I had a new Acer Aspire V3-772G-9829. This post discusses how I configured its hardware.
The 772G came with a 5400 RPM hard drive; I ordered a replacement 7200 RPM HGST Travelstar drive. I planned to hold the original drive for a while, in case I wanted to return the Acer; eventually, I thought I might sell it with the unwanted virgin Windows 8 preloaded, to anyone who would actually want that operating system. I also ordered a Kingston Digital 120GB SSDNow V300 2.5″ solid state drive (SSD).
I didn’t order RAM at this point; I had to investigate the possibilities, and Acer’s product webpage and hard-to-find support page weren’t making that too clear. Later, when I had installed CPU-Z, I was able to take the screenshot shown here. Its output seemed to agree with the Crucial System Scanner: Crucial was steering me toward DDR3-1600 PC3-12800 RAM, CL=11, unbuffered, non-ECC, 1.35V (204-pin SODIMM). The DRAM frequency was just under 800, for a double data rate (DDR) of 1600. The timings shown are 11-11-11-28. I hadn’t needed more than 12GB even on a desktop machine running all sorts of things, so I figured I might eventually add one more 4GB stick, presently around $40. (Later, I did add this RAM. The Moo0 SystemMonitor indicated that I was now using the page file less frequently and was, after all, using more than 8GB of RAM.)
Adding more RAM did not seem urgent; the SSD had lots of space for a pagefile. There were concerns that constant reading and writing with a pagefile would shorten the SSD’s life; the reply seemed to be that, with lots of RAM, the pagefile would rarely if ever be used, and anyway SSDs were reaching a point where, for various reasons, this was less of an issue. A Lifehacker article did point out that a RAM drive would be up to 20 times faster than an SSD (70 times faster than an HDD), so if pagefiling did seem to be taking place, I could consider adding a RAM drive for the pagefile.
Before installing any hardware, I wanted to make an image of the virgin Windows 8 partitions on the original 5400 RPM drive. My way of making an image was to boot the machine from a YUMI USB thumb drive containing working copies of Acronis TrueImage Home 2011 and Macrium Reflect. To get the Acer to boot the YUMI drive, I had to change the way the system booted up. By default, it was set to UEFI, and as such it did not seem willing to recognize the YUMI drive when I hit F12 at bootup. To fix this, I kept hitting F2 at startup, so as to open the startup settings. There, I went into the Boot menu and changed the Boot Mode to Legacy. Turning off UEFI would have been problematic if I had intended to boot Windows 7 from a single 3TB drive; it did not seem to pose issues where I was booting Win7 from a separate SSD and neither drive was over 2TB.
With the Boot Mode set to Legacy, the system would respond to F12 at restart, giving me a boot menu that included my YUMI drive. By this point I had plugged in my external USB hard drive. (Sometimes I had found that an external USB drive would prevent a system from booting, if it was connected at the very start.) I told the system to boot from the YUMI drive. On that drive, I went into Acronis. The Acer touchpad was not responsive, so I used an external USB mouse. I made backups in both Acronis and Macrium. Macrium gave me this list of partitions on the virgin internal drive:
Recovery – NTFS primary – 400MB
ESP – FAT32 primary – 300MB
Unformatted primary – 128MB
Acer (drive C) – NTFS primary – 914GB
Push Button Reset – NTFS primary – 17GB
Those sizes were partition capacity. Actual contents were considerably less. Depending on compression, each image of all five partitions totaled around 26 GB. Now I was ready to install my new SSD and hard drive. I did this by following the advice in an instructional video. Basically, I had to unplug the power and battery and remove the big panel on the back of the machine. The screws did not actually come out of the panel; I had to gently pry the panel out with the loosened screws still partway inserted into it. A Kingston video that I found after the fact advised that I should have pressed the power button after removing the battery and the power cable, just to be sure there was no lingering power in the system.
As shown in a video, this Acer had two drive bays and one mSATA slot. For the one installed hard drive, I removed the two obvious but tiny screws and slid the drive out, away from where the screws had been. Now I could see that, while Crucial incorrectly reported two RAM sockets, I actually had two in each of two banks, for a total of four sockets, with a reported maximum capacity of 32GB.
