HP Mini (110-1134CL): First Pass Evaluation

This HP Mini netbook was the oldest of three laptops in this NUCC-sponsored research project. As a netbook, it was a very limited and basic machine even when new, and that was around ten years ago. A lot has changed in the computing world since then.

Today, its 32-bit only CPU limits robot brain applications, as only the older ROS Kinetic LTS released prebuilt 32-bit binaries. Outside of robot brain applications, any modern graphical user interface is sluggish on this machine. From Chrome OS up through Windows 10 and everything in between. When running Ubuntu Mate, it actually felt worse than a Raspberry Pi running the same operating system, which came as a surprise. Both had ~1GHz CPUs and 1GB of RAM. And even though a 10-year old Atom could outperform a modern ARM CPU, the 10-year old Intel integrated graphics processor has fallen well behind a modern ARM’s graphics core.

So it appears the best position for this machine is in running command line computing or data processing tasks that work well on old low-end Intel 32-bit chips. It would be a decent contender for the type of projects that today we would think of running on a Raspberry Pi. With the caveat of weaker graphics effects, it offers the following advantages over a Raspberry Pi:

  • Intel x86 (32-bit) instruction set.
  • Higher resolution screen than the standard Raspberry Pi touchscreen.
  • Keyboard (minus the N key in this particular example)
  • Touchpad
  • Battery for portable use
  • Actual data storage device in the form of a SATA drive, not a microSD card.

It is also the only one of the three NUCC machines to have a hard wire Ethernet port. As someone who’s been burned by wireless communication issues more than once, this is a pretty significant advantage over the rest of the machines in my book.

 

HP Mini (110-1134CL): Command Line Adept

So far I’ve determined a ~10 year old netbooks lack the computing power for a modern desktop graphical user interface, even those considered lightweight by today’s standards. Was it always sluggish even in its prime? It’s a little hard to tell from here, because even though computers have undoubtedly gotten faster, our expectations have risen as well.

But there’s more to a computer’s capability than pushing pixels around, so we fall back to the next round of experiments with command line interface systems. And since we’ve already established that a solid state drive was not a great performance booster on this platform, I put the original spinning platter hard drive back in for the next round.

This time instead of Ubuntu Desktop, I installed Ubuntu server edition instead. This minimalist distribution lacks the user friendliness of a graphical user interface, but it also lacks the graphics processing workload of displaying one as well. As a result this machine is quite snappy and responsive. I found it quite usable, especially now that I’ve learned about virtual consoles and use the Alt key plus F1 through F6 to switch between up to six different sessions. Simple tasks like running Python scripts and running a basic server were done easily and quickly.

I started experimenting with Ubuntu 16, because Ubuntu did not release prebuilt installation binaries for 32-bit Ubuntu 18. However, once Ubuntu 16 server was and and running, I was able to rundo-release-upgrade to move up to Ubuntu 18. From minor tinkering I didn’t notice any significant difference between them.

Then I remembered I had played with an even more minimalist Ubuntu earlier, on an even older machine. Ubuntu 18 Snappy Core is available for 32-bit i386 processors, and it installed successfully on this laptop. Now I have one more incentive to learn how to build my own snaps to install on such a system. I just have to remember to that I can only connect to an Ubuntu Snap machine via SSH, and the list of valid keys associated with an account do not auto-update. I typically generate a SSH key every time I reset a machine, and I no longer have the keys to access my previous snappy core experiment. I ended up reinstalling snappy core to pick up the current set of SSH keys.

HP Mini (110-1134CL): Ubuntu Mate and Chrome OS Slow Even With SSD

After a ~10 year old netbook was upgraded with a solid state drive, we can now confirm the hard drive is not the only thing holding back performance. The following experiments indicate the old Atom CPU at the heart of this machine lacks the power to run any modern operating system graphical user interface.

First up at bat was Ubuntu Desktop 16.04 i386. It ran sluggishly when loading from this machine’s original spinning platter hard drive, and it was not significantly better when loading from the upgraded solid state drive. Watching the HDD activity light earlier, I thought this might be the case, but wanted to verify firsthand, which I have.

Next candidate was Ubuntu Mate, which has a 32-bit installer for 18.04. (Mainline Ubuntu stopped supporting 32-bit in 18.) Even though Ubuntu Mate advertised itself as a lighter-weight alternative to mainline Ubuntu, it was unfortunately still far from pleasant to use. But if needed, one reason to run Ubuntu Mate 18.04 is for the longer supported timeframe of Ubuntu 18. According to Ubuntu releases list, 18 is supported until April 2023.

I then tried an even more constrained operating system: install Chrome OS and make a faux Chromebook out of this thing. I had known Neverware CloudReady as a build of Chrome OS that anyone can install on an old laptop to turn it into a Chromebook. I had trouble making it work before on an old machine before, and wanted to try again.

I noticed the minimum recommended amount of RAM has increased as I remembered it was 1GB, now it is up to 2GB. But that was just a recommendation and I was able to load CloudReady on this netbook with just 1GB RAM. Once launched and running, CloudReady proved to be about as sluggish as Ubuntu Mate 18.

