Log The portaNUC - turning a NUC into a UMPC handheld system

[CC Ricers]

Not much going on with updates on this build, as I haven't touched this PC in days, but it does seem to be charging well and I can put it on sleep mode for about a whole day, completely on battery power.

The screen sometimes refuses to properly display video when waking up the PC, and I have to manually turn the screen off and on to make the video work again.

Biggest obstacle I'm trying to solve right now is how to make the enclosure. The BMS's heatsink on the MOSFETs is what allows it to take up to 40 amps of current. I don't need anywhere near that much current, and the heatsink never gets warm during extended use. But it creates problems designing the case and I don't want to remove the heatsink just yet.

Placing the BMS facing opposite of the battery compartment, so it faces the wall of the case, looks like the best option. The enclosure just needs a separator so that there's no risk of shorting out the batteries with metal contact from the BMS. Maybe, for this first version, I will leave a hole open for the heatsink so it is exposed "hot rod" style. No harm in doing that, since it's separated enough from anything electrical.

 

[CC Ricers]

I'm finally making more progress with the middle part of the enclosure. A lot of care has been done to make sure all parts and cables can be properly routed and shaping the nooks and openings the best way I could.

Top view with the rear I/O exposed

Bottom view with the front

The following view shows most of the interior of the pieces. The top has several notches to hold the screen in place. A separate piece will be modeled with extra notches. It would be screwed on to the front after the screen is placed in the middle. This will prevent it from falling inwards.

The larger side holes are pass through-holes for cables. The longer one in the back is for the controller inputs. Since the buttons are located on the left and right sides, there needs to be a cable to connect them in between.

There are screw holes for joining the top and the bottom pieces, and for joining the end pieces which are yet to be made. Also, the front and back is made in mind for replaceable panels for different I/O. For now I will just lave the back wide open with the current set of cables that I have. It's still just a prototype so I am not making everything 100% neat and precise. I also don't expect it to be extremely ergonomic, just prioritizing that everything can fit inside with no problems.

The top half is more complete. I can expect to have the top part ordered for a 3D print later today or early tomorrow. Probably clear PETG or ABS. Polycarbonate would be nice, but probably better for a more refined build.

 

[aromachi]

Watching! Your CAD skills are impressive. Awesome project!

 

[CC Ricers]

The top half pieces have been ordered for printing. I chose to print them all in black because it was still cheaper than printing one part clear and one part black.

I'll finish on modeling the bottom half, but the the top half is now "locked" in. The final dimensions of this section are 190mm x 120mm x 33mm. Even without a controller, this is a self-contained computer unit that contains its own power source. It's only 0.75 liters, which would make it interesting to pit it in the Performance Per Liter thread!

 

[jakejm79]

The top half pieces have been ordered for printing. I chose to print them all in black because it was still cheaper than printing one part clear and one part black.

I'll finish on modeling the bottom half, but the the top half is now "locked" in. The final dimensions of this section are 190mm x 120mm x 33mm. Even without a controller, this is a self-contained computer unit that contains its own power source. It's only 0.75 liters, which would make it interesting to pit it in the Performance Per Liter thread!

I think the powerbrick needs to be included in the PPL, regardless of battery power or not, also wonder if you'd be able to complete the required benchmarking on a single charge :-)

 

[CC Ricers]

I think the powerbrick needs to be included in the PPL, regardless of battery power or not, also wonder if you'd be able to complete the required benchmarking on a single charge :-)

If that's the case I'll have to use the stock Intel NUC adapter because it is a bit smaller than the one used to charge the batteries

Looks like I need to run three benchmarks, one for at least 30 min. I'm more concerned about compatibility than battery life, with the Time Spy benchmark I wonder if it will even run on Intel graphics. But in the future when I get a Ryzen powered NUC, things will improve a lot.

 

[jakejm79]

Should run on the NUC no problem.

 

[CC Ricers]

The top half of the shell has been printed, and it's shipping fast. Still working on the bottom half, though...

Even so, I'm already brainstorming design ideas for the next iteration. The next one will stack parts in three layers, and have a sleeker design while using mostly the same dimensions. It should also be less complicated in figuring out the hole placement for wires and connectors.

 

[CC Ricers]

Well, this has been long due for an update, but it's a bigger one.

In the following week and a half I have received two versions of the top shell that I ordered to be made. The first didn't quite fit the screen, and neither did the second one, even after making adjustments. I have filed the design a bit but the openings were way too tight for the screen, especially seeing that they have tabs that hold the screen in place.