To remove an existing module, I would use a fingernail to push the clips aside. Here, the photo shows the clips for Module 1 already partly pushed aside. The module would then pop up. If the clip sprang back into place, it might be necessary to start on the other side and/or to lift the module a bit. To add another module instead of removing one, I would just push it into the empty socket (shown here as Socket for Module 2) and then push it down until its clips (Clips for Module 2) held it in place. (When I did add the third module, later, I had to play around for a while before I figured out how to get it to go in all the way. It was feasible; it just wasn’t super-easy.)
Installing two hard drives meant removing the mounting bracket (also called a caddy or a tray) from the stock drive and installing it onto the SSD, and also getting a second bracket for the replacement hard drive. (The bracket is the light-colored piece of metal shown in the drive bay at the bottom left of the photo above.) I had not realized that it would be necessary to buy a mounting bracket. It was hard to find one. The part number stamped onto the existing bracket was 13N0-7NM0302. (Those are zeroes, not ohs.) A search turned up very few references to it. One user complained that it would cost $30-40 to buy this little piece of metal. I eventually found it: 25 Euros (about $34) — and that was apparently for delivery within Europe. This was not a situation calculated to inspire love of Acer. Designing a substitute would not be easy: the bracket had a little fin on each side, at the end opposite the electrical connectors, to fit into a slot in the Acer’s case; those fins and slots would support the drive largely in midair, for cooling.
It looked like a bracket for an Acer Aspire 771G might also fit. A search for one of those led to iTecForLess and its apparent alter ego Jtec, both in Canada ($19 plus shipping; micro-screws not included!). I sent them an email asking if they had a caddy for a 772G. They eventually pointed me to exactly the right item in their product list. But by that point, I had worked out my own solution. First, I decided to use the existing caddy for the replacement hard drive, since (unlike the SSD) it had exposed circuits that might come afoul of whatever paper clips or pieces of rubber I might shove into the bay to support it in lieu of a genuine mounting bracket. I swapped the bracket from the stock drive to the replacement drive, put the replacement HDD into drive bay 2, and focused my attention on the SSD and drive bay 1.
One possibility was to use tin snips to cut up a soup can and bend the resulting pieces into four appropriately sized and shaped brackets, one for each corner of the SSD. I reasoned that it might not be necessary to have the two tongues that screws would affix to the case (see photo, above) if I could instead design springy brackets that would push up, down, and sideways at the opposite end of the SSD. This sounded like a lot of work and perhaps several sliced fingers from the sharp edges of the tin, however, and anyway I figured most people do not have tin snips and therefore my solution would be unhelpful (though some kinds of ordinary scissors might also be sacrificed to the mission).
I noticed that the drive bay and the inside of the back panel came with pieces of foam and solid rubber installed. I guessed that drive temperatures probably would not be so hot as to melt those pieces. So it seemed that I might use rubber to keep the SSD in place upwards and downwards. In the interests of cooling, I wouldn’t want to use big pieces that would block a lot of drive surface or hinder airflow. The SSD came with a sort of rubber gasket; it looked to be about the right thickness to cushion between the SSD and the back panel. I removed the protective tape from that gasket (except for the part near the electrical connectors, which I didn’t want to become adhered to the gasket – although maybe the lingering protective tape would eventually become enfouled with the connectors anyway). I put the gasket in place.
Turning to the opposite side of the SSD (in this case, the side with the Kingston label), I found that I had a thick piece of double-stick foam. Not quite thick enough. I fortified it with a layer of double-stick tape. I also put some double-stick tape onto the preexisting foam cushions near the connector end of the bay. I left the protective tape in place on the top layer of double-stick foam; no need to permanently secure the SSD into the bay. This three-point contact arrangement was not ideal, but with the length of the SATA connectors, it seemed unlikely that there would be twisting forces sufficient to cause damage. So now I had my SSD cushioned from above and below, lying at the same level as the HDD in the other drive bay.
