There are plenty of resources out there for replacing an LCD, but having been through this twice during the week I thought I would provide a novice’s perspective. It’s tricky! Also you should be aware that any attempt at home repair is going to invalidate the warranty.

I managed to drop my phone onto tiles from quite a height, and caused what looked like a hairline crack under the glass. It was barely visible but over a period of a couple of weeks it started to leak liquid crystal, causing what I’ve seen referred to as an ‘ink stain’.

I bought a pack from eBay which actually included the glass [I figured I’d keep this in reserve] as well as the LCD panel and tools. There are a few videos on eBay, and this is the one that I used.

There are a couple of steps that I found particularly tricky. The first is the placing of the LCD onto the metal backing plate. There are a couple of lugs at diagonal corners which guide its seating. The LCD assembly is very flimsy, and I managed to break my first one, possibly by pressing it onto the back panel too firmly [there is an adhesive backing]. Because of the amount of flex in the plastic panel, I really can’t imagine how you could detach it from the metal plate without damaging it, so attaching it via the adhesive is basically a one way ticket.

The second step that is nerve wracking is the bending of the LCD ribbon around the screwdriver, which is recommended in the video I’ve linked to. I had kittens doing this, and actually skipped it the second time round.

When I went through the whole repair the second time and reassembled the phone, I thought I’d broken the Home button. After checking all of the connections I happened to notice that a little plug numbered 4, which you shouldn’t have to go near in the repair, had managed to work itself free. Be warned that there are a couple of metal contacts which protrude from the front assembly, and are part of the Home button mechanism [I think]. You need to be careful not to snag these.

I’m glad I tried it and managed to do it successfully – after two attempts – but it strikes me like the sort of job that could go badly wrong quite easily. If you damage anything on the main board that the various components plug into, you could have an attractive paperweight on your hands.

The phone I did the work on was well out of warranty, and I’ve recently upgraded to a 4S, so it was worth a punt.

I got an OCZ 120 gig 2 1/2 inch SSD for my Mac Pro and had a go at installing it this morning. A few comments on the choice of drive before we launch into the saga of the install itself. There are a slew of new SATA 3 drives coming onto the market at the moment. My decision to go with the SATA 2 was simple: I didn’t want to go to the expense of upgrading the controller on the Mac to accommodate SATA 3. At £160, the disk strikes me as pretty good bang per buck, though I’m sure I’ll come back to this post in a couple of year’s time and find the price laughable.

There is a great article on the Mac Performance Guide site that I took as the starting point for what I wanted to do, which was to install the disk in the spare optical drive bay. Leaving the disk dangling loose in the bay rankled with me, so I picked up a pretty cheap 2 1/2 inch to 5 1/4 inch converter [from an outfit called Lian Li].

OCZ SSD and converter mount

It was almost a complete waste of time. I took the optical bay out of the chassis and tried screwing the mounting into it, but only the front two screws would line up. I thought this would be good enough, but the daisy-chain SATA cable that comes off the DVD drive wasn’t long enough, as fixing it into the bay left the mount relatively deep, and the cable wouldn’t stretch.

SATA cable

So I left it in the bay, but unscrewed. At least this would constrain the amount of movement that the disk could be exposed to.

I restarted the machine, which picked up the drive straight away, and I created a single partition and default journalling file system on it, and then sparked up Carbon Copy Cloner. I’d given the next stage some thought, but filtered through what turned out to be a couple of deluded assumptions. My Mac Pro came with a default single 1 TB disk which I’ve managed to use 3/4 of in the space of about 5 months, with a combination of images and movie files. I figured that, aside from the root level directories that make up the OS install, I would just copy across the applications that I use most and that I’d appreciate the speed-up from the SSD with [so Aperture, Safari, and a few others]; the rest I’d just run off the original disk. I figured that if there were any library dependancies, they would have the same pathnames, and everything would be fine. In short, whether this works is a lottery. In some instances, the application will load fine, and in others you will see a pop-up saying that the application can’t open because ‘it may be damaged or incomplete’. So iPhoto [for instance] doesn’t work, while Opera happily will.

So I had to take a few bites at the selective set of files that I was looking to copy over with Carbon Copy Cloner. Some were just daft mistakes on my part, like forgetting to copy over the utilities folder underneath the applications folder.

