The IBM Thinkpad X41 and X41 tablet are very well packaged laptops. The X41 tablet was the first tablet among the thinkpads I think. It was released in 2005. After a few revisions, current offerings like the X201 remain pretty similarly packaged, with the exception of the exotic X300/X301 which was ultimately a failure (see this Businessweek article http://www.businessweek.com/magazine/content/08_08/b4072042350389.htm)
The biggest problem with the x41t for me was the hard drive. It really is a bottleneck for everything. It was packaged with an early adoption of a 1.8″ form factor PATA drive standard which didn’t pan out. Some hardly attractive options have appeared since for SSD replacement. I was considering replacing my x41, even though it is still in great condition, until I came across another option:
I decided to try this out, to avoid getting a brand new X201. With a new (“new”) third party battery, maybe I can get a few more years out of this X41, with the advantage of the peace of mind and performance of an SSD.
So PIMP ME UP!
Figured it was worth the risk, so I got an Intel X18-M generation 2 or whatever from amazon. Came very quickly. From Walther’s page it looked like it would be pretty easy. I decided to post the results to share with the English community, and to verify that the method works and that there is not much signal degradation from the SATA wiring.
First, take out the screws labeled for the keyboard I guess, and take the keyboard out carefully. Get the x41 maintenance manual, and remove the screws necessary for removing the top plate/cover. Remove the hard drive cover, hard drive, and hard drive cage. Here’s what it looks like:
Really nice packaging. Things are alot tinier than I thought, from Walter’s pictures! Holy crap (Scheisse)! Even with the smallest soldering iron pencil tip, things will be difficult. Make sure you have the right tools! You need a decent temperature controlled soldering iron/station (clean tips and in good condition), and a small hot air rework station to remove the PATA-SATA Marvell bridge chip. Also the PCB and components are dusty and oxidized a little, and a way to carefully apply micro-dollops of no-wash flux is necessary maybe. For wiring, AWG 30 or smaller hook-up wire.
Anyways, the X18-M doesn’t quite fit in the bay as-is, as described by Walter. He used a dremel to remove raised parts of the (magnesium?) alloy case. I did this with pliers, to prevent having to remove the PCB. I put a piece of paper to cover the PCB to shield from debris, but the case is brittle and pieces break off very cleanly, so I have to recommend that (clean and fast).
Leaving the PATA header connector creates a nice way to keep the drive anchored. Gluing a piece of foam underneath for feedback pushes the drive against the header. Overall, it fits very nicely, barely clearing a mounting hole for the top cover on the left. Also, plastic tabs on the hard drive cover piece have to be snapped off with pliers to accomodate. Subsequently looks like this:
Anticipating the soldering work, inspect the areas by the Marvell chip and the PATA header carefully. Its tiny stuff, you will definitely need an inspection scope or magnifier of some kind, if not while soldering then at least to inspect the traces afterwords. I took images with a cell phone camera through a 50mm fl magnifier:
The region by the Marvell chip. Soldering pads of the (QFN?) package are tiny, < 1mm pitch. For the signals B+ and B-, the target should be the coupling capacitors to make things easier. Its interesting that AC coupling was used here, doesn’t it require that the data transmitted is DC balanced? Anyways for A+/A-, Walter taps in somewhere else. I followed the YouTube video and tapped at the Marvell chip solder pads.
To remove the Marvell chip, you probably need to use a hot air rework station. I did this with the PCB still attached to the alloy chassis. This makes it harder to heat up the area quickly. Walter said 400-500 deg C at lowest air flow. I started low (usually solder pastes flow at 225 deg C or so). This part was unnerving, after a few minutes, the chip comes off at 400 deg setting. No other components really got hot enough to flow, just keep the nozzle close to the chip. The chip should be yanked as soon as possible I guess, in case parts start to drop off the underside of the PCB. After it was removed, some clean up was required to un-short some of the solder pads, by reheating and running a sharp point through the melted solder, they will attach to the pads by surface tension etc. Most of these traces are for the PATA 32 data lines, and are unconnected since we removed the drive. Once the chip is gone, theres really no going back!
Here, the PATA connector is shown. This will be tapped for 3.3V power and GND. Having a patient/impatient girlfriend added a very interesting time component to this challenge, luckily Walter’s page decodes the pin ordering and makes that aspect trivial. Print out his stuff ahead of time to maximize time! You should also print out the X18-m datasheet, etc. The Marvell IC datasheet couldn’t be found via google, so I really trusted Walter’s site for this.
Started making connections at the drive end. I couldn’t waste time cutting out a micro SATA cable as shown at the other sources (Also i didn’t have a micro SATA cable). So I soldered directly to the gold pins of the X18-M drive. Here you really have to be careful of strain. Ultimately, a hot glue gun really does wonders here (I was skeptical after seeing pictures). If the modification is not mechanically robust, then there is really no point. Especially since the strength of the SSD is vibration resistance.
To make connections, I guess I recommend adding a small stub between a stripped SATA cable and any solder points (I used wire wrap kynar wire for flexibility). On the other end of the SATA cable (red, white), I added stubs (blue) which were first soldered to the points near the Marvell chip. Heat shrink tubing applied (black). Not my finest work, but oh well (see below). The guy in the YouTube video did not use a SATA cable, and the differential impedance was not well defined. This may be OK over the few inches of distance needed to reach from the PCB to the drive. Also I didn’t see the SATA cable fragment grounded in Walter’s images. Probably not that important, I left it floating.
Because of the size, the connections are very fragile. I see the feasibility of this SSD modification resulting from two things: the relatively few connections needed (just 6 pins) and the liberal application of hot glue / glue gun to encase all the tenuous connections in solid plastic. I’m more used to 5-minute mercapthenol epoxy stuff, so I didn’t like the hot glue idea. However, it is AWESOMES!! (to the max). Hot glue (thermoplastic adhesive) comes in a variety of recipes with all sorts of ingredients. Ultimately, it measures unreadable resistance with the multimeter, so it is good to go:
Above is the glue gun. $10 at your local crafts store or Jo-ann’s.
Hot glue all over the crappy soldering job. Saved! (~Solid as a rock~) Oh baby…
Hot glue at the drive end saves a crappy soldering job again. Zupah!
One more time. Hyper-hyper!
Cover fits, everything is solid. Give it a good shake once or twice to make sure! Needs to survive on the lap in front of the TV!
Miraculously, the computer recognizes the drive. Change the BIOS setting to get past the warning screen about the non-IBM drive. After restoring a drive image with a Norton Ghost boot CD, Windows starts. Time to take it out on the autobahn!
HDTune Pro (free trial) shows SSD with TRIM support. Sweeeet!
>100 MB/s read! (Write test doesn’t work for whatever reason)
Random access performance.
Anyways, I hope this helps the English community with this SSD upgrade. Post your experiences!