up and pull down resistors, I'd pull the anodes low and pull the cathodes up?

I don't see why you would need them (unless driving a 74C14 or MOSFET).

For the npn solution I'd hook the base of the transistors via a resistor to the segment and grid driver pins?

Yes, but you only need base resistors for common emitter transistors.

[Ian.M]

The 10937p-50's a royal PITA to use with LEDs - you'd be far better off chucking them or it back in the parts bin and either using it with a VFD or, if you must drive starburst LEDs, using a string of 74HC595 shift registers and a handful of MOSFETs (or TPIC6C594 power shift registers) and segment limiting resistors, all run from a single positive supply, with the Arduino providing the segment pattern for each character.

However if you *MUST* use it, possibly for a LED conversion of an existing VFR display, a VFD has active high character grids and segment anodes but a starburst LED display has either active high character anodes and active low segment cathodes if its common anode, or visa-versa if its common cathode.

A 10937p-50 is only good for up to 20mA of segment or character drive, so if you are using it with starburst LEDs, it strongly favours using common cathode characters, with a NPN transistor for each cathode so it can sink the worst case 200 mA of a character with ten segments on. Each transistor needs a base resistor to limit the base current and each segment anode needs a current limiting resistor to stay under the 10937p-50's 20mA output rating.  You could use  ULN2003 or ULN2803 ICs (depending on the number of characters you need) for a reduced parts count solution with all base resistors integrated.

If you are stuck with common anode LED characters, you'll need a NPN emitter follower on each 10937p-50 ADx pin to get enough current, and you'll still need 14 or 16  inverting drivers for the segment cathodes, + resistors to limit the segment currents to whatever the display is rated for.  One could use 14 or 14 discrete NPN common emitter transistors + base resistors but the parts count is fairly ridiculous so using 2x ULN2803 or ULN2003 is preferable.  Due to the voltage levels involved a 74HC14 is useless.   In all cases,  to minimise dissipation in the segment resistors, unless you are driving blue LEDs or ones with multiple series dies per segment, the ULN2803 or ULN2003 chip's common emitter pin (GND) should go to 0V not your -10V 10937p-50 Vdd supply. 

[hydrolisk1792]

The 10937p-50's a royal PITA to use with LEDs - you'd be far better off chucking them or it back in the parts bin and either using it with a VFD or, if you must drive starburst LEDs, using a string of 74HC595 shift registers and a handful of MOSFETs (or TPIC6C594 power shift registers) and segment limiting resistors, all run from a single positive supply, with the Arduino providing the segment pattern for each character.

However if you *MUST* use it, possibly for a LED conversion of an existing VFR display, a VFD has active high character grids and segment anodes but a starburst LED display has either active high character anodes and active low segment cathodes if its common anode, or visa-versa if its common cathode.

A 10937p-50 is only good for up to 20mA of segment or character drive, so if you are using it with starburst LEDs, it strongly favours using common cathode characters, with a NPN transistor for each cathode so it can sink the worst case 200 mA of a character with ten segments on. Each transistor needs a base resistor to limit the base current and each segment anode needs a current limiting resistor to stay under the 10937p-50's 20mA output rating.  You could use  ULN2003 or ULN2803 ICs (depending on the number of characters you need) for a reduced parts count solution with all base resistors integrated.

If you are stuck with common anode LED characters, you'll need a NPN emitter follower on each 10937p-50 ADx pin to get enough current, and you'll still need 14 or 16  inverting drivers for the segment cathodes, + resistors to limit the segment currents to whatever the display is rated for.  One could use 14 or 14 discrete NPN common emitter transistors + base resistors but the parts count is fairly ridiculous so using 2x ULN2803 or ULN2003 is preferable.  Due to the voltage levels involved a 74HC14 is useless.   In all cases,  to minimise dissipation in the segment resistors, unless you are driving blue LEDs or ones with multiple series dies per segment, the ULN2803 or ULN2003 chip's common emitter pin (GND) should go to 0V not your -10V 10937p-50 Vdd supply.

Could I use td62783apg instead of the uln2003 solution because I have a lot more td62783apg than uln2003.

[Ian.M]

No.  The td62783ap is non-inverting and is high-side not low-side, so it cant replace a ULN2003 or ULN2803.  If you can tolerate its higher voltage drop, you could however use it instead of the discrete NPN emitter followers for the character anodes if driving common anode starburst displays.  You'd still need two ULN2003 or ULN2803 chips for the segment cathodes.

[hydrolisk1792]

Okay I understand now, that makes sence why you selected the uln chips.

So to sum it up, I'll put the uln chips at the grid drivers and as for the segment drivers, just limiting resistors?  Just making sure I understand what you are telling me without a schematic.  I may throw a schematic together off the original one I posted to visually see what you are saying.

