If wiggling jumpers isn't helping then it is quite likely the display needs replacement. Luckily they have standard pinouts and it isn't difficult to find a replacement.

That is a good point -the jumper may have corrosion or may not be tight any more. Let me give that a go. I did notice that flicking the case really hard with my finger brought a dim light back to the faulty segment.

I had one that I received burned out. (2 or 3 non-lighting segments, 3 dim ones) First I tried reflowing the solder joints, then cleaning all the jumper pins.. didn't solve it. I just ordered a new 7-segment display for it for $3 and that solved it. I'm guessing the prior owner connected voltage directly to the segment jumper pins by mistake, or it was just old and died.

It is possible to burn out segments by lighting up only one or two of them (while setting up the numbers). There's usually only *single* current limit resistor for all the segments and it functions right only if there's many segments lit, with just one or two the lit segments will have huge current through them and are easily damaged.

Did these things ever happen to establish a common footprint for the display PCB? That is, the mounting hole(s), size and position of the board relative to the window, etc.. My Pentium case uses a harder-to-find 3-digit display that fits in the same area as the common 2-digit displays. It's definitely different than the 2-digit display PCBs that I have, but all three of those seem to be somewhat interchangeable. (Although they are probably from the same manufacturer -- the original cases are similar as well.)

The reason I ask is that I've already laid out a new PCB for the 3-digit display. The original board's controller IC didn't work, and I couldn't figure out the single-letter legend over all the control pins anyway, but the LED module itself was fine. On my board, I used two 10-bit DIP switches to set turbo and non-turbo speeds, and an ATmega164PA to read the inputs and multiplex the display -- with current limiting per-segment. This could be adapted to a 2-digit controller easily enough, and I could make the files available for anyone who wants to spin their own. No idea how many people have the appetite for dealing with SMD parts and flashing microcontrollers, but IMO setting the display in binary is a lot easier than dealing with the weird common / turbo / non-turbo jumper arrangements they tend to use.

I see two fairly standard sizes for the display box, the larger and smaller boxes. Both come in 2-segment and 2-sgement+1. I haven't checked the the PCB mounting leads though.

Anyway, moving on and off the jumpers didn't have any effect on my display. I had to flick it several times in harder and harder consecutive flicks to get that particular segment of the LCD to light up. If I then flick it one too many times, the light goes out again. So I have to start flicking it again with my finger. The first flick gets the light to just come on, the next flick, its a little brighter, and so forth, but the best I can get it to illuminate is at about 80% intensity - one more flick and the light goes out again. Anyone know what this means from a physical stand-point?

Sounds like a bad solder joint to me. You might have to do more localized flicking to find out.

I guess the LED could have failed, structurally. I've never heard of that happening though. AFAIK they're entirely encapsulated so the bond wires wouldn't have much room to move.

does that segment fail in both states? HI and LOW?
I have a display where it will only light up in 1 state but not the other, seeing if the segment behaves the same in both states could help you rule out a few things

Yes, both states it fails. Does that point to a solder joint or the display itself? Are these LED's or LCD's? I always assumed they were liquid crystal, but some of you are calling the LED, so now I'm confused.

It won't hurt to melt some fresh solder onto the join and mix it, pretty quick way to rule out a cold/cracked joint, while your at it checkout the whole pcb under a magnifying glass

I'd have to pull out my SCSI2SD, my CF, and floppy drive to access the display, also I'd need to remove all the jumper wires. It is working at 80% brightness now and is hardly noticeable, so I think I won't take further action until it goes out again. I'm pretty burned out with moving around the SCSI2SD, CF, and floppy. The mounting is rather unconventional.

yeah those are pretty annoying to take out the display In have that won't show a low state I left like is too much trouble, but one of my other displays actually comes our forwards so can extend it out without removing anything behind it, I accidentally discovered this but every case is different. I was thinking would be a good cheat if you could flip the display upside down somehow then the burned out segment would be the empty top on top of the 4

I thought of that as well, but some digit will still be affected in one of the states. Low is 20 and high is 40. In the future, it will be Low is 33 and High is 66. But I need to figure out how to retain a removable heatsink to the QFP version of the SXL2-66.

was worth a mention, have you tried a different jumper?
Once I was short 1 jumper so made my own by bending a cable and the segment with the homemade jumper always took longer to light up then the rest

If it acts the same in either L or H mode, then that would point to the segment joint. Although, since the segment is common to both speeds, if your display is like mine there's a separate jumper position for L Only, H Only, and Common. So it could be the Common jumper that's failing. Although you said you fiddled with the jumpers and it made no difference, so I'm sticking to the segment joint.

They are LED, by the way -- not LCD.