Hi all in the recent "Anyone in here interested in electronics" thread there was a discussion about the use of hot air for desoldering. Since I was about to give that a go I said I'd give an update on my experience. Here it is...
For background, I had a Western Digital 1.5TB hard drive die on me. It had 'valuable data' on it, with no backup. I wanted to try and ressurect this is possible. The error could be seen via the output of 'dmesg' (linux) when the drive was inserted via my caddy. I didn't go through the fine detail of the steps that are taken to recognise the drive, but rather little was happening. I had another similar drive, and by experimenting with different combinations of driver controller board, and drive 'body' (with platters and heads etc) decided that it was the controller board that had failed. The failure mode here seems to often be a fuse in the 5V rail, but this was not the case for me. I made a note of the exact model number of the controller board and managed to find a replacement on eBay (in France, as it happened). It cost me about £12. With this replacement board fitted to the drive body, dmesg showed that we were getting further in the process, but there was an 'IDENTITY ERROR'. I had half expected this; apparently the controller card and drive body are keyed together, via the drive firmware which is held in some flash memory. If these do not match then the drive cannot be mounted. So the problem now became, how to get the data in the replacement controller's flash to match that of the original controller. The Flash is a SPI Flash part, WINBOND 25x20, in an SOIC8 package.
I first tried to use a SOIC8 test clip to attach to the controller board and read the data. This failed - my little CH341A-based programmer dongle would not recognise the part. I could have looked at the signals, and perhaps played with the power regime, but instead I started to think about taking the flash off the old controller board and putting it on the replacement board.
I have a decent modern soldering iron, some SMD soldering skills, and a magnifying lamp. I decided that although buying some more kit might be overkill for this project, it might be useful to have in the long run. So I ended up paying something like £35 for a NOS small hot air desoldering station (Yihua 959D) on eBay
I also bought a syringe worth of flux; I already had desoldering braid and good tweezers.
I practiced using the hot air station to remove similar sized packages on an old/broken laptop motherboard. I found that although this was quite easy, it was also easy to accidentally dislodge adjacent small parts with the tweezers. Also, any nearby thermoplastic was at risk. FWIW I used a temperature of 350C, the smallest nozzle, and the airflow knob about midway.
The Flash part was very near one of the SATA connectors on this hard drive. Although this did not seem to be thermoplastic, I used caution and cut up an old aluminium takeaway carton to get some think Al foil. I fasted this down with Kapton tap so an area only a little larger that the SOIC was exposed.
The SOIC came off fairly easily - phew. I wanted to backup the data from it, and hoped to use a SOIC8 ZIF socket I bought (eBay) to allow we to do this easily. However this didn't work for some reason, even after I had cleaned the solder off the pins. I suspect the pins had shifted in vertical position as part of the pick'n'place process - some new parts I had were able to be read/written fine using the same ZIF socket.
So in order to read the contents I had to solder the flash onto a little adapter PCB I had (SOIC8 to DIL), and then connect that to my programmer. This worked fine, but meant more desoldering of the part. And the contents of the binary file looked plausible at least.
I then followed the same process to remove the flash from the replacement controller board. I plan to also read that and compare the contents, but I haven't done that yet.
I then soldered the Flash from the old controller onto the new controller board. A clean up with IPA and an inspection with my microscope ... all seemed good.
I nervously screwed the controller back onto the drive body and put it in the caddy. A bit of a wait ... SUCCESS! The drive was recognised and the partition table looked sensible.
I don't doubt that for some other (larger?) parts, using a hot air desoldering station will not be the best choice. And G*d knows this has been a bit of a roundabout exercise. But FWIW I found the process fairly straightforward and am glad to have my new tool for the future. And I have got my data back, hurrah! (yes, I will do more backing up...)
Hope this is of a bit of interest J^n