vacuum fluorescent repair - is it possible?

[alank2]

First I have a 4x20 VFD in an enclosure and probably everyone knows that twang kind of sound you hear if you set a VFD down with a small knock.  If I lift the enclosure just on one edge perhaps 5mm to 10mm and let it drop on my desk, I'll hear it and also you can see some of the display vibrate if it is on.

I decided to test a display that is not in the enclosure to see if it would exhibit the same behavior, so I connected it to a pcb, and gave it a few small knocks.  I know these are made of glass so the knocks were very very light, more like taps really.  The one outside the enclosure began to die!  It started fading in the center, kept fading, and eventually was completely faded.  I was surprised as I would have thought these things would be more durable.

I called the manufacturer and they were super and are sending me a replacement, but he told me that he doesn't think it is a loss of vacuum based on the color of the dot in the corner of the display.  He said it would turn white if the vacuum is lost.  If it isn't a loss of vacuum, could the shock have caused an overcurrent situation on the main pcb that is part of the problem?  He mentioned the fuse, but the fuse is still closed and it still draws 300mA when the factory test screen is supposed to be on.

Any thoughts on what I might look at first for fun to see what I can figure out?  Perhaps the circuit that generates the higher voltage to drive the glass?

[McBryce]

Possibly the knocks were causing the filaments to fail? Can you see if all the filaments are still in place? They are the bits that are most likely making that twang noise.

McBryce.

[bktemp]

It sounds like a failed vacuum, but that should be easy to tell when the getter turns from silver/black to white.

If you knock hard enough, the filament can touch the grid electrodes, shorting the outputs of the grid drivers. But then the display would fail instantly, and the driver ics would probably fail also shorting the anode/grid rail to gnd.

Could you take a picture?

[alank2]

It sounds like a failed vacuum, but that should be easy to tell when the getter turns from silver/black to white.

That is what I thought, it was a progressive failure, I saw dimming, more dimming, and then eventually out.  Then again, I've seen it where part of the display goes out or appears when power is ramping up or down too.

If you knock hard enough, the filament can touch the grid electrodes, shorting the outputs of the grid drivers. But then the display would fail instantly, and the driver ics would probably fail also shorting the anode/grid rail to gnd.

It wasn't instant for sure.

Could you take a picture?

Trying to, having an issue...  give me a minute to fix.

[alank2]

[bktemp]

It looks like the display itself is fine:
The vacuum is good (getter is still silver/black) and all the filament wires are straight lines, none of them is broken or stuck to one of the grids.
That's good news.
The bad news is, all the grid and anode drivers are embedded in the display itself. So you can not measure the signals directly only their supply voltage. It should be in the range of 30V-60V (there should be some large caps at the back of the module).
Is the filament still glowing (should be clearly visible at low ambient light)? If not then either the boost converter failed, or something is shorting it (probably the anode/grid drivers).

[alank2]

Even in a dark room there is no glow at all.

It does consume around 300mA when I try the factory pattern test on it.

I can see the electronics are under the glass and it is only soldered on one side, so I might be able to lift it up assuming the adhesive pads aren't too strong...

[alank2]

Good display with test pattern enabled = 240mA, I measured the pins to the display and one of them is 37V.

Bad display with test pattern enabled = 350mA, The pins all read near the same with meter, the one that had the 37V on the good one also had 37V on the bad one...

[bktemp]

Is the filament voltage the same on both displays? The 3 pins at each end should be the filament voltage. This is either connected to the 5V supply voltage or to the boost converter without any rectification.

[alank2]

The right most pin in the middle header is the 37V pin.  All the other pins seem to be 0-5V, but I've only used a Fluke 189 on them.  The bad display might be just a little higher 37V vs 36V or so.  Do you want me to read the voltage on the right most three pins between the two?

[bktemp]

There are 3 pins on the left side, marked VFD1 and similar 3 pins on the right side. Measure the voltage between the pins on the left side and the right side in both AC and DC mode. The voltage should be around 3-5Vrms.

[Andy Watson]

Can you check the integrity of the filament circuit - as well as delivering power to heat the filaments, there should also be a DC return path such that the filaments behave as a cathode. The dimming/fading in the middle suggests that you have AC drive but no DC path.

[alank2]

measuring left 3 pins vs right 3 pins:

bad
   res   6.2 ohms
   dc   2.21 V
   ac   0.149 mV

good
   res   6.2 ohms
   dc   3.91 V
   ac   10 mV

[bktemp]

This clearly shows something is wrong with the filament driver. The filament cooled down, therefore the segments dimmed because of the reduced emission. That explains what happened, but not why.
Since it is DC, one end is probably connected to +5V and the other end to GND with a resistor in series to reduce the voltage.
You could try to reduce this resistor to bring the filament back to its nominal voltage.
But this is only a workaroundd. The actual question is what went wrong? Why does the filament draw more current? Something seems to conduct the heat away from the filament, lowering its hot resistance.

[alank2]

Left side goes to what looks like vcc.

Right side goes to a mosfet or transistor labeled "DE 711".  It has one large leg on one side and the pcb has a block of copper to distribute heat.  The right three pins are connected to this leg.  I'd estimate it switches to ground through a 7.5 large sized resistor with a 220 smaller sized resistor in parallel.  Measuring those 3 pins to ground on the bad one is 2.64V, on the good one only 1V...  I did try shorting the large pad to the 7.5 resistor and current increased from 350ma to 400ma but still no screen...

I lifted the glass off of the circuit board and have access to the top surface of the pcb now...

[bktemp]

If the filament does not glow even with the full 5V applied, the display is not repairable.
My guess is, air has leaked into the display and is now conducting the heat away from the filament. But it is strange why the getter is still silver/black. Maybe it takes some time to turn white.

I had a damaged VFD (broken glass). Since it was already broken I tried to burn the filament (just to have some fun and see the wires smoking). The typical filament voltage was 4.5V. With air inside I had to increase the voltage to more than 40V to get the filament glowing (and burning). It dissipated almost 50W.

[alank2]

If the filament does not glow even with the full 5V applied, the display is not repairable.
My guess is, air has leaked into the display and is now conducting the heat away from the filament. But it is strange why the getter is still silver/black. Maybe it takes some time to turn white.

It did fade over the course of 2-3 minutes or so, so perhaps there was a small leak somewhere.

So - one question just so I understand.  With only 9 pins other than the 3 on the left and 3 on the right, does this mean there is still considerable chip on glass circuitry inside the glass component itself?  The main pcb has 3 fine pitched ic's as well...

I had a damaged VFD (broken glass). Since it was already broken I tried to burn the filament (just to have some fun and see the wires smoking). The typical filament voltage was 4.5V. With air inside I had to increase the voltage to more than 40V to get the filament glowing (and burning). It dissipated almost 50W.

Wow!!

[bktemp]

So - one question just so I understand.  With only 9 pins other than the 3 on the left and 3 on the right, does this mean there is still considerable chip on glass circuitry inside the glass component itself?  The main pcb has 3 fine pitched ic's as well...

Yes, but is only ics, no passives:
https://www.noritake-elec.com/display/cig_driver_vfd.htm
Depending on the display it is only a simple shift register + level shifter or it can contain the complete display controller with RAM, character font and grid/anode drivers.
For example this display: https://www.mikrocontroller.net/attachment/197395/16-BT-131INK.pdf

[alank2]

That is a pretty cool way to eliminate a large number of pins...