It's was just too difficult to slide the screen in. And having applied a lot of force, I eventually cracked the screen from the inside and it's now useless, the panel just shows a lot of lines.

Going back to the drawing board for a moment, I decided to take a look at some other screens to use, since it was lacking in a few features I wanted, especially touch controls. So I bought two touch screens for DIY purposes. One is a Geeekpi screen and the other is from Uperfect, the same maker of the screen that I broke. Both are touch screens but the Uperfect has built in speakers. And they actually work this time! Here are both screens, the one with speakers is on the right.

Having received the Geeekpi first, I spent more time with it, but the Uperfect one showed up just a couple days ago and I favor this one by a mile. The Geeekpi screen doesn't have a IPS panel unlike the other 7-inch screens I have, so viewing angles are not so great and colors have a blue-ish tint that needs to be offset with image tinting software. Other than that, there are no speakers on the thing and I would have to figure out another way to add sound. But mainly it's the colors that I don't like. Something that was also not noted in any Amazon reviews is that there is noticeable color banding in very dark and very bright colors. The panel likely supports only 6 bits per color channel. This would also have needed software workarounds (like from your display card driver) to enable dithering and lessen the effect.

So the Uperfect screen wins this one. Better colors and viewing angles, no banding, and built-in speakers means I don't need to solve the problem of adding sound (freeing up another USB port). About the only thing I don't like is the space it takes up. The mounting holes are on tabs that stick out farther compared to the Geeekpi, which potentially increases the size of the case I would put it in. However it's been mentioned by a few users that these can be chopped off with no problem.

Also, there are four permanently fixed standoffs on the rear for mounting a Raspberry Pi. I'm not going to use a Pi but I can't remove these standoffs and don't want to force them out and risk breaking the board. Interestingly, there are unused standoff points for a Raspberry Pi Zero (it's even labeled as such) but my accessories didn't come with extra standoffs and the manual doesn't talk about installation with a Pi Zero. I just thought that was peculiar.

Anyhow, if the standoffs can be trimmed I can at least make the profile of the handheld system narrower. But it needs to be done cleanly. So I VERY CAREFULLY used a Dremel cutting wheel to chop off two of the standoffs about halfway through, and then I used a sanding bit to remove the sharp metal bits and smooth it out some more.

With all that done, I tested it again and it still turns on. Awesome, I did not wreck the screen.

Now to go back and re-work the design of the enclosure to fit this new screen. I probably will be chopping off at least two of the mounting holes on the corners, and have it fit through the edge of the board only.

 

[CC Ricers]

The PortaNUC is harder than I thought to make more slim. Currently have it around 38mm to 40mm of thickness when I stack the parts on top of each other. Rather on the chonky side for a portable, or about the same width as a regular 2-slot graphics card.

Good news is I have my new 3D printer already set up and maybe I should print some rough drafts to mock up how it feels to hold that kind of case in your hands.

 

[Valantar]

With some rounding and indents in the right places that thickness should be very comfortable. There's a reason you have tons of third party grips and the like for the Switch - holding something that thin and flat just isn't what human hands are good at.

 

[aromachi]

4cm isn't that bad. Heck, the original gameboy was just over 3cm in width and that was in my hands for 5+ hours a day when I was a kid lol

 

[CC Ricers]

I'll go with that, then. Chunkier in the middle and perhaps narrower at the ends with some shaped grips.

The battery balance board is giving me more trouble than expected to place it in. It doesn't have any mounting holes and need to design some kind of battery compartment with clips to hold it in the flat areas where no components are sticking out.

I also got Wii U controller board with buttons. Going off the custom modded Wii U that @thewizzard1 posted, it will save me the trouble of designing my own custom buttons and circuit boards. That also means it will adopt the Wii U layout of both thumbsticks being towards the top. Curious how that will be like while playing.

 

[CC Ricers]

I have 3D printed a rudimentary bezel design for the screen. It ended up wider than I expected, but the black color does a decent job with blending in the irregular borders of the screen and it has a bit of a "pristine" feel with the glossy surface of the bottom layer. It will probably even look better if I used a tinted acrylic sheet to mask out the borders more.

I printed two bezels because one had more inaccurate hole placements. The second one still is a bit inaccurate because increasing the chamfer moved the hole placements- oops. Gotta watch out for that with parametric modeling. Instead of printing yet another one I'll just drill offset holes through the board and going with this for now.

When playing full screen games or watching full screen videos, the bezel is less apparent, especially in darker lighting. Note the reflective surface in both the screen and bezel makes it more seamless under these conditions.

