site  contact  history  index

Librem 5 and Pinephone assembly

October 15, 2019 — BarryK

The first prototypes of these phones are built and shipped to some developers.

The Librem 5 is planned to be manufactured in batches that will be progressively refined. The first one is named "Aspen", to be followed by "Birch". Here is a video showing Aspen being disassembled and reassembled:

Interesting, it has no heatsink. A heatpipe is planned for Birch. The speaker, Tod Weaver, states that he has to charge the battery twice per day, but is hoping it will eventually get down to once per day. That would be very light usage without heatsink. I wonder how many hours of actual usage in the day?

Whatever, that is so far off we expect with modern phones. My Huawei rarely gets below 90% in a day, in which I might, say, go on a train ride to the city and be browsing all the way there and back. If drive somewhere and use the GPS, it does consume a bit more.

The SoC is, I think 28nm technology, and I think I read that late in 2020 they plan to progress to a 14nm NXP SoC. The two m.2 sockets with external modem and ...what's in the other socket, can't remember ...anyway, that configuration is going to be current-hungry.


The Pinephone is an alternative. They both have modem separate from the SoC, in the case of the Pinephone it is soldered in, so the phone is much slimmer. Here is a video showing assembly of the first prototype:

Interesting, it looks like the back of the LCD is being used as the SoC heatsink. I haven't found any information on battery life, but doubt that it will be good news ...but premature to give any opinion.


One good thing about the Pinephone, it has a modem with frequencies suitable for Australia.

Tags: tech

Two inlet nozzles works

October 14, 2019 — BarryK

I posted yesterday that the inlet design was not spreading the water wide enough:

I have jury-rigged a setup with two nozzles. Externally, from the linline-filter there is a T-junction and two 4 litres/hour drippers, then 4mm trickle pipe taped onto the distiller:


...the 4mm pipes are taped on the up-flow-side of the aluminium channel, so as to flow horizontally somewhat before seeping underneath the channel.

This looks good. I managed  to take the photo at an angle where you can just see how the water has progressed down the cloth. It is covering the width of the cloth. Interesting, it looks like the left-hand outlet has a very slightly higher flow rate.

Next step, redesign the inlet to have two pipes. 

Tags: nomad

Inlet pipe design not satisfactory

October 13, 2019 — BarryK

Today I setup a test of the inlet pipe in the solar water distiller prototype #3:


...from the water container, there is an inline filter, then the flow-reducer as described in an earlier post:

Then there is a 13mm-to-4mm trickle-pipe reducer, and silicone tube into the distiller. Water comes out of the top-side of the aluminium square-section channel, so flows along the top before working its way underneath the channel and down the wicking cloth.

The design was a punt. I thought that the spreading effect would work, but didn't know how much. In the photo, the water has trickled down about halfway in the centre, and if you squint at the photo, you might just make out the spread. It is good, but not wide enough to reach the sides of the cloth.

So, need to rethink the design. Two holes would probably do it, but the problem is to get equal water flow through them. I don't want to put any trickle parts inside the distiller, as the temperature might get up to 65 degrees C and the plastic might soften. Thinking to have two flow-reducers externally. 

EDIT 2019-10-14:
Problem solved, see post: 

Tags: nomad

Attempted to compile Chromium

October 11, 2019 — BarryK

The attempt was on Easy Buster 2.1.6, booted off USB-stick. Although it failed, it would be good to document the steps...

It is a massive project, with a lot of dependencies. /tmp requires 20GB free space, which was the first hurdle, as easy has a tmpfs mounted on /tmp -- which is done in the 'init' script before switch_root -- so I modified 'init' not to mount /tmp. A partition with 50GB space is required, and at least 8GB RAM is recommended.

I found some build instructions, inside a source tarball downloaded from here, size about 1GB:

The instructions, though, downloaded the complete source again, did not use that tarball. Following those instructions:

# git clone
# export PATH="${PATH}:/mnt/sde1/chromium/depot_tools" # mkdir chromium
# cd chromium
# fetch --nohooks --no-history chromium
# cd src

There is a script to downloaded any missing dependencies, however, it aborted. It requires certain versions of either Ubuntu or Debian, and although Easy uses Debian DEBs, it is not Debian. This was the attempt:

# ./build/ --unsupported

I looked at the script, and was able to determine what deps are required, and manually installed them, and their dependencies, via the PETget Package Manager. This is what I installed (as well as their deps):

ninja, cdbs, devscripts, libbrlapi-dev, libpam0g-dev, libspeechd-dev, openbox,
 p7zip, python-cherrypy3, python-crypto, python-numpy, python-opencv, python-openssl,
 python-psutil, python-yaml, rpm, librpm-dev, ruby, openjdk-11-jre

I then put in path variables for Java:

# export PATH="${PATH}:/usr/lib/jvm/java-11-openjdk-amd64/bin"
# export JAVA_HOME=/usr/lib/jvm/java-11-openjdk-amd64
# export JAVA_BIN_DIR=/usr/lib/jvm/java-11-openjdk-amd64/bin


# gclient runhooks

The instructions explained some variables that can be applied, however, didn't explain how to apply them, so I took a punt and preppended them:

# enable_nacl=false symbol_level=0 blink_symbol_level=0 gn gen out/Default

Now to start the compile:

# autoninja -C out/Default chrome

After awhile, fail, with errors like this:

/usr/include/x86_64-linux-gnu/bits/stdio.h:58:10: error: use of undeclared identifier '__getc_unlocked_body'
return __getc_unlocked_body (__fp);

Googling returned nothing, however, I think the problem may be changes in stdio.h. Chromium is, I think, compiled on earlier versions of Ubuntu and Debian (16.04 and Jessie), older glibc. I left it at that. 