But that’s not the biggest problem:

CloudReady 32-bit EOL

CloudReady, originally advertised to help give old machines new life, has been forced to leave 32-bit CPUs behind. After seeing this notification I went online to find their announcement, as well as confirmation that Chrome OS v76 was about the right vintage for this to happen.

For interactive graphical desktop use, it really doesn’t get any more lightweight than Chrome OS and this machine still struggles. It looks like we need to fall back to a text-based server edition of operating system software.

Sewing Machine at CRASHspace Wearables Wednesdays

I brought a “naked” sewing machine to the February 2020 edition of Wearables Wednesdays. Wearables Wednesdays is a regularly occurring monthly meetup at CRASHspace LA, a makerspace in Culver City whose membership includes a lot of people I love to chat and hack with. But Culver City is a nontrivial drive from my usual range. So as much as I would love to frequently drop by and hang out, in reality I only visit at most once a month.

The sewing machine belongs to Emily who received it as a gift from its previous owner. That owner retired the machine due to malfunction and didn’t care to have it repaired. At our most recent Disassembly Academy, one of the teams worked through the puzzle of nondestructively removing its outer plastic enclosure. There were several very deviously hidden fasteners holding large plastic pieces in place.

Puzzling through all the interlocked mechanisms consumed most of the evening. Towards the end, Emily soldered a power cable (liberated from another appliance present at the event) to run its motor, which was the state I brought in to Wearables Wednesdays.

This event was focused on wearables, so everyone has some level of experience with a sewing machine. And it is also an audience who have experience and interest in mechanical design, so it was a perfect crowd for poking around a sewing machine’s guts.

When the outer enclosure was removed, a broken-off partial gear fell out. The rest of the gear was found to be part of the mechanism for selecting a sewing pattern. At the end of Disassembly Academy, our hypothesis for machine retirement was because of its inability to change patterns due to this broken gear.

Further exploration at CRASHspace has updated the hypothesis: there is indeed a problem in pattern selection, but probably not because of this broken gear. We can see the large mechanical cam mechanism that serves as read-only memory for patterns, and we can see the follower mechanism that can read one of several patterns encoded on that cam. However, pushing on the internal parts of the mechanism, we couldn’t get the follower to move to a different track.

New hypothesis: There is a problem in the pattern mechanism but it’s not the gear. The pattern selection knob was turned forcefully to try to push past the problem, but that force broke the little gear. It was a victim and not the root cause.

Exploratory adventures of this sewing machine will continue at some future point. In the meantime, we have a comparison reference from a friend who owns a sewing machine that predated fancy pattern features.

HP Mini (110-1134CL): SSD Upgrade

Installing Ubuntu Desktop 16.04 LTS on this ~10 year old netbook resulted in a very sluggish computing experience. Since it was not a speed demon even when new ten years ago, I doubt its performance was only held back by its old school spinner platter hard drive. As a quick experiment, I’ll perform a SSD upgrade and see how much of an performance improvement it would result.

Opening the door at the bottom revealed a memory module as expected. It appears to be a DDR2 memory module with a capacity of 1 GB. Unfortunately I don’t have any DDR2 laptop memory modules to attempt a memory upgrade, so I left it alone and continued trying to open the laptop.

HP Mini 110-1134CL remove easy stuff

There are clips around the perimeter. Trying to open them up, I find resistance at the corners due to screws hidden underneath each rubber feet.

HP Mini 110-1134CL screws hiding under feet

Removing them did not help release the top and bottom halves of the machine, so there are even more fasteners I have yet to see. Looking for what I might have missed, I found three screws with a keyboard icon inside the battery compartment.

HP Mini 110-1134CL keyboard retention screws

Removing those screws allowed the keyboard to be popped open, exposing the hard drive plus additional fasteners holding the laptop together. The focus at the moment is the SSD upgrade, so I’ll hold off on further disassembly for now.

HP Mini 110-1134CL keyboard removed

The bright piece of metal had the right shape and size for a hard drive cage, and it was also conveniently labeled with an instructional diagram. I thought that was great — it’s going to be as easy as 1, 2, 3! Except it wasn’t. For reasons I don’t understand, they neglected to mention two more screws that had to be removed before I could proceed with steps 2 and 3.

HP Mini 110-1134CL hard drive removal

Once removed, the stock hard drive was held to this cage with four standard mounting screws. From there it was straightforward to install a 2.5″ SATA SSD for further adventures in netbook computing.

Learning About Airflow From Three Years Of Dust

Three years ago I started this custom computer case project to build something tailored to run FreeNAS. The primary purpose was to get some hands-on time working with laser-cut acrylic, and I learned a lot building it. Now that I’ve decided to upgrade my home server to a different configuration, there will be too much hardware to fit in this box. I’ll start with a commodity PC tower case but I might build another custom case later. Either way, this little acrylic box will be retired.

Since the computer has been sitting in a corner unobtrusively serving up files for my home network, it has also collected three years of dust. The top layer is not particularly interesting, as they were deposited by gravity. The remainder, though, serve as indicator for airflow through the system and serves as a record of comparison against my intended airflow design for the box.

The biggest lesson for me was that convection played a much smaller role than I had expected. Most of the dust indicating flow was proportional to the size of their air channels, there’s no visible sign of convection altering the flow. The most visible example is the ring of dust near the CPU fan on my front panel. I had expected it to be slightly teardrop-shaped to reflect heat rising, but it is almost a perfect circle.

The most unexpected cluster of dust is on the auxiliary CPU power cable, running to the right side of the CPU fan alongside the USB wires. It appears most of the dust there were carried by air drawn in through the front panel gap. I hypothesize that, since it is a very narrow gap, airflow through that route is slowed and thus more likely to deposit dust on that cable bundle.

There were a few minor smudgest of dust whose origins are a mystery. Two up top near the PSU fan, and one on the bottom at the rear end of the PSU. I’m curious what they were, but their fine dust particle size implies they were not a significant factor, so I’m content to leave them as mysteries for now. Maybe they’ll make sense for me in the future once I learn more about designing for airflow. In order to preserve this information (all this dust will be disturbed and cleaned up when I disassemble this box) I shot a video for future reference:

(Cross-posted to Hackaday.io)

FreeNAS Box Decommissioned After Three Years

I’ve decided to retire my current home server running FreeNAS. It has worked well for three trouble-free years and will likely continue working for a few more. But I have enough motivations for an upgrade beyond its current capabilities.

First, I learned that FreeNAS has been making more and more use of its boot drive in its recent releases. At one point all the SATA ports on a FreeNAS box could be dedicated to storage devices, because FreeNAS itself is happy to boot from a USB flash drive, load to RAM, and run from there. Thus the boot drive is touched very little, minimizing wear on flash memory. However, FreeNAS documentation explained this has not been the case for several years. I have yet to run into any problems with the USB flash drives I’ve been using as mirrored boot volumes, but after three years of service I decided not to wait until problems crop up.

When looking at a boot drive for a modern operating system, my default choice is to use a solid state drive. SSDs were still an expensive luxury when I first started playing with FreeNAS, but they are now quite affordable and there’s been enough hardware churn for a few of my own SSDs to drop out of circulation and thus available for use. My first two Intel X25-M SSDs still report over 85% of wear life remaining. Their modest 80GB capacity is pretty cramped for modern desktop use, but quite spacious for a FreeNAS system drive. That capacity also means a lot of elbow room for flash wear-leveling.

The downside, of course, is that I need a SATA port on the motherboard to connect to my old but still functional X-25M. In order to have a X-25M as my FreeNAS system drive, I had to upgrade beyond this MSI AM1I motherboard with only two SATA ports.

Another motivation was an interest in hosting more functionality on the home server. While code with FreeBSD support can run in a jail, I needed virtual machines to run non-FreeBSD code such as ROS. When I started looking at FreeNAS, virtual machine support via bhyve was an experimental new feature. It has since grown to be a mature part of the feature set. With virtualization I can use the same physical box to host other software projects.

But a virtual machine also locks out a portion of system memory as any RAM allocated to the virtual machine is not available to the rest of FreeNAS. I have many 8GB DDR3 memory modules, but there were only two memory slots on an AM1I motherboard for 16GB. Moving to a motherboard with 4 memory slots will double the available memory to 32GB, plenty of room for playing with VMs.

With these points in mind, I powered off my homebuilt FreeNAS box built of laser-cut acrylic. The two storage drives will be moved to a commodity PC tower case. But before I take it all apart, I wanted to make note of a few observations from this computer’s three years of sitting on a shelf quietly running FreeNAS.

HP Mini (110-1134CL): Slow At Ubuntu 16 Desktop

A quick hardware orientation tour of this retired netbook found that we should be able to run ROS Kinetic Kame on this computer. Getting a simple ROS Kinetic environment running would be a baseline test to see how it might perform as a robot brain. And for that, we’ll have to erase the Windows 7 Starter Edition on this hard drive with Ubuntu 16.04 “Xenial Xerus”.

A sticker at the bottom of the machine identified the default operating system as Windows 7 Starter Edition. Since this machine predated Windows 8 mechanism for embedded licenses, we know this hardware would not have an embedded license for Windows and erasing this drive would mean the loss of a Windows license. I decided a Starter Edition license was no great loss and proceeded to install Ubuntu Desktop 16.04 LTS, i386 (32-bit CPU) edition.

Installation was successful on this netbook, but it was annoyingly slow to use. Every action required a few seconds, starting from activating the logon screen to every single interaction after that. I don’t know if running Windows 7 Starter Edition was any more responsive on this computer, but I wouldn’t have wanted to run an end-of-life OS even if it was faster.

What was the bottleneck here? Was it the CPU? Was it the RAM? Was it the hard drive? Or perhaps a combination of the above, like a lack of RAM triggering virtual memory activity that is hampered by a slow hard drive? For diagnosis I appreciated the fact this little netbook had a hard drive activity light, a feature that has been dropped from most modern machine. Judging by the lack of activity on that light, I suspect the problem is a slow CPU and upgrading the drive to a SSD would have limited benefit.

Even with this pessimistic view, I wanted to give it a try. I had a spare SATA SSD already on hand so it shouldn’t take a lot of time to test.

HP Mini (110-1134CL): Hardware Specifications

After a quick check to make sure this machine comes to life after charging, I started researching its hardware specifications. A sticker below the machine identified itself as a HP Mini 110-1134CL.

Based on reviews online for the HP Mini 110 product line, this is roughly ten years old powered by an Atom N270 processor. This is a big strike against using this computer as a robot brain running ROS, as the N270 is limited to 32-bit software. The latest longer-term support distribution of ROS “Melodic Morenia” only officially supports 64-bit Intel/AMD chips. If this machine is to run ROS, it would be limited to the previous LTS of ROS “Kinetic Kame” which is not ideal, but at least it will be supported until April 2021.

According to BIOS readout, there is 1GB of RAM installed on this computer. An access door with a memory module icon is visible on the underside of the machine, so a RAM upgrade is probably possible. But I don’t have any old memory modules on hand and I’m not inclined to spend money upgrading a ten year old computer. A single gigabyte is expected to be very limiting, but it is possible to get a very basic ROS installation running on a Raspberry Pi 3, which also only has 1GB. It might get annoying but is probably not going to be a deal breaker.

The BIOS hardware list also describes the hard drive as an old school spinning platter type, which was as expected. Thankfully the Fujitsu MJA2160BH G2. A 2.5″ 160GB 5400 RPM is at least a standard SATA drive and not the compact variant that stumped me earlier. This means I have the option to try upgrading it with one of the SATA SSD drive I already have on hand.

Now that I’m oriented, it’s time to take on the first experiment: see if it can run Ubuntu 16, the basis for ROS Kinetic Kame.

MatterHackers 3D Printing And Space Event

Even though Santa Monica is technically in the same greater LA metropolitan area as my usual cruising range, the infamous LA traffic requires a pretty significant effort for me to attend events in that area. One such event worth the effort was the “3D Printing and Space” event hosted by MatterHackers, Ultimaker, and Spaceport LA.

Like the previous MatterHackers event I attended, there is a nominal main event that is only part of the picture. Just as interesting and valuable is the time to mingle and chat with people and learn about their novel applications of 3D printing. Sometimes there is a show-and-tell area for people to bring in their projects, but it wasn’t clear from event publicity materials if there would be one at this event. I decided to traveled to Santa Monica via public transit, which meant Sawppy couldn’t come with me, which was just as well since the exhibit area was minimal and mostly occupied by items brought by members of the speaking panel.

I started off on the wrong foot by mistaking Matthew Napoli of Made in Space for someone else. Thankfully he was gracious and I learned his company built and operates the 3D printer on board the international space station. It was tremendously novel news a few years ago, and the company has continued to evolve technology and widen applications. Just for novelty’s sake I tried printing that wrench on my Monoprice Mini some time ago, with very poor results. Fortunately the Made in Space printer on board ISS is a significantly more precise printer, and Matthew Napolo brought a ground-printed counterpart for us to play with. It was, indeed, far superior to what I had printed at home. A question he had to answer several times throughout the night is whether FDM 3D printing in space still require support materials, which we use to hold melted filament up against gravity. The answer is that (1) their testing found that even though there’s no gravity, extruded filament nozzle has momentum that needs to be accounted for, and (2) Made in Space design their “production” parts to not require support material when printed either on earth or in space.

On an adjacent table were several 3D printed mounting brackets brought by Christine Gebara. Each of them had identical mounting points, but they had drastically different structural members connecting them. Their shape appeared to have been dictated by numerical evolution algorithms becoming available under several names. Autodesk calls theirs “generative design“. Learning how to best take advantage of such structures is something Christine Gebara confirmed was under active development at JPL.

Kevin Zagorski of Virgin Orbit brought something I didn’t recognize beyond the fact it had bolt patterns and fittings to connect to other things. During the discussion he explained it was part of a test rocket engine. While the auxiliary connecting pieces are either commodity parts or conventionally machined, the center somewhat tubular structure was 3D printed by a metal sintering(?) printer. 3D printing allowed them to fabricate a precise interior profile for the structure, and the carbon deposits inside a testament to the fact this piece was test-fired. He also described a development I was previously unaware of: they are using machines that has both additive and subtractive tooling. This meant they can build parts of a metal structure, move in with cutters or grinders to obtain a desired surface finish on the interior of that structure, before proceeding to build remaining parts. This allows them to get the best of both worlds: geometries that would be difficult to make by machining alone, but with interior surface finishes that would be difficult to make with 3D printing alone. Sadly he believes these machines satisfy a very narrow and demanding niche, so this capability is unlikely to propagate to consumer machines.

I didn’t know about Spaceport L.A. until this event, but I had been dimly aware of a cluster of “New Space” companies in the area. Southern California has been a hotbed of aerospace engineering for as long as that has been a field of engineering, though there have been some painful periods of transition such as severe industry downsizing at the end of the Cold War following collapse of the Soviet Union. But with SpaceX serving as the poster child for a new generation of space companies, a new community is forming and Spaceport L.A. wants to be the community hub for everyone in the area.

But even though some portray “Old Space” companies as dinosaurs doomed to extinction, in reality they are full of smart engineers who have no intention of being left behind. Representative of that was Andrew Kwas from Northrup Grumman and the entourage he brought with him. He said several times that the young Northrup Grumman engineers in his group will take the company into the future. It was fun to speak with a few of them as they had set up shop at one of the tables presenting pieces from their 3D printing test and research. One of them (I wish I remembered her name) gave me my first insight into support materials for laser sintering metal 3D printing. I thought that, since these parts were formed out of a bed of metal powder, it would not need support materials. It turns out I was wrong, and support materials are still required for mechanical hold and also for thermal dissipation. I don’t know if I’ll ever have the chance to design for laser sintering printing, but that was a valuable first lesson.

And last but not least, I got to talk to Kitty Yeung about her projects that express love of space through 3D printing. It’s a little different from the other speakers present as she’s not dealing with space flight hardware, but they are an important part of the greater community for space enthusiasm. In between esoteric space hardware, it’s great to see projects that are immediately relatable to hobbyists present.

I look forward to the next MatterHackers public event.

HP Mini (110-1134CL): Relic of the Netbook Era

After I had completed my first pass at the Toshiba Chromebook 2, I moved on to the third and final machine in this research project made possible by NUCC: the HP Mini 110-1134CL. It is an example from “netbook” era, a category of computers launched by the Asus Eee PC. Netbooks were an early attempt to build minimalist computers for basic internet activity, years before Chromebooks defined their own product category.

Like Chromebooks, the first netbooks ran a custom operating system that offered a web browser and very few other basic applications. But unlike Chromebooks, netbooks quickly abandoned custom OS and started running a special “Starter Edition” of Windows. Was it due to customer demand? Manufacturer preference? Microsoft offering carrot and stick? I have nothing useful to contribute here.

Like the other two machines, when I received this machine the battery was flat. Nothing happened when I first pushed the power switch. Fortunately it could be charged with the “type I” tip for my Targus universal laptop AC power adapter, same as the Toshiba Chromebook 2 I looked at earlier.

While it was charging I looked over its physical condition. There is a missing “N” key on the keyboard, but nothing else was outwardly wrong. Once the battery was charged, I turned it on to take a quick look at its condition. I saw what I recognized as the Windows 7 boot up sequence, but after some time staring at “Welcome…” I never got to the login screen. It threw up an error of “failed to start” without any error codes I could investigate, and suggested “rerun installation” which I don’t particularly care to do as we’ve long passed the end of life for Windows 7.

That’s a problem to be looked at later. For now, I’m glad the machine showed signs of life making it worthwhile to spend some time looking over its specifications.

Mars-Bound Rover Perseverance

NASA has just announced the winner of their Name the Rover Contest: Perseverance. The contest invites K-12 students within the United States to submit names for the Mars 2020 rover along with an essay explaining why it is the best name. Sadly, not one of them suggested naming Mars 2020 rover “Sawppy” but I’ll just have to get past that personal disappointment somehow.

I’ve already seen a few people calling it “Persie”, “Percy”, and variants thereof. The official Twitter handle is @NASAPersevere and judging by the first few tweets, this account will follow the trailblazing Phoenix PR effort and tweet from the first person. I’m doing the same for Sawppy, and @SawppyRover has followed @NASAPersevere for future updates. If all goes well, Perseverance wouldn’t even be the only Mars-bound robotic explorer: the European Space Agency has designed one of their own. ExoMars 2020 is an interesting engineering marvel in and of itself, but that’s for another blog post.

Back to the naming contest: I really like the cartoony rover they used. I’ve spent a lot of time analyzing photos, videos, and CAD models of Curiosity and Mars 2020 Perseverance when building Sawppy. I see a lot of technical accuracy represented in the illustration. From the robot arm joints to the rocker-bogie suspension components, I’m very impressed at how well it blends technical details with classic cartoon techniques for appealing character animation. I’m definitely keeping this guy in mind for my future rover work.

In the meantime I have to get on various parts of Sawppy’s online documentation and update mentions of Mars 2020 to Perseverance. I’ll definitely update greeting pages, but I probably won’t bother to update every past project log. The rule of thumb: If it has a date, people can tell it was written before the announcement, so I will probably let it slide. If it doesn’t have a date, I should probably fix it.

Toshiba Chromebook 2 (CB35-B3340): First Pass Evaluation

Taking apart this Chromebook immediately after taking apart the HP Split was a very instructive contrast in varied approaches taken when building a laptop. Even though the two devices were only 2-3 years apart, they ended up at very different destinations reflecting the philosophy of the software they were designed to run.

HP Split was built for Windows 8, together they positioned themselves to be all things to all people including ability to convert between tablet and laptop modes. And as a result the overall package is big and bulky. It tries to do everything, but it’s not especially great at any single task.

A Chromebook runs Chrome OS, which is a thin shell built around the Chrome web browser. It does not try to be anything else, and its simplicity in software was also reflected in a thin lightweight laptop built for a singular purpose and doing just that one thing well.

Windows 2-in-1 machines have evolved a lot since the first generation of devices like the HP Split, and while the penalty for compromises have been greatly reduced, they still exist in the form of weight, space, cost, or some trade off between them. In contrast, Chromebooks have remain thin and light machines in terms of software, hardware, and price tag.

If I wanted to turn this Chromebook into a robot brain computer, though, I need to break out of the Chrome OS sandbox and put it into developer mode. Unfortunately that menu is only visible on the primary display, which is broken. I could spend money on a replacement screen, but it seems wasteful just to use it to toggle the developer mode switch. A robot brain computer would not need a screen! So if I am to buy the screen, I should probably use it as a Chromebook laptop. The cost/benefit for that isn’t great, because Chrome OS may drop support for this hardware platform pretty quickly. When that happens,  security upgrades stop coming.

I will set this machine aside while I debate what to do. In the immediate future I have the third and final machine to examine in this research project.

Toshiba Chromebook 2 (CB35-B3340): Hardware Internals

While I contemplate buying a replacement screen to bring this Chromebook back to function, I removed the bottom panel as well, just for a look. Ten screws held the bottom panel in place, two of which were hidden under two rubber feet. (Top two as shown in this picture):

Toshiba Chromebook 2 CB35-B3340 bottom

Once removed, we could see the surprisingly roomy interior.

Toshiba Chromebook 2 CB35-B3340 internals

This was a relatively thin laptop for its age and screen size, so I had expected components to be packed densely. The battery, a single large module in the center, dominated the volume as expected. On either side are speakers, each with the luxury of the largest enclosures than I’ve ever seen in laptop speakers. Far larger than the volume allocated to speakers within the HP Split. I’ve read that interior volume for audio tuning is at an extreme premium, with sound designers fighting for millimeters, so it was a surprise to see this. And even with that use of volume, there are still room left unused near the corners.

Electronics occupied the area close to the hinge. They covered less than half of the available surface area and far less than half of the available volume within this laptop. Most of the main circuit board was covered by a metal shield, so I removed it to see components underneath. (In this picture, the machine was rotated 180 degrees relative to previous picture.)

Toshiba Chromebook 2 CB35-B3340 internals without EM shield

I see the CPU, RAM, and flash storage are all tightly integrated and soldered on board. No RAM or storage upgrades for this machine, which is consistent with the Chromebook ethos. About the only core component not soldered down is the commodity WiFi card, which I interpret to mean there was little to no cost savings to integrate an unit.

I had known about Chromebook’s concept of reduced complexity, but it was mostly in terms of software and maintenance. Since Chrome OS was running a Linux kernel under the hood, I expected the hardware to be just as complex as any other laptop. But apparently not this one, which I found very interesting. Now I’m curious if all Chromebooks have electronics guts simpler than equivalent full PC laptops. If I have the opportunity to take apart more Chromebooks in the future, I’ll keep an eye open to see if this is actually common across all Chromebooks or maybe the simplicity of this model is just good work on the part of the Toshiba Dynabook team which designed this Chromebook.

This was an interesting and instructive look inside the machine, time to put it back together and take stock of the current situation.

Toshiba Chromebook 2 (CB35-B3340): Replacement Screen Shopping

I have an old Chromebook that was pretty obviously retired due to a broken screen and I freed the damaged module for a closer look. I had no expectation that I could repair the display module, as there’s a visible crack. Interestingly the crack is inside the glass and not present at the outer-most surface. There’s also discoloration surrounding the crack hinting at more severe damage underneath. I could probably go online and find information on the display module used in this particular Chromebook, but getting the make and model is only a secondary objective. Before I contemplate a replacement, I wanted to first make sure I could install the replacement with low risk of damage. Hence the removal exercise to verify the lack of an impenetrable wall of glue or similar impediments.

It turned out getting to the actual module label was useful because this device was apparently sold in multiple configurations. There’s at least one variant with a minimal 1366×768 low resolution panel, and this device was an upscale version with a 1920×1080 panel. A search on Amazon marketplace found replacement new LP133WF2(SP)(A1) available for roughly $75 (*) and eBay sellers in a similar price range.

The price was the last piece of information I needed, now I need to make a decision about this project. $75 isn’t a terrible price to pay to bring a laptop computer back up and running, but it doesn’t compare very favorably to what else that money can be spent on. Even if we limit ourselves within the Chrome OS ecosystem.

New Chromebooks can be had for a little over $100 with the occasional sale, though at that price point we’re limited to 1366×768 resolution displays. I’ve seen 1920×1080 resolution Chromebooks at around $150 on sale, or roughly double the cost of a replacement panel. A new Chromebook would have access to newer developer features like Crostini that this Chromebook does not. A new Chromebook will also receive Chrome OS updates for at least five years, where support for this 2014 vintage ‘swanky’ Chromebook would end considerably sooner. Even if I put Ubuntu on this machine via Crouton, it is still dependent on Chrome OS for Linux kernel security updates.

At $75 for a new replacement panel, the economics is a tough call. I may contemplate buying salvaged panels which are available for less. (As of this writing, as low as $47.) I’ll keep thinking about this for a while. In the meantime, I want to look at the rest of this Chromebook out of curiosity.


(*) Disclosure: As an Amazon Associate I earn from qualifying purchases.

Toshiba Chromebook 2 (CB35-B3340): Screen Removal

If I wanted to turn this Toshiba Chromebook 2 into a robot brain, I need to put it into developer mode to install software outside of Chrome OS. Unfortunately the menu to put it into developer mode is only visible on the primary display which is damaged and illegible. Time to explore what it would take to bring the screen back.

The screen bezel are held in place by two screws and a lot of plastic clips. Once the screws were removed, I could go around the perimeter and pop loose all the plastic clips.

Toshiba Chromebook 2 CB35-B3340 screen bezel screws

After the bezel was removed, I noticed a small magnet held inside. This allows the base to detect if the lid is open or closed via a Hall effect sensor. If we end up going without the screen and need to spoof lid state, this is where we would place a magnet.

Toshiba Chromebook 2 CB35-B3340 screen bezel magnet

The lid had heat-set inserts that looked like they could be used to help fasten the screen in place, but no fasteners. I thought maybe someone had investigated screen replacement before me and just didn’t put the screws back, but as it turns out, screws wouldn’t have done anything.

Toshiba Chromebook 2 CB35-B3340 heat set insert but no fastener

This screen is held by four adhesive pads, one at each corner. Perhaps a different model used a different screen with metal brackets that would have been held using screws, but this screen was held by sticky pads.

Toshiba Chromebook 2 CB35-B3340 screen freed from adhesive pads

Once freed from the four pads, the final point of attachment was the electrical connector. The plastic tag is actually adhesive tape, helping to keep two sides of the connector together together. Once plastic is peeled away from metal, it was easy to unplug the connector.

Toshiba Chromebook 2 CB35-B3340 screen cable

Disconnecting that connector freed the screen, and we can take a closer look.

Toshiba Chromebook 2 (CB35-B3340): Developer Mode

The upside of a Chromebook is that the user never has to worry about hardware specifications, applications minimum requirements, or any of the typical headaches of computer ownership. It’s all handled through the Chrome browser. The downside of a Chromebook is that the user is not allowed to install traditional computer applications. Or at least, not by default. Under the hood, Chrome OS runs a Linux kernel, and it’s possible to use that as a foundation to extend computing experience outside the walled garden of Chrome OS. I started learning about Crouton, a project using chroot capability of Linux kernel to allow a variant of Ubuntu or Debian to run on Chromebooks.

Documentation for Crouton referenced the Crostini project, a way to get a Linux shell and container support without putting a Chromebook into developer mode. It sounds like a great thing to try first! But unfortunately this particular Chromebook is not supported. More specifically, this Chromebook hardware generation with the code name ‘Swanky’ does not meet the hardware virtualization support required for Crostini.

Especially frustrating is the explanation that, while the Intel spec sheet says the Celeron N2840 has the required hardware virtualization support, ‘swanky’ Chromebooks actually use a special variant of the chip without such support. I guess it saved them some money at the time? Keeping in mind the original intent of Chromebooks, it made sense to cut out virtualization support. But that decision now cuts this laptop off from Crostini.

So no Crostini for this machine, back to looking at Crouton. And the next critical step is to switch this Chromebook into Developer Mode. Holding down ESC + Refesh then pressing power, I can see the broken display is illuminated but the external monitor is not. It appears the recovery/developer mode menu is shown only on the built-in display, which I can’t read. And unlike the power wash menu earlier, the screen mirror key combination has no effect on the developer mode menu.

I searched online for a complete procedure to put this Chromebook into developer mode. Unfortunately all I found were”press control-D from recovery screen and follow menus” which isn’t helpful when I can’t read the screen!

It appears if I want to venture outside the Chrome OS sandbox, I have to look into screen replacement and the first step is investigating its removal.

Toshiba Chromebook 2 (CB35-B3340): Hardware Specifications

The old Chromebook retired due to cracked screen has been updated to latest Chrome OS, so I started searching for its technical specifications. I stumbled right out of the gate, failing to find anything on toshiba.com. Eventually I learned Toshiba sold a majority share its laptop business to Sharp and the new company (does that make it a joint venture?) is called Dynabook.

This would explain why every single product support page for the CB35-B3340 I found was under the Dynabook domain. There is not a lot of detail here, as a Chromebook is supposed to be low maintenance and they carried that concept through to reduced number of things a user has to worry about. A Chrome OS user shouldn’t ever have to worry about gigahertz or gigabytes.

