Fluke 45 VFD suddenly starts ghosting

[mdijkens]

From one day to another the display of my old Fluke 45 starts to lid extra segments.
The weird thing is that after a long time waiting it gets less but whenever I press a button it's there again.
The segments that unwanted lid up are everytime the same segments; the middle dash of all primary digits and the lower-right of all secondary digits, together with AC-symbol .

I have measured all power supplies for the VFD and they are all within spec
What can I measure next to narrow down the problem?

[mdijkens]

I've updated the post above since my first measurements where wrong.
Now I've measured all power rails and they all seem good.

[coromonadalix]

are the red voltages values corrected on the schematics ?

You're 5.3vac on the filament seems high a little, maybe add an resistor to drop the voltage a little bit, and see what the results it give ..  your tp13 and tp14 are wrong  -26v is not good ...   

Are the capacitors good, not swollen, leaking etc ....   check if you have some ac voltage on top of the dc voltages,   maybe change all the capacitors in the psu would be a good start ??

[mdijkens]

Thanks for your help.

According to the service manual the fillament should be between 5VAC and 6VAC on top of -25VDC to -27VDC.
I've also visually checked all caps. nothing suspicious.

[coromonadalix]

I would not say on top, the Filament transformer center tap is shifted in voltage compared to the ground line and the 30vdc voltage need for the display driver.

There was talk here over defective or leaking display driver chip (ic) used ...

[floobydust]

The power supply voltages all look high but the DC-DC is unregulated? I'm not sure what U11 LM3578 is regulating to. You could try lower VIN on the battery side.

I think the VFD driver MCU NEC uPD75212 is the same family as the uPD7527 used in the HP/Agilent 34401a.
It's a PMOS part that does not have push-pull outputs (open-drain, can't sink current beyond high value pulldown resistor) and my theory is it can't really cutoff the VFD if leakage currents exist, due to an aging VFD or aging PMOS.

Discussion here: https://www.eevblog.com/forum/repair/hp-34401a-dmm-with-leaking-segments/msg1524538/#msg1524538

It's a bit up in the air, as some people replaced the VFD to fix the (34401a) problem.

If I was desperate, I would pull the 5.1V zener VR3 and use an adjustable TL431 to change VFD driver IC bias voltage VPRE (VEE). Or put in a higher voltage zener like 6.2V; NEC says "VPRE= VDD - 9V + 1V", with VLOAD at -30V they use a 9.1V zener. It's totally confusing. It's a pre-bias voltage for the MCU PMOS output stage.

[mdijkens]

Thanks.

It is a pity Fluke does not specify very strict voltages for these parts. My measured values are all within their specs.
My VFD looks fine but several pins are heavily corroded. I also suspect the NEC to leak voltages to adjacent pins.

I've ordered a new (2nd hand) VFD display board on eBay which I wait for first (2 weeks). If the problems persist I'll try the TL431 but that is challenging for me (as a beginner)
I was still hoping for some contact problems 

[Specmaster]

Thanks.

It is a pity Fluke does not specify very strict voltages for these parts. My measured values are all within their specs.
My VFD looks fine but several pins are heavily corroded. I also suspect the NEC to leak voltages to adjacent pins.

I've ordered a new (2nd hand) VFD display board on eBay which I wait for first (2 weeks). If the problems persist I'll try the TL431 but that is challenging for me (as a beginner)
I was still hoping for some contact problems 

When I first read your original post, my first gut feeling was that it sounds very much like a bad connection to me, possibly a ground connection on the display but your last post now does actually say that the display does indeed have some badly corroded pins which reinforces my thoughts.

I'm not familiar with the connections on these meters, is the display socketed or soldered ? If its the former, I would unplug the display and use some switch cleaner on the pins and the socket and then plug and unplug it a few times too try and wipe the contacts cleaner). If it is a soldered connection that I'd be tempted to try and remake the solder joint with a soldering iron and some new solder and see what that does for the problem. 

[mdijkens]

I have tried to resolder a few pins of the VFD but it was almost impossible.
I first carefully brushed the pins, then put on a lot of flux and leaded solder, but the solder hardly sticked and pcb and pin turned black (never seen that before).
So for the worst pins I soldered small wires between the VFD and driver chip. The respective segments/grids became somewhat brighter, but the leaking/glitching is not over. Not sure yet how chip and vfd are grounded...

[Specmaster]

The fact that you say that some segments have become brighter when you ran wires from the pins of the display driver chip, does seem to indicate an issue with poor connections somewhere.

[mdijkens]

Yep

Attached a picture after adding 1 wire.
You can see when I first tried to solder, pin 4 and pcb around it became black.
Pins that look bad are: 4-6, 9, 18, 21, 35-39
I've added wires to the worst of them

[james_s]

I would check all the vias, if something has caused corrosion you may have a corroded via.

Be very careful with those displays, I'm amazed you were able to find a spare, the VFD panels are made of unobtanium.

[Specmaster]

I would check all the vias, if something has caused corrosion you may have a corroded via.

Be very careful with those displays, I'm amazed you were able to find a spare, the VFD panels are made of unobtanium.

Tell me about it, I've after one for some time now.

[mdijkens]

I found one on eBay and there is still another one also.
Of course too expensive and no guarantee 

I measured all connections. resistance is between .2 and .4 ohm.
But I am unsure if that is stable and could cause ghosting also?

[Specmaster]

Well it certainly won't help it thats for sure. I had a similar thing but with LED's on a Black Star counter, and running an hot iron of the joints with a bit of fresh solder and the LEDSs were also socketed so I extracted them and gave the sockets a dose of contact cleaner, reassembled it and the counter worked just fine from then on.

[floobydust]

The VFD pins are Kovar and your solder makes a difference. Pb-free solder and water-soluble flux don't work well.
I think the factory uses an acid dip on the pins, so with age it might make a poor connection with the original solder.

[Specmaster]

Oh no, lead free solder, I never use that shit, good old 60/40 solder is still the best in my view, never had an issue with it yet.

[mdijkens]

The new old display arrived today; problem solved