And unrelated, but here are the spare Wii U gamepad parts that I bought. I was originally going to use all the parts, but decided against it because I wanted a more Xbox-like layout where the right thumbstick is under the buttons instead of over it like the Wii U. Which means I will use some prototype boards to secure the buttons. The joysticks will be used, however. Instead of using its short 5-pin cables, custom wires will be soldered underneath the points to connect them to the microcontroller.

(the object in the bottom right is not another thumb stick- it's the top of my torx multi-screwdriver )

 

[CC Ricers]

Work has been quietly underway on the handheld case. Yet again I have redesigned most of the case, but I found a very good middle ground for portability and space efficiency. The middle area is still 38mm thick for the housing of the screen and NUC but the side handles are 27mm which is just enough to fit the batteries and the PCBs for the face buttons. The joysticks are taller and have to be positioned in the space behind the batteries, meaning it will go back to a Wii U layout.

The balance board is the smaller board right of center, which has its own housing and uses the standoff screws in the NUC to secure it in place.



A few of these parts have been printed already. More notably the screen bezel is now narrower and sleeker, and makes the screen appear bigger.

Also, a "bump" piece needs to be made to hide the cables coming out of the HDMI and DC power ports.

 

[CC Ricers]

Now that the Steam Deck has been announced it has the opportunity to trounce crowd-funded systems like the Aya Neo with its very competitive pricing. Its form factor is interesting- judging by the 7 in screen in the middle, it is wider than my own handheld and more space is given for the controls. And I thought my handheld would be large at 265mm long.

Even so I will continue building my handheld. I also had the idea of a smaller one with cut-down specs, running off a PC stick. I know that hundreds of Raspberry Pi powered handhelds exist, but I want to be able to run some PC games natively and Box86 isn't quite there for Pi-based systems yet.

 

[Apotessar]

Even so I will continue building my handheld. I also had the idea of a smaller one with cut-down specs, running off a PC stick. I know that hundreds of Raspberry Pi powered handhelds exist, but I want to be able to run some PC games natively and Box86 isn't quite there for Pi-based systems yet.

That's the spirit. And I will continue to read diligently :-)

 

[Elaman]

Now that the Steam Deck has been announced it has the opportunity to trounce crowd-funded systems like the Aya Neo with its very competitive pricing. Its form factor is interesting- judging by the 7 in screen in the middle, it is wider than my own handheld and more space is given for the controls. And I thought my handheld would be large at 265mm long.

Even so I will continue building my handheld. I also had the idea of a smaller one with cut-down specs, running off a PC stick. I know that hundreds of Raspberry Pi powered handhelds exist, but I want to be able to run some PC games natively and Box86 isn't quite there for Pi-based systems yet.

The Valve engineers probably visit the forum frequently. Maybe they saw my all-AMD Arch-based SteamOS build and saw the error in their ways, decided to go AMD APU Arch-based themselves. Or maybe I am just kidding. Or maybe they might take ideas from this build log as well!

But anyway for what it's worth, it is more interesting to me to follow here your process and own conclusions. Customization all the way.

Regarding Steam Deck itself, it is a good observation that the market has been building up for a while. OneXPlayer already exists, and furthermore there's been the GPD Wins and the Aya Neo if one wants to go full-AMD. I don't know about benchmarks, but I expect Intel GPUs to be very competitive in this niche market.

 

[CC Ricers]

It has given me more motivation to finish it earlier. Feels like I'm now racing against it

However I did have some photos from last week showing the parts that I have already printed. I now have the thinner bezel of the new design along with the controller grip bottom parts. You get a better idea of how components will be grouped together. These parts still need some tweaking before they are ready to be attached together completely.

Playing some Mirror's Edge here, a game that never made it to the Switch. And it's still fun.

Empty shell piece of one of the controller ends, where the joysticks and batteries are mounted.

That's going to be a tight fit with the analog joysticks. And I still need to place the shoulder buttons beside it...

 

[CC Ricers]

A quick update here- The biggest part was printed last night which is the bottom shell where the NUC is mounted. I can attach the side pieces and finally hold something that resembles a case in my hand. The screen part does not comletely close all the way but another part can be made for that to adjust it. Turns out I didn't account for the little speaker connectors that touch the NUC heatsink when attempting to close the thing.

I did about an hour of gaming with the NUC inside the shell and did see the drawbacks of PLA+ when part of the bottom shell started bending under the heat.

The CPU gets up to 85C while gaming but it's actually the RAM that is the closest component to the plastic. The RAM has no heat spreaders and you can really feel the heat in the middle. I might get a 3rd party maker to print it with a more heat resistant material.