But one of the applications that I particularly wanted to leave on the original drive was iTunes. When I tried to start this, I got a terminal error message which said, ‘You do not have enough access privileges for this operation’. The application would die about 5 seconds later. I spent a lot of time trying to figure this one out, as I’d assumed in the first instance that it was a neighbour of the library problem. I started with a full delete and re-install in the SSD, which didn’t work. I then tried various file permission changes. Again no joy. To cut a long story short, you absolutely have to have a /Users/Shared directory, as explained in this Knowledge Base article. I saw the folder when I was doing the selective copy, figured it was something like a dropbox a la ~/Public, and didn’t bother with it. That’s a couple of hours of my life that I won’t get back :).

I managed to carve out 24 gig of the original 730 gig on the pre-installed hard drive, but clearly not without problems. I think, in the general case, you are going to run into problems with this type of approach. Storage is expensive with the SSDs, but if you are trying to be selective in what you copy across [as opposed to the much simpler approach of cloning the original disk], if you don’t know what a directory does, assume that you need it!

I still have a lot of tidying up to do. Because of my earlier excitement with the Aperture vault, it’s absolutely enormous [around 100 gig]. So I need to slice and dice this to take full advantage of the SSD and, if I’m completely honest, the original reason that I bought the disk. That said, all of my ongoing imports will be on the SSD.

I’ve also left the original disk bootable. I think this makes sense for the time being as the root level OS directories don’t take that much room up. I’ll use soft symbolic links to put all of the apps that I launch off the HDD into the Applications folder [I’ve already tested this and it works fine].

Btw: boot speed. Before the install: 42 seconds; after: 41 seconds.

I’ve had the strangest problem with Aperture, and I’ve absolutely no idea what went wrong. I’ve been using it for over a year, and merrily backing up to a vault on my NAS device, as referenced files. It’s always been a bit glitchy with the ‘finding’ of the masters if I subsequently want to make a change: I would do a path update, and still have the offline master warning if I tried to relocate it. The only way that I could reliably get this to work en masse was by doing a vault update – no big deal as these were sporadic enough to coincide with backup activity anyway. I should mention that, because the NAS device is noisy and slow, I only turn it on when I have to.

Anyway, the arrival of my AppleTV around the end of January has prompted a lot of looking back at old projects, editing, and metadata slicing and dicing as a consequence – smart folders organised by Faces etc. I decided over the weekend that I’d do mass ‘consolidate master file’ update over about a year’s worth of  projects, and it crashed at some stage. It’s hard to say when as I left it running overnight to do it.

The next night I decided to to the path update and vault sync, and rather than taking about 5 minutes, the sync took about 4 hours. I decided to have a look at the NAS directory structure and the vault appears to have turned itself inside out. Having read up on this a bit, if you don’t specify a location for the masters when you load the images from your camera [or some other storage], Aperture will put them inside the vault package. I’ve duly looked and it’s almost empty. I now find myself with about 5,500 pictures in two separate directories, one where the vault is, and one in the directory above. By a process of elimination [i.e., deleting them, then trying to resync, and watching Aperture delete previews from the library. I had backed them up before!] I’ve discovered these are the masters.

The naming convention is peculiar, with the files have been renamed in numeric order in the format <unknown>_nnnn.jpeg [I have a lot of RAW and a few Tif files as well] – the angle brackets are literal.

So the long-term path to recovery is to consolidate the libraries for the period of time that Aperture has been managing my imports; export them to a directory structure reflecting year, month and event, and then creating a brand new library from scratch. From now on I’ll set the image import preferences to copy the pictures from my camera to this manually assigned directory structure – something I should have done from the start.

The masters are all there, in the exploded version of the vault. It’s just that they are mixed up with pre-Aperture days images, and with unrecognisable names. This is going go take quite a while. Especially as Aperture isn’t finding everything… Oh dear…

The irony is that part of the reason I did this in the first place – the mass consolidation – was so that Apple TV would have access to the highest quality images. This is almost certainly daft: I can’t imagine that it uses anything other the previews – dumping 20 meg raw files down the wireless network for slideshows is extremely unlikely.

One last piece of weirdness: my entire library has converted itself to referenced, right across the board.