VFD ghosting - help needed

[rfdes]

Hi -
I am attempting to resolve segment ghosting on a Fluke 6080A signal generator VFD display.  I purchased this item 20 yrs. ago from a govt auction.  My unit had only a few hours on it and it has mainly being sitting in the garage for most of its life with me.  In fact it only has about 20 total hours of use.

Anyway, I have discovered that the ghosting issue is due to ‘leakage’ between the anodes.  There are actually two VFD displays in this unit.  The non-ghosting VFD display shows no leakage when using an ohmmeter when measuring between the anodes.  The troubled display shows ~2MegΩ down to ~50kΩ between anodes at the worst.  Segment brightness is not an issue.  This design uses an anode voltage of 37V with a 6Vac filament and 6Vdc bias.  The design used UDN6118A drivers that I initially suspected as being the source of the leakage; however, I have removed all of the drivers and completely isolated the VFD display in order to measure the leakage.  I have also alcohol cleaned any crud around the VFD PCB pins hoping that the leakage was due to contaminants.   The getter appears OK (i.e. nice and dark).

This is my first experience with VFD displays and am wondering there are any solution my problem.  Finding another display will likely be impossible.
I would appreciate your opinions whether my display is ‘toast’ or if there is a remedy.
Thanks - Jim

[inse]

I once managed to burn away the 'short' between the characters of a VFD, but I don't remember voltage and current applied.

[floobydust]

There's many things that can cause VFD ghosting between digits or segments.
Old VFD driver IC's especially the NEC 4-bit MCU PMOS ones, the mosfets age badly and they get lazy and unable to drive elements into cutoff. This is also aggravated by an old tube which has low emission. Or a bad filament bias voltage check C35, CR15, C34, C35 for the 6.2V bias. But you have UDN6118 BJT drivers so I would not suspect the driver IC's.

You can recover emission by carefully overvolting the filament, to burn off impurities and oxides. I mentioned it here and that thread's OP had success with his Fluke 6060A.
I have a second stage VFD rejuvenation that involves burning oxides off the grids and the tube has to be removed to do that. Gotta find my notes about it, but it sure helps a lot with unevenly lit, dim segments.
If the VFD's phosphor is ages/burned, you can't do anything to rejuvenate that.

[rfdes]

floobydust:  Post 2

Thanks for the input.  As I earlier mentioned, it is not an emission or driver issue.  I have removed the UDN6118 driver parts from the board to isolate the display anodes(plates).  Using an ohmmeter, I can see a worst case leakage path 'between' the anodes of about 50k ohms (typical 200k).  I see no leakage between the grids or grids to anodes.  Only anode to anode, which I don't understand.  I thought maybe the leakage was due a dirty/contaminated PCB.  The display is still soldered to the PCB just the drivers removed, so I did a thorough cleaning of the board with alcohol, particularly between the display mounting pins, but no luck.

I am puzzled as to this failure mechanism and was hoping someone would understand this better than I.
Take care -
Jim

[floobydust]

Sorry- I didn't really get what you're measuring. It's difficult for aging to create a resistance in a vacuum. I'm tending towards thinking it's not the tube because I've never seen or heard of that problem.
I can't find the Fluke 6080A service manual showing the A1 Display PCA schematic. UDN6118 is basically an emitter-follower output with 125k pulldown, and you say you have pulled... all these 5 IC's? I would guess there's some other parts in-circuit like a SIP resistor network (Z1-3?) giving you the ohmmeter readings.
I can only speculate ion migration on the phosphor between segments for this weak conduction but I've never seen it. I thought phosphor was non-conductive.

The VFD pins are Kovar and sometimes flux leaves a crusty buildup on the display, maybe look around the glass as well for contamination.

[james_s]

My bet would be on some kind of external contamination. Give the board and display a good wash and let it dry thoroughly, then check for leakage.

[rfdes]

Yes, I removed all 5 UDN6118 to isolate the display.  I, too, thought about other components or external contamination.  I cleaned the PCB using alcohol at the VFD pins as I could see some crustiness as you mentioned but I saw no difference in my readings.  If there is an external leakage path, I will need to remove the display to confirm.

I appreciate your comments.
Jim

[james_s]

I suppose you could connect a relatively high voltage between two anodes and see what happens. There should be a hard vacuum inside the tube, so if it flashes over anywhere it ought to be external. Obviously this needs to be done with a bare display, not connected to the drivers. You might also be able to work around it by adding external pulldown resistors to the anodes.

[rfdes]

I think i will try to remove the display from the PCB first and re-measure.  Not sure if I will have difficulty removing the display without damaging the PCB.

[floobydust]

Usually the PC board holes are much larger than the VFD pins, to make factory assembly easier. If you have a good desoldering gun or vac pump it should be OK.
One trick I do is mount the board vertical, and on the VFD side, heat the pin (at the board) with a soldering iron- while desoldering (suction) from the other side.
It's a two handed operation because you do need a lot of heat on the pin but desoldering (heat) only on one side of the board takes a long time for the heat to flow to the top layer and melt the solder through. So I also apply heat with a soldering iron on the top side.
If it's not working on the pin, stop and apply fresh solder and retry. Don't wiggle the soldering tip against the pad or it will tear. I also use a poker to make sure pin is loose afterwards and then move on. Careful not to bash or drop the display.

You only have to isolate a few pins to see if the leakage resistance is in the display or the board.

I do remember reading about a VFD that was designed poorly, the internal segment bus traces were too close to each other for the voltages and problems were some kind of coupling between them so the display was redesigned with better spacings. But I'd think that was a capacitance problem inter-segment.

[james_s]

Desoldering it should be very easy as long as you have a vacuum desoldering gun. I wouldn't attempt it with wick though.

[rfdes]

I removed the VFD display and it went smoothly.  Performed a thorough cleaning of the display, particularly between the pins of the anodes and verified that there is indeed leakage between the anode pins.  I've attached a photo of the display and the location of the anode connections. 

Does anyone believe it would be possible to rectify the leakage using excess current between the pins?
Very puzzling.
Jim

[james_s]

That's fascinating, I've never seen that happen before. Maybe it would be worth contacting Noritake and asking about that phenomenon? I think I would be inclined to try pulldown resistors on the anodes, I think you said the leakage was a couple megs, so 470k-1M resistors might do it.

[rfdes]

I cannot see where pull downs are going to benefit me with this issue.  The UDN6118 drivers are essentially common emitter bipolar drivers with 125k resistor pull downs when OFF. 

[james_s]

Well in that case I don't know, I assumed the anodes were floating up due to the leakage.

[wraper]

Did not try this but for similar problem for CRT suggestion was to connect a capacitor charged to high voltage to the shorting pins. As leakage is relatively small, I'm not sure there will be enough current to burn contamination though, unless charged to 1000+ volts.