Dave Platt wrote:

[...]


The fact that swapping the filament pins didn't affect the problem,
makes me suspect that there's a fault inside the tube. Possibly
there's an internal short between one end of the filament, and ground.

I haven't been able to measure any short between filament and the rest of
the internals (grid, anodes). I've also spent some time checking for shorts
between grid-grid, grid-anodes and even anodes-anodes. Since characters
are generally made correctly, I didn't expect any short, but I was out
of ideas.


So, take a look at the voltages and waveforms on the filament drive
pins. Ideally, they should be exactly mirror-opposites of one
another, with respect to ground. I suspect you'll find that they are
not.

That's the first thing I checked actually. The voltages are not symmetrical
respect to the ground, so I swapped the filament pins expecting to have
an opposite behaviour, but, the filament supply didn't change (it did not
follow the pin swapping) and the brightness situation of all the characters
didn't change.
The filament waveform is a square wave with rather low duty cycle (and
this also probably explains why they got away with this non symmetrical supply
in the first place), the DC component is about -11V and it is present on both
filament pins, and the square pulses go to about -21V and are applied on one
side of the filament. It is obtained with a NE555 based circuit, the duty
cycle is probably ok as I measured an RMS value of 6V on the filament and
this particular VFD wants 5.8V nominally.
I can't still exclude a rather obscure (at least to me) problem in the
filament supply, but components are very few and they are all ok.
This instrument is rather large with more than 100 ICs scattered around, so
I'm not going to reverse engineer a schematic out of it in my lifetime.
All the tests in my opinion points to lazy phosphors or low emission from
the filament in the last digits.
If it was the asymmetrical filament supply, the dim digits would follow
the pin swap of the supply.
Maybe the hypothesys that the last digits were the ones most used isn't that
bad after all.
VFD are really nice looking displays, but I must remember to avoid old
instruments with those
Frank