Tags: easy

EasyOS Buster-series version 2.1.6 released

October 09, 2019 — BarryK

Version 2.1.3 was released 3 weeks ago, see the blog announcement:

There are some bug fixes, but the big news is the incorporation of the 'nm-applet' GUI tray applet, for network management. The source is patched so as to integrate with EasyOS. In particular, it can be popped-up by clicking on the "connect" icon on the desktop -- after booting 2.1.6, try it!

Networkmanager is now better integrated, so that the user can switch between the older network management systems, such as SNS and PupDial. The Connection Wizard is still available by right-clicking on the "connect" icon, or in the Setup menu.

Release notes are here:


If you need assistance with installing, read this:

There is a thread in the Puppy Forum for feedback: 

For any interested developers, here is a tarball of woofQ as used to build Easy Pyro 1.2.5 and Buster 2.1.6: 

Have fun!

Tags: easy

Installing EasyOS to laptop is too easy

October 08, 2019 — BarryK

I have a very cheap Acer Aspire1 laptop, with Apollo Lake CPU, 4GB RAM and 64GB eMMC storage, passive cooling. It cost AU$240, purchased January 2019. It came with Windows 10 S, with free upgrade to 10 Pro. I have posted about problems with Windows on that machine.

Anyway, I don't use Windows, well, extremely rarely. The laptop sits beside my lounge chair, for comfortable browsing, and it has EasyOS on a USB-stick, protruding from the side of the laptop -- awkward, when the laptop is balanced on the lap.

So why not take that extra step and completely replace Windows with Easy, then no need for a protruding USB-stick?

OK, booted up EasyOS Pyro on USB-stick, with the 'easy-1.2.5-amd64.img.gz' file on it, then just ran:

# easydd easy-1.2.5-amd64.img.gz

It gave me a choice of installing to the internal eMMC (/dev/mmcblk0) or the USB stick (/dev/sda) and I chose the former. That's it, installed.

Powered off, removed the USB-stick, powered on, and EasyOS started up. Didn't even have to setup anything in the UEFI-firmware. How's that for easy?!

I wrote about the same type of installation, to the entire drive, in my Mele mini-PC: that case, I did have to set the UEFI-firmware to boot the SSD. 

Tags: easy

EasyOS Pyro-series version 1.2.5 released

October 08, 2019 — BarryK

EasyOS 1.2.5 has been released. This is the latest in the "Pyro" series, version numbering 1.x, which is supposed to be in maintenance mode, as development is focussed on the "Buster" series, version numbering 2.x.

However, the Pyro series is getting more attention than just maintenance. I have compiled 'network-manager-applet' with my patches, so there is now a nice GUI, replacing the text-mode 'nmtui'.

Release notes here:


...English, French and German builds, and there is an ISO if anyone needs it.

Installation notes:

Feedback welcome in this thread of the Puppy Forum: 

Tags: easy

Assembling inlet of water distiller

October 08, 2019 — BarryK

Continuing construction of solar water distiller prototype #3, previous post here:

Holes needed to be drilled to mount the inlet pipe. The small holes are 4mm diameter, the large one is 6mm:


...there was a bit of messing around deciding exactly where the holes will be drilled. Exact measurements will be in the final DIY page. I assembled the central glass with silicone mat and wicking cloth, then placed the inlet pipe on top, to find out where those holes needed to be.

I didn't really want to glue anything together, as the aim is that the distiller can be totally pulled apart. However, the mat sits on the glass and want the edges folded up so as to contain the water. The top up-slope end of the mat is a problem, want it to stick up about 90 degrees from the plane of the mat. I used Selleys 401 silicone sealant to adhere the mat to the glass and achieve the top lip.

Here is the inlet pipe assembled:


...using the black towel used in prototype #2, which has a 6.3mm aluminium rod inserted along the top of the towel. The towel is cotton, from Kmart, it is plain black, no embossing. The aluminium rod is from Bunnings.

The rod was AU$2,32:

The towel was AU$6:

The great thing about the towel is that it already has a folded-over-end through which the aluminium rod fits snuggly:


As you can see in the second-last photo, the silicone mat top-end lip is not sticking up at 90 degrees, more like 45 degrees. So what I did was fold it right over 180 degrees and apply silicone sealant:


...left that overnight. Yes, that did the trick, it sprang back to about vertical:


The next exercise will be to construct the distilled-water runoff. That is tricky, as distilled water will be running down both top and bottom glass sheets. I have considered various ways to construct it, and trying to do it without any more gluing.

Tags: nomad