Downside on the 8845A/8846A is the VFD display, which quickly fades, and "burns-in". But if you buy the multimeter new, it is likely not a problem, another story with equipment powered on 24/7 in a production line or something...
The other forum member who said Fluke wanted $400 or so to replace the display was not pleasing to me (You can buy the 34461A display online for $60 or so I think I saw), but perhaps the $400 is for the entire display and front panel pcb with input buttons, etc. I looked through some of the Noritake stuff I picked up when they ran a 50% off sale a year or two ago and I think I actually might have the same glass on one of the VFD modules GU256x64D-7000, so if it dies, I've got a VFD glass I can try down the road.
Just for fun I hooked up the logic analyzer to it to see if I could discern the signal. These are chip in glass "CIG", but they are not full high level driver in glass. The CIG is meant to just lower the pin count so it is really some shift registers and grid drivers. Just enough to reduce pin count, but not more than that. Noritake has a helpful document on this here:
https://www.noritake-elec.com/display/cig_driver_vfd.htmBut, they actually go further than that - they combined the grid and data spi into a single spi. The first 256 bits are data, and the final 64 bits are grid.
From GU256X64D
pin name description
6 HV 60v
7 gnd
8 gnd
9 vcc 3.3v
10 sig1 4.57k, positive pulse 16us blanking
11 sig2 4.57k, positive pulse 1us latch
12 sig3 2.25k, square wave
13 gnd
14 nc data out?, 800ns wide pulses (1.25MHz clock)
15 nc data out?, 500ns wide pulses (2MHz clock)
16 sig4 data in
17 sig5 clock
18 sigA data or clock ?
19 sigA
20 sigA
21 sigA
22 nc data out?
An entire frame is 14.21 ms so refresh rate is 70.3729. I wonder if Fluke may be running it slower in their implementation of driving it because out of the corner of my eye I can see it updating sometimes.
So, I exported the data out of the Saleae and wrote a quick C program that converts the captured values to an image to see if I could do it. It was a bit perplexing as the order of the bits changes between columns. It has 64 four bit columns. So a byte is two rows tall. First it is 7654 with 3210 under it, and then it is 4567 with 0123 under it. See the Noritake doc, probably has to do with overlapping grid scan.
The bottom line though is that with only a few pins and a microcontroller, one could convert the 256x64 bitmap display output to practically anything. OLED, LCD, maybe even composite for precision meter meets old school TV output... Might be a fun project down the road for 8845/8846's with bad displays.
The PNG is from a GU256X64D test screen: