I want to build a cyberdeck (part 1)
background – what is a cyberdeck
Like all mediocre blogs, I’m going to start with my life story. Not for no real reason or just because I want to share but because I want to help communicate my state of mind with this so that my decisions might make a little bit more sense. There are hundreds of ways to achieve what I’m setting out to do but I chose this way because it is what I wanted and almost definitely not because it is actually practical.
So what even is a cyberdeck? A cyberdeck is a term originally coined in the mid ’80s by author William Gibson (though the concept itself may originate elsewhere, idk, I wasn’t there) and it is largely just a small (compared to ’80s tech), portable (mostly), and powerful (enough) computer built for a specific task. If you’ve played the game Cyberpunk 2077, you’ve probably used one yourself or at least seen these in use. While an active part of most play styles, the player characters’ cyberdecks never actually appear rendered in-game (just an item to equip in your inventory), one does still appear in-game. In most fiction that has devices like this, the primary interface is virtual and thus the devices do not need explicit screens. Reality isn’t fiction, however, and the software just isn’t there outside of VR games (and while I do have a programming background, I ain’t writing a whole UI for this project), so we’ll be keeping the screen.
I don’t know when I actually started this project – it’s probably been floating around in my head since I read some of William Gibson’s books in the late ’00s. Even with the idea floating around in my head, I didn’t actually start working on this until I started typing up this blog page. So with that in mind, let’s start with a seemingly non-sequitor tangent. I like e-books. I prefer reading books off a tablet specifically. I’ve tried e-readers (like the kindle) but the main draw (e-ink screen) never really appealed to me so I stuck with more fully featured tablets instead so I can easily read comics/manga on the same device. It works for me and I like it and the alternative is not reading at this point so I’ve come to terms with it.
Over the last 11 years, I have purchased four different tablets for this purpose. All have done exactly what I wanted up until the internal batteries degraded too much. And because of who I am as a person, I kept most of these.
hardware donors (and more background)
So most cyberdecks appear to be constructed with Raspberry Pi hardware, portable screens (from 3” to 12” typically though it’s HDMI so you can use whatever you want really) but I figure I’m making a portable computer, why not start with one? If I use one of the older tablets I already have, not only will it be less expensive, I’ll be reusing stuff that’s effectively e-waste already, and I figure I have a leg up on the competition since I don’t have to worry about software support, picking a screen, I/O, power management, etc. So it sounds like the older tablets I bought over the years for reading would be perfect for this.
The first option (and my first suitable tablet) was a HP Stream 8. The alternatives at the time in my pricerange and desired specs were a second-hand first gen iPad Mini (before they came with retina displays) or a secondhand Nexus 7. Other devices weren’t on my radar for whatever reason. On paper, the first gen iPad Mini looked like garbage compared to other two. Lower resolution screen, weaker looking specs, and so on. Plus, it would have had to be used to fit into my budget which also meant no real warranty. The Nexus 7 (2013) looked A LOT better than both alternatives on paper but I had already (briefly) had a Nexus 7 (2012) and just really didn’t like the Android 4 experience on a tablet. Again, to fit into my price-range, it would have had to be second hand and it would have had no real warranty. Looking back, I probably should have went with this anyway. Instead, I walked over to the Microsoft Store in the mall and just bought the HP Stream 8 brand new for a bit over $100 after tax. At the time, this seemed like the best value but I soon learned that it didn’t really what I wanted all that well. It was slow in 2014 (and still is today) but it did a remarkable job as an e-reader and is still technically supported in regards to software and security updates (at least until the EOL of Windows 10 in 2025) but it’s way too underspecced these days to bother fixing up. It’s still fully functional, including the original battery, even though I’ve damaged the LCD itself (just the backlight somehow – there are spots right in the middle) and the battery only lasts about an hour. I’d want to replace the LCD and engineer a new charging and power management system because the battery will need to be replaced and this tablet has limited IO (charge port is the data port). Between the screen being far less than ideal and the general slowness of the device, I am not picking this tablet for this project.
The second option is the device I replaced the HP Stream 8 with. I bought a Xiaomi MiPad 2 (the Windows version with 64gb eMMC, not the Android version with 16gb eMMC) in 2016 ish. It seemed well reviewed at the time and the price was right. Most importantly, it fixed two of the biggest issues I had with the Stream 8, the general slowness of the tablet and the screen. It was still slow, don’t get me wrong, but it was so much faster than the Stream 8. The screen though. Oh my god the screen. This device was an eye opener for me in terms of how good a tablet screen can look. This device appears to use the same panel as the Retina iPad Mini. Same size and resolution though perhaps a different ribbon cable added during manufacture. The Stream 8 wasn’t bad but this was just so much better. It was at this time that I started reading comics/manga too instead of just e-books. Unfortunately, the MiPad 2 has exactly zero IO on the device EXCEPT for the USB 2.0 Type C port (and a headphone jack). Yes, USB 2.0 via a Type C port. At the time, I didn’t understand the implications of this, I was just happy that things were using Type C now. I tried loading comics on the device and quickly learned that I’d be constantly shuffling data on and off this device and I would have to anticipate what I wanted to read in advance since there was no expandable storage with which to store my media collection. Not an unsurmountable task but definitely tedious. 64gb was all you got and most of that was taken by the Windows install. About 13 months after purchasing, the battery capacity fell off a cliff. It went from at least two hours (that’s how long I’d use it in one sitting) to less than 30 minutes so I made the decision to just sell it for parts and move on.
Well. That was a journey all by itself. I couldn’t restore it to factory with the recovery USB I had made when I got it because the device could not use USB devices and charge simultaneously. I didn’t want to sell it for parts since I knew everything worked aside from the battery but more importantly, I wanted to ensure my data was properly wiped before I parted with it. At the time, I chalked this up to me not having the proper dongles but I learned in 2020 that this was just how the device works. So I bought a parts MiPad 2 for the battery on ebay. The seller had the android version of the tablet and had managed to hard brick the device. As the seller described, it would turn on to a black screen and then sit that way until the battery died and it wasn’t even detected over USB (so no ADB/fastboot options either). While not exactly ideal for a “fixer upper”, it did tell me that the battery did at least work. With seemingly no way to fix this, I just popped the rear covers off both devices and swapped the main PCB which put my 64gb windows model into the housing of the broken but otherwise pristine 16gb android version. From there, I was able to restore the newly rebuilt device and I decided to try my hand at selling the thing on ebay while it still had value and just use my HP Stream 8 in the mean time while I looked for a suitable replacement device. The Stream 8 still got at least 2 hours on a charge so it was still serviceable for my purposes, even if less than ideal. Well, long story short, the tablet got damaged in shipping and ebay refunded the buyer and I was now out the costs of repairing it the first time, the fees ebay decided to keep, and the costs to repair it again (but I did still have the original chassis of the device, complete with working screen). In shipping, the screen developed a nasty flicker that made it really distracting to use. I swapped the main board back to the original chassis and then began the process of lifting the adhesive to transfer the battery over (original chassis had already had battery recycled). Good news was that this did indeed work and I got a working MiPad2 back! I decided to forego selling it at this point and just keep using it instead of the Stream 8 as the tradeoffs were much more manageable. I got like three months out of the battery before it swelled up and popped the screen out (lucky for the original buyer because it sounds like they dodged a bullet).
At this point, I decided to try out the aftermarket for this device and I bought a well reviewed third party battery. The MiPad2 was not a brand new device when I got it and it had already been a few years since then. Xioami had already discontinued the MiPad2 and moved on to the MiPad3 and was probably nearly done with the MiPad4 so I figured that any OEM batteries would have been years old already with degraded capacity (if they worked at all). The value had already fallen off a cliff and while it still mostly worked for what I wanted, I figured it was worthwhile to just extend the lifespan of the device and wait for a better replacement next year or the following. I was really happy with the performance of Windows tablets at the time and so that’s all I was looking at and when Intel retired the Atom series of CPUs, smaller Windows tablet markets totally stagnated. Bigger stuff like Surface Pros and other 2in1s were all thriving but the 8” tablet market was effectively dead or using the same hardware from years before. As it turns out, the rumors were highly exaggerated, and Intel did continue making low power CPUs, just not Atoms. Even still, it did take a few years for the low power Celerons (J series chips) to start appearing in 8” tablets. So I bought the battery. It worked about as you’d expect but the fuel gauge is built into the main board of the device instead of the battery so that required recalibration. It took about four battery cycles before the low power automatic shut downs would occur when the battery was actually low instead of after 20 minutes. The tablet worked great for another 10 months until this battery popped the screen out of the chassis (again). Exasperated and already ready to give up, I ripped the cells out and put the tablet away (with the parts tablet too) and started the search for a new device. The HP Stream 8 was back off standby again (this time with even less battery life!)
Third option is the device I purchased following the retirement of the MiPad2. I’d argue that my experience with the Xiaomi device was less than ideal and I hoped it was an exception. That being said, the anec dotal evidence I already had said HP made long lasting tablets and Xiaomi makes disposable ones. Seemed solid. I had already been looking at an HP Pro Tablet 608 G1 and it seemed to combine everything I liked about my previous two tablets, plus some extra features that I typically only bother with in higher end devices. The new HP tablet has the same great screen as the MiPad2 (but not flickering!), the same Intel Atom x5-z8500 but with double the RAM again (4gb vs the 2gb in the MiPad2 or 1gb in the HP Stream 8), expandable storage, and proper USB C support. This was, by far, the best device yet. Bonus, it was a surplus business machine which usually means that it’s built to higher standards or at least more repairable. I figured after the experience with batteries, both OEM and aftermarket, on my previous devices, this would come in handy later. Well, unfortunately, parts for this model are basically non-existent. I didn’t think to check in advance. I believe I got the longest active use out of this device before the battery finally popped the back housing off though. It kinda surprised me because the Stream 8 never expanded, it just lost capacity. All three batteries in the MiPad2 tried popping the screen off after the capacity fell off a cliff (though I did catch the first battery before it expanded enough to initiate self-disassembly). The battery in the 608 G1 didn’t seem to lose a significant amount of capacity. Further investigation lead me to discover that the MiPad2 (and frankly, every other device I had that popped batteries in storage) was using a LiHV battery with a higher than typical nominal voltage. Standard Lithium-Ion batteries (like the one in the Stream 8) are nominal 3.7v and have a max charge of 4.2v. LiHV is 3.8v or 3.85v nominal with a max charge voltage of 4.35-4.4v. It does make sense though. Battery tech has stagnated and the easiest way to get more capacity out of Li-Ion is to just charge it to a higher voltage. There is supposed to be an additive to the electrolyte solution that allows the battery to be safely charged to the higher voltage but clearly the tradeoff is that the cells have an even shorter effective life. This is no good for what I want. More on this later.
Just for continuity sake, the fourth device and the one I purchased to replace the 608 G1, I am now using a 6th gen iPad Mini. I regret not getting one of these sooner. The lack of expandable storage (and the lack of a headphone jack) is a real bummer but I haven’t actually found myself missing these features. Everything else just totally blows away the previous devices again though. I am seriously impressed. I figure this time around, if I stick with a more mainstream device, I won’t be so shit out of luck when it comes time to replace the battery. If nothing else, Apple will still offer service on this device long after any other manufacturer would (though I anticipate I’d just DIY). This device is not in consideration for this project but my experience with it has been so positive that I’m considering buying another one (probably an older model though and maybe an Air or Pro) for this. I’m still using this and I intend to keep using it indefinitely.
Last option, and the device that really got me focused on this project is an Acer Switch Alpha 12. Unlike the previous devices, this device was not purchased for media consumption. I bought this device to use as my primary computer and it worked great for years. I ended up giving the device to my family to use after I replaced it with a newer model but I was always impressed with the performance in that form factor. As it turns out, a regular laptop with an actual hinge and not a kickstand works a lot better for me and that’s the only real reason I replaced this device. Nonetheless, it worked great until a few months ago when the batteries initiated self-disassembly (notice a trend?). Compared to the other options for this project, this device is a monster between the size, performance, specs, I/O, screen, and cooling. Best of all, I can power the device completely from the DC jack, no internal battery required. The device does support USB PD charging but for some reason cannot be powered only via USB PD. Regardless, this means I have a lot fewer hurdles to cross and I’ll ultimately have a more functional device (albeit somewhat larger – more on this later too). Best of all, if we bypass the CPU whitelist and then manually install the Intel OEM video driver instead of the Acer OEM video driver, it’s fully compatible with Windows 11 (with all the bells and whistles turned on, including memory integrity and core isolation – features that don’t even work on some of my officially “supported” devices). Something will probably break in the future but this device works great with most linux distros out of the box too (unlike the previously mentioned devices) so I’m not too concerned. Biggest con to using this device is that the battery life sucks, even under ideal circumstances (only 34Wh from the factory). However, since I have to rebuild the battery anyway, I can double or even triple the capacity to fix this issue.
hardware diy (and plans to actually do this)
So I think I’m decided upon the Switch Alpha 12 for the base of this. Should I fail to get it working or accidentally destroy it with my modifications, the next runner up will be the 608 G1 or, if I cannot get either working for whatever reason, some beat up older model iPad. I’ve decided to just recycle the other tablets at this point. Either way I go, there will be some decisions I have to make regarding the chassis I build and input devices (like keyboard and mouse) that will be relatively static regardless of the device I choose. On that note, the wider Switch Alpha 12 should not be an issue for the chassis since it’s already at the minimum width I intend to make. I want to support a regular size keyboard so the 8” tablet runners up would have ended up with giant bezels or a massive offset on one side anyway. Due to the full size keyboard I want to implement and the massive 12” screen on the donor tablet, this device is going to end up looking like a chunky laptop.
I anticipate the bottom half of the chassis part being one of the easier parts since I have experience designing and building custom enclosures and custom keyboards so I’m going to leave the finer details for this until it’s all that’s left. That being said, there’s still a lot of work to be done, especially around the top half of the chassis and the hinge. My primary focus over the next few weeks/months/years will be to: * Investigate internal power. If I can restore internal cells to the tablet (relocated though to bottom of chassis for counterweight), that’s ideal since the tablet already has perfectly working battery management and power isolation circuits. I have a few ideas on this – more later * Investigate external power. Even if it’s impractical, I’ll end up with extra space in the base of the unit. So why not more batteries? Failing an internal replacement I’m happy with, I anticipate being able to just use a USB PD battery bank hot wired to the DC jack (19.5v vs 20v, shouldn’t be an issue). I’d need a battery bank that can charge and power simultaneously though and that probably means it will be the single most expensive part of this project. And I want two of them. * Find a donor trackpad. I’m tempted to just repurpose a thinkpad keyboard and then I get to easily integrate a trackpoint. I’m leaning more towards a custom mechanical keyboard though (probably Kailh Choc switches to cut down on thickness). Tablet has exposed USB interface over the dock connector that I can probably use for this. Trackpads sucked for decades but they got pretty decent in the last couple years so I’d like to integrate a recent one. I’ll have the vertical space due to the aspect ratio of the screen so I might as well install the biggest one I can find. I’m fairly certain these are all USB devices these days (last trackpad I modded was PS/2) but I’ll have to make sure since I only have a USB bus available. No UART or PS/2 and I don’t want to deal with adapters. * Come up with a secure method to mount the tablet to a chassis frame. If it aint smooth but rigid, I’ve done a poor job. * Plan the hinge. Clam shell is the best form factor for a device this size and I already know from experience that I don’t want to deal with a kick stand (else I might as well just replace the battery and move on). Slider would be neat but wildly impractical for a device this size. I’ve got junk laptops I can steal the hinges out of so I’m not too concerned on that topic. Weight might be an issue but there’s no reason I cannot just keep stacking hinge assemblies until it works. It looks like the tension is adjustable on the most hinge assemblies too so it probably won’t even come to that. Additionally, the hinges from the donor tablet kickstand might be suitable for a clamshell hinge too (though these ones are NOT adjustable). * Anything else I’m not thinking of right now. I’m aiming for a rolling target so things may change as I get closer to finishing or even after I finish it.
So now I have a list of tasks to accomplish before I can begin assembly. I think I’ve got a solid plan to actually start working on this though. I’m going to start with power though because if I cannot get that working to my satisfaction, the entire project won’t work.
power
The biggest benefit of using the Switch Alpha 12 for this project is that it does not require an internal battery to boot. This means that I don’t actually have to modify the tablet and I can just power it off a battery bank and move on. The downside is that the built in fuel gauge will not work, unfortunately, and will likely be the most expensive option. This is plan c. The DC adapter is 19v and USB PD supports 20v so I wager I can just mash 20v into the DC port and call it a day. The tablet will charge over the USB C port and will negotiate 15v for this so I wager there’s a generous window of supported inputs (though it cannot be booted without a battery when powered via USB C). I do not intend on blindly testing this if I can avoid it.
I found these LG batteries and these Sanyo batteries after doing some research last time I tried doing this project. I believe these should be a direct swap for the damaged cells I already have. I can just make a simple PCB that contains mounting points for assembly, battery holders for 18650s, and then connects to the original salvaged battery board. This should be the least amount of effort and probably the cheapest option. However, I don’t anticipate that these cells will last as long as lower voltage cells (though I’d have more time away from the wall, I think I’d end up having to replace them after a year or two) so I intend to leave this as plan b.
My next course of action is going to be to disassemble the tablet again and pull the main board out. I need to begin reverse engineering the internal BMS so I can determine safe input voltages. The ace in my pocket and plan a is that I’m hoping to reprogram the profile or restrict the charging profile to top out at 4.2v instead of 4.35v so I can just slap a half dozen 18650s in the base and call it a day. I do not know what hardware this tablet uses but I wager it’s using a TI chip and if so, I’ll have all the documentation I need to accomplish this. Hopefully Acer didn’t use custom parts or deface the chips. Hopefully I don’t find a rockchip BMS in there either but I think this device is more high end than those devices those parts usually end up in.
And that brings us to today. I don’t anticipate working on this regularly so I have no idea when (or even if) the next update will be. I don’t plan on doing this regularly but I do intend to keep these shorter than this next time. Thanks for the time and I hope this was entertaining at the very least. makho 2023-02-09