But I did get some useful information implying this machine meets requirements for running robot operating system (ROS). The Intel Celeron N2840 processor is 64-bit capable. The maximum clock speed is up to a very respectable 2.58GHz. But it is constrained to run under 7.5 Watts, so it’s still an open question whether it has enough processing power in practice. Typical web browsing only need CPU power in short bursts: render a web page, then wait for the user to read the page, before rendering the next page. But robot intelligence puts a consistent high workload and if the machine needs to stay under 7.5 Watts it might have trouble sustaining maximum clock speed.

Its 4GB of memory and 16GB of flash storage meet bare minimums for even contemplating an Ubuntu installation. I’m not sure if 16GB storage is enough for the full suite of ROS nodes, but it can certainly run the subset necessary to operate a particular robot. And just like how CPU operating pattern differs between Chrome OS and ROS, the storage I/O pattern will be very different between typical Chrome OS and ROS. There’s a risk the flash storage will wear prematurely, a concern to keep in mind.

But first, we have to get this system to a point where we can install ROS, because ROS doesn’t install on bare Chrome OS.

Toshiba Chromebook 2 (CB35-B3340): Reset and Restart

I have a Chromebook with a damaged screen. Fortunately I could connect an external monitor via HDMI and mirror the primary display. It makes the machine usable, which is an improvement, but I have to command display mirroring every time there is a change in state. I have to press the magic key combination after every boot-up, every user log-in, and every user log-out. It is rather less than ideal but at least I can proceed.

The first order of business is to erase the system. The Chrome OS login screen showed the image of someone’s Google profile. I don’t know who it is, and I don’t care. This person’s data is none of my business and, since it is a Chromebook, I know all their data is still available online with Google.

Chrome OS has a “Log in as Guest” option, which allowed me to access the system settings menu. I thought this was where a system reset could be commanded, but after coming up empty handed I went online for more research and learned a Chrome OS reset is actually triggered by a particular key combination upon power-up. I was worried the reset process would be restricted to the primary monitor, fortunately I could mirror that display to initiate the reset process which someone at Google decided to call a powerwash. Cute.

Once up and running on a blank slate, the next order of business is to update the system. There’s no telling how long it has been since this machine received a security update.

Chromebook 44.0.2403.156

This model Chromebook launched in 2014. Chrome OS 44.0.2403.156 was released August 19, 2015. Since Chrome OS auto-updates itself, it appears this particular machine lasted less than two years in use, possibly far less, before its screen was damaged. This makes me feel bad for the original owner.

Chromebook 53.0.2785.154

Update was a multi-stage process, I assume due to its age. I’m not familiar with Chrome OS development history but each of these steps probably transition across large architectural changes. The first round of update only took it to 53.0.2785.154 (October 2016).

Chromebook 72.0.3626.122

Another round brought it to 72.0.3626.122 (March 2019) This update resulted in a very different looking user interface and many items were moved around.

Chromebook 79.0.3945.123

Yet another round of updates brought it to 79.0.3945.123 (Jaunary 2020) which is the latest available. This is a pleasant surprise, as I had not expected this device to still be supported but it looks OK on the Developer Information For Chrome OS Devices page. This model launched in 2014, and Google only guarantees Chrome OS support for 5 years, so this device is probably living on borrowed time.

We’ll worry about that later. Now that we are up and running with latest Chrome OS, time to start looking at technical information.

Toshiba Chromebook 2 (CB35-B3340): Cracked Screen

By the time I wrapped up investigation of the HP Split tablet/laptop convertible, the Targus type I adapter arrived. I needed it to charge the battery in the remaining two machines of my research assignment from NUCC. I set aside the HP Split and started charging the Toshiba Chromebook. Once the charge LED turned from orange to white, I turned it on and the answer to “why was this machine retired?” was immediately apparent.

Toshiba Chromebook 2 CB35-B3340 screen detail

There is a large diagonal crack across the middle of the screen. Sometimes when a screen is damaged we could still read the content around the crack, but not here. The entire screen is illegible. Turning the machine on and off a few times, I saw the content is not consistent between runs. Either Chrome OS is booting to a dynamic splash screen every time, or what’s visible just have no correlation with the intended content.

Fortunately, this Chromebook has a HDMI video output port. Plugging it into a monitor, I see a very pretty picture of a night time landscape. This is probably a background picture, but without any controls, it is merely the secondary screen. The login prompt is still on the primary display I can’t read.

Since this was my first Chromebook, I didn’t know if there was a key combination I could press to toggle from this “extended” mode to “mirrored” mode where the login screen is sent to both displays. A little bit of research implied that there was, and a few minutes of fumbling found the magic keystroke: control plus an icon that might be maximizing a window or possibly going full screen.

Chromebook keyboard closeup

I’m not sure if it’s common to all Chromebooks or specific to this model, but at the moment it doesn’t matter. I can see the ChromeOS login screen on my external monitor, time to get down to business.