Update time.
I was away for a couple of weeks, but came back with various ideas about how to tackle this.
First up, I checked the voltage regulators, and they were all behaving correctly. I then traced a 1khz test tone through one of the input hybrids, and it was appearing on the differential outputs which feed the ADC. The fact the display included a trace, and there were no display distortions lead me to believe that the display processor was fine, only leaving the ADC to display connection. Unlike an ADC in a typical audio circuit, the ADCs collect the data to a circular buffer, and the display processor reads this buffer from the ADC - these scopes only have a short buffer of 4k samples if I remember, so this style works well, especially when you realise they run at 5Ghz.
So, either all 4 ADCs are dead, or the processor is failing to read the data from the ADCs, and these are connected on a dedicated bus for this purpose. My thinking was that there was a dubious solder joint on an address pin, or something like that, leading to the bus being corrupted when the processor reads, leading to weird values appearing in the display.
To test this theory, I decided to desolder the left hand 2 ADCs - the idea being that this would either change things, or change nothing (pretty scientific eh?). If it didn't change anything, i'd then at least have some easy pads to probe to try and learn more.
Anyhow, desolder two chips, clean the pads, and bingo, the scope was then working.
Next up was to clean and resolder the chips to see if there's a dodgy chip, or whether it was dodgy soldering. This went well, and all 4 channels are now working.
So then i've reassembled the instrument, run some tests, a bit of soak testing, and it seems to be back in one piece.
There's one outstanding issue, which is the ethernet is not working on this instrument, but this is a common problem caused by dubious settings in the NVRAM, so i'll hopefully work on that, but that's software niggles and nothing to do with BGAs.
The other thing i've done is to flash the firmware to a newer version to enable stuff like FFTs (it was a paid option in earlier scopes and they then bundled it with the later software). This involved trying to find 4 floppy disks, and a machine with a floppy drive. I have a USB floppy drive which came to the rescue.
Overall, success! I think i've learnt to reball BGAs, although my processor didn't work (sadly). I do feel I could tackle this if needed again, so that's another skill to now at least understand enough to not be totally scared by the prospect. Given the likelyhood that more everyday appliances will require a microscope and a heatpad to fix in future, I think i'm well prepared
I was away for a couple of weeks, but came back with various ideas about how to tackle this.
First up, I checked the voltage regulators, and they were all behaving correctly. I then traced a 1khz test tone through one of the input hybrids, and it was appearing on the differential outputs which feed the ADC. The fact the display included a trace, and there were no display distortions lead me to believe that the display processor was fine, only leaving the ADC to display connection. Unlike an ADC in a typical audio circuit, the ADCs collect the data to a circular buffer, and the display processor reads this buffer from the ADC - these scopes only have a short buffer of 4k samples if I remember, so this style works well, especially when you realise they run at 5Ghz.
So, either all 4 ADCs are dead, or the processor is failing to read the data from the ADCs, and these are connected on a dedicated bus for this purpose. My thinking was that there was a dubious solder joint on an address pin, or something like that, leading to the bus being corrupted when the processor reads, leading to weird values appearing in the display.
To test this theory, I decided to desolder the left hand 2 ADCs - the idea being that this would either change things, or change nothing (pretty scientific eh?). If it didn't change anything, i'd then at least have some easy pads to probe to try and learn more.
Anyhow, desolder two chips, clean the pads, and bingo, the scope was then working.
Next up was to clean and resolder the chips to see if there's a dodgy chip, or whether it was dodgy soldering. This went well, and all 4 channels are now working.
So then i've reassembled the instrument, run some tests, a bit of soak testing, and it seems to be back in one piece.
There's one outstanding issue, which is the ethernet is not working on this instrument, but this is a common problem caused by dubious settings in the NVRAM, so i'll hopefully work on that, but that's software niggles and nothing to do with BGAs.
The other thing i've done is to flash the firmware to a newer version to enable stuff like FFTs (it was a paid option in earlier scopes and they then bundled it with the later software). This involved trying to find 4 floppy disks, and a machine with a floppy drive. I have a USB floppy drive which came to the rescue.
Overall, success! I think i've learnt to reball BGAs, although my processor didn't work (sadly). I do feel I could tackle this if needed again, so that's another skill to now at least understand enough to not be totally scared by the prospect. Given the likelyhood that more everyday appliances will require a microscope and a heatpad to fix in future, I think i'm well prepared