I do have a lot of electronics experience and work with electronics for a living so I think I get the gist of what you are saying.

[Ian.M]

Yes I think you've got the idea.   State whether your starburst LEDs are common anode or common cathode, 14 or 16 segment, their LED Vf (measure one segment @ 10mA if you don't have official data) and post a schematic as PNG or GIF and I'm happy to check it for you.

[hydrolisk1792]

Okay. I have some common cathode displays, and I usually go by the 2.1v 10ma thing when dealing with unknown leds.

Quick question, when using these chips with the actual vfd glass, what is a good filiment voltage to use? AC or DC? Working on another one because I found that one of the glass are still good that I have.

[Ian.M]

LED Vf is highly dependent on colour  Your 2.1V 'guesstimate' will only be typical of GaAS (Gallium Arsenide)  amber, yellow and green LEDs and *SOME* GaAS red LEDs. 

However if your LEDs are over 2V Vf, due to the voltage drops in the 10937p-50 segment drivers and the ULN2003/2803 inverting cathode drivers it looks like there wont be enough headroom to run the LEDs between the +5V rail and ground and still have enough drop across the segment resistors to accurately set the current , so you'll have to return the  cathode driver to your -10V rail.  If you wanted to run them between 5V and Gnd, you'd need to use MOSFETs (with gate pulldown resistors) for the cathode drivers to avoid the high Vce_sat drop of the Darlingtons in the ULN2003/2803.

Whether you need to run a VFD filament on AC or DC depends on the filament length, and how much brightness variation you can tolerate across the display due to the voltage drop across a DC filament resuting in a different anode-cathode voltage for each digit.   Get it going on DC, and if you don't like the results, build an AC filament driver.

To find the filament voltage for an unknown VFD with no data,  if possible, measure the RMS voltage across the filament (using a scope or true RMS meter that can handle the filament drive frequency) in the equipment it came out of to give you a starting point.  Otherwise, in a dimly lit room, connect just the filament to a PSU that goes right down to 0V. Observe the VFD,  shielding it from as much external light as possible, and *SLOWLY* starting from 0V turn up the voltage till the filament has a barely perceptible dark red glow.  If you can observe the glow under normal domestic room lighting, you are almost certainly over-driving the filament.     

[hydrolisk1792]

So for the vfd question, I went ahead and measured one of the original displays in working condition.  It measured spot on 5vac with my fluke 87 true rms DMM.

[Ian.M]

So, next check that as described above with a bench supply, wind the voltage up to 5V DC (while keeping an eye on the filament to make sure it doesn't get hotter than a very dim glow) and measure the filament current, which you'll need to design a supply for it.

Also, while you can, measure the supply rails to the VFD driver,  the DC bias on the filament (check both ends), and the negative voltage feeding the grid and segment pulldown resistors,  to get an idea of the anode-cathode voltage it needs, and the extra grid bias for cutoff.   See Fig.4 Partial System Schematic in the  10937p-50 datasheet to get an idea of what voltages you are looking for.

That gives you enough data to put together something that should work first time - either a 6.3V filament transformer + a dropper resistor, or drive a MOSFET H-bridge powered from your 5V rail, with a 50% duty cycle squarewave, then capacitively couple its outputs to the ends of the filament so you can bias the filament to a suitable   negative voltage.  If you use a BJT H-bridge, you'll need a somewhat higher supply voltage to compensate for its Vce_sat voltage drops.

Either way you'll need a resistive divider across the filament to fake a center-tapped 5V transformer.   I'd start by using resistors of about 5x the filament hot resistance.  Again see fig.4 - your drive circuit replaces the CT transformer and AC source shown directly below the VFD.  The center tap of the two resistors goes to the cathode of the bias Zener Ek.

The pro-grade alternative of a Royer or similar power oscillator driving a miniature transformer to provide both a CT 5V RMS secondary and a suitable output to rectify to get the -Vdisp supply is just too much design + experimental work for a one-off to make sure the filament voltage doesn't change excessively as the load on -Vdisp varies, unless you can simply reverse engineer the oscillator + transformer circuit from the donor equipment and reuse the transformer frim it.

[hydrolisk1792]

Actually the last part of your post is exactly what I'm doing for the one off vfd I'm doing. I was given another one off a jukebox that had a busted glass and I'm going to use that transformer. My mate also gave me a schematic for that display, so I don't even need to reverse engineer the board.

As for the LED verson in the OP, I'm waiting on more displays to come in the post.

Now I'm assuming on the LED verso I don't need the negative bias for the pull down resistors, I can just put them at ground, right? Or should I keep that true to the datasheet?