After three evenings with a Fluke 867B

[shakalnokturn]

"New" sad meter turned up a few days ago, sold as "dead - battery leaked". I was told that there was no display by the seller.

I didn't bother turning it on after peeping at the state of the PCB! Removed and dumped ROMs, removed buzzer  and battery springs then took it for a good wash, then spent some time scratching away at corroded tracks, tinning, continuity testing.

When I thought I was done, reassembled, gave it fresh alkaline power and... No joy... nothing more than the backlight.

Started probing around (luckily there's a schematic for this thing) power good, U24's oscillator is running, PCLK and RIC_CLK outputs are there, CPU reset not stuck low.
Nothing on the LCD connector, nothing on video RAM, not much happening on U13 either on the whole, here reset is stuck low.

CPU handles reset for the video IC, maybe the CPU isn't able to load the firmware?
Probed around some more and found an odd looking U11-O5 data line, that turned out to be a cut track I missed on the not so badly corroded side of the buzzer.

Now it displays, the function switch is very fussy, resistance reading gives 40-60 Ohms with inputs shorted, no corrosion about U30 so why...?
Having had similar unexplained issues on other Flukes I decided to reflow all the TQFP's (U30 should have been enough), reassembled and tested again. Now all switch positions bring up the "LOGIC" test... It's driving me nuts by now...

SWVOLTAGE line is almost shorted to ground (1 Ohm), peeping at the CPU pins I end up spotting a solder blob under the CPU, not a blob caused by reflowing the TQFP but one caused by tinning the tracks on the other side, some solder has crept through the vias and created a mess under the CPU. I gave it a hard try to remove it from there and ended up admitting that the CPU would have to be removed. Bugger!

After cleaning-up and before fitting the CPU back I covered the hidden surface with a layer of kapton tape just in case I had to rework the tracks on the other side once more.

So finally I'm getting spot-on resistance readings and although I haven't been through checking other function calibrations it's beginning to look good.
There's still one oddity but maybe that's just a feature of this GMM: When the meter is busy there are small noise bars in the display, they don't appear when the display is still though. Is anyone else seeing this on their Fluke 86x? (Edit: As seen in last photo)

[mahi]

Excellent job on the repair!

There's still one oddity but maybe that's just a feature of this GMM: When the meter is busy there are small noise bars in the display, they don't appear when the display is still though. Is anyone else seeing this on their Fluke 86x? (Edit: As seen in last photo)

Yes, that is normal. Mine does it as well. When the screen contrast is high you'll see the noise most pronounced in the light areas as in your picture. When the contrast is low it shifts towards the dark areas. You can't get rid of it, though.

[shakalnokturn]

Thanks mahi for confirming it's a "normal" operation.
I'll try probing around at the LCD connector to figure out what causes this.
I can only imagine it is caused by variations on the Vee line, maybe due to sharing a common ground track with some of the digital IC's.

[Wytnucls]

I don't see much noise on mine:

[shakalnokturn]

Possible different PCB revision?
I have a "867-3001 Rev. M" PCB.

I don't notice it on a stable reading without graph, get a little noise while it updates the graph, heavier noise while it is auto-ranging, atrocities while pushing some keys in ohms mode.
The most obvious is in ohms when pushing the save/print key and the screen refreshes with a dark background.

[mahi]

The noise on light background is easier to see when looking down at the display (but before the image inverts) and the dark noise when looking up. Setting the backlight to maximum brightness also makes it more visible. As said by shakalnokturn, pressing the Save/Print button makes it very visible for a brief moment when the display inverts.

I have "867-3001 REV J" (made in USA) and "867-3001 REV L" (made in Holland) and both show the same noise.

[shakalnokturn]

I did plenty of probing on the LCD FFC on both ends and was unable to spot a difference there between with / without displayed noise.

The two ground lines (3&6) are linked to the buried ground layer just after the FFC socket.
On the LCD side they're independent 3 being the frame ground and 6 the bus ground.

Could the Hitachi LMG7135PNFL LCD panel cause this by itself?

[harrimansat]

It seems a "bug" in firmaware, switching from text to graph page. May be isa problem due different response speed in 867 and 867b. It seems thant isn´t a bad o defective LCD. The efect is only visible when there is a moving graphic section in LCD.

https://www.eevblog.com/forum/testgear/fluke-867b-lcd-issue/msg3241296/#msg3241296

[shakalnokturn]

Yes, I came to that conclusion too in the end, blamed it on the gate array...
As not everyone seems to have the problem they may have fixed it on later models. I'd be curious to know the gate array revisions with/without the bug.