There remained the need to exert lateral pressure from the non-connector end of the SSD. There was about 1/4″ of space to fill. To solve this problem, I jammed an old shoe in there. (Kidding.) A few inches of thick electrical cable might have done the job: sturdy but still somewhat springy. A box of matches would be a particularly bad candidate for the task. I remembered that I had sometimes temporarily placed desktop internal hard drives on cardboard boxes while doing maintenance. On that basis, I decided to go with a rolled-up piece of cardboard backing from a notepad. This piece wanted to unroll itself, and was therefore at least somewhat springy. On further reflection, I saw that it covered an indentation in the left end of the drive bay, and might thus impede airflow. So I cut it into two pieces, one for each side, leaving the center open.
Unfortunately, this did not work. When I tried to reinstall the rear panel, I saw that it had a tab extending down into the open space between the left edge of the SSD and the case — exactly where my cardboard was. On the positive side, it looked like the tab came within about 1 mm of the SSD. The tab was thin but not flimsy; it had a bit of buttressing on its left side. It appeared it might suffice to hold the SSD in its socket.
I went with that hypothesis. I removed the pieces of cardboard, restored the rear panel, plugged in the battery and power cable, and turned it on to see what I had achieved. F2 at bootup told me I did indeed have Kingston and HGST drives installed. BIOS options on this machine did not seem to provide temperature information or permit temperature monitoring. F12 at reboot took me into the YUMI drive, whence I booted Ubuntu 13.04 . . . not. The Acer froze. Well, of course. I was trying to run the AMD64-specific version on an Intel CPU. For some reason, it appeared that I would have to use a 32-bit version to get an Intel-specific version of Ubuntu 13.10 desktop. I installed that ISO on the YUMI drive and tried again. Still a freeze. I tried booting Ubuntu 11.10 from DVD. A different kind of freeze, but a freeze nonetheless. That was odd — I had used that DVD many times in the past. I tried booting Knoppix from the YUMI drive. That ran. I went into System Tools > System Profiler and Benchmark. It reported only 2251MB of RAM. There were mysteries here that would perhaps be resolved as I proceeded to install Windows 7.
It would eventually develop that I had put the SSD in the wrong bay. Apparently Acer anticipated that some people would want to add SSDs; hence, they left drive bay 0 open for the SSD. In other words, looking at the photos above, it seemed that I should have put the SSD in the bay on the right, not on the left. Then again, an Amazon user’s comment seemed to say that drive bay 0 was SATA III while drive bay 1 was only SATA II. I would rather have the faster SATA III access for large data files and leave the SSD to run at SATA II.
Another thing about the SSD: updates. At Kingston’s downloads webpage, the closest I could get to finding firmware updates for my SV300S37A/120G model was their SV300S3. The FAQs at that page said I would not need special drivers and advised against defragmenting. There didn’t seem to be any special downloads for my SSD.
The next time I opened the rear panel of the Acer, about two months later, I saw that the double-stick tape was still firmly in place. The SSD was also largely still in place, though it had shifted slightly to one side. To prevent that, I found two screws with relatively flat heads, screwed them into the holes on the sides of the SSD, and adjusted them until they made lateral movement impossible.
Installing the third RAM module did seem to make a difference. Acccording to my Moo0 SystemMonitor, I now tended to be using more than 8GB of RAM, especially if I had a lot of browser tabs open, and I was also using the pagefile less frequently. I was frankly surprised to be using the pagefile at all. It seemed there might actually be a case for having even more than 12GB of RAM, though presumably that would use even more battery power when I was going portable.
Installing the SSD also made a difference. After I proceeded to install Windows 7, I found that it made a very noticeable difference was in the speed of reboots and also, it seemed, in the restoration of data from the paging file and in other program-file accesses. It did seem to have been a worthwhile investment. Later, I added a 256GB mSATA drive in the sole remaining bay, and used it partly for oft-accessed data files and partly for the paging file. It did not have an especially noticeable impact on performance; as people had advised, putting the Windows program drive (C) on SSD gave more bang for the buck. Since the mSATA drive was not the primary hard drive, it did not improve my Windows Experience Index score (Control Panel > Performance Information and Tools). To the contrary, for some reason, the Primary Hard Disk component of that index actually slipped a bit, from 7.9 (the highest possible) to 7.8.