Prototypes Updates and next Talks – October 2021

Prototype delays due to electronic components shortages

As already stated a few times, we are still victims of the electronics industry supply-chain difficulties. Back in July, we informed you that “98% of more than 2000 components are now secured and will be delivered on time. The hunt is still ongoing for the remaining forty components left, and finding them is crucial not to miss the October deadline.”Some of the power management components are currently unavailable, so the electronic designer had to search for their replacements. Soon we will publish the resulting updated PCB design reporting all new components. The production factory has not yet received all the required components that we already ordered, and there are still that so far cannot be found anywhere on the market. In particular, we are facing problems getting the HDMI connector (part number 2041481-1) that could fit inside the Eclipse notebook chassis. If you are able to help us find such a connector, please contact us. We are urgently looking for 3 pieces of this connector for the 3 prototypes. Additionally, we are also looking for a solution for the larger batch production.

Unexpected increase of 1000 euros for the prototypes

We are very happy about the generous participation of all donors that allowed the prototype campaign to exceed 90% of the final goal. Thank you very much!

During our surveys on the electronic markets last September, we observed a skyrocketing increase of the prices. We are a group of hobbyists that have zero power to sit down and bargain with electronic companies. Even the well established Raspberry Pi foundation was forced into increasing their prices (see https://www.raspberrypi.com/news/supply-chain-shortages-and-our-first-ever-price-increase/)

As a result, every prototype increased its final cost by around 300-320 euro including 22% local VAT, for a total amount of 1000 euros including PayPal fees for the three prototypes. Long story short, we have to increase the campaign goal from 12500 to 13500 euros.

We currently cannot tell if the market prices will go back to lower prices, and, moreover, when the current electronic components shortages will be finally over. We all hope that the situation will get better by the time the full batch production starts.

Now, the bright side of the overall situation is that being forced to wait for electronic components is quite compatible with the slow pace of our donation campaign, so please, continue donating!!

MXM Video Cards

You may get an idea of the current electronic shortages by the fact that we ordered an AMD Radeon E9172 MXM GPU (approximately 295 EUR with VAT) and an AMD Radeon E9174 MXM GPU (approximately 380 EUR with VAT) back in May 2021. Well, the expected delivery date is 27th of November 2021!

At the moment, the cost of the three MXM cards needed for the prototypes are not covered by the donation campaign, but we ask your financial support for those cards also.

In the meantime, we had the chance to buy an ATI Radeon HD4650 1GB DRR3 MXM 3.0 card and, thanks to the kind donation by Stefano, a new collaborator from Italy, we have now two AMD FirePro M4000 GDDR5 1GB MXM 3.0A cards.

ACube Systems, our partner taking care of building the prototypes, has also purchased a PCI to MXM adapter. The adapter will allow us to test MXM cards before we have the prototype ready, as it will be used in conjunction with the motherboard “Sam460ex” made by ACube Systems. Tests will be performed under AmigaOS 4.1, a native PowerPC operating system.

Switch to the Cern 2.0 License under evaluation

We are currently evaluating the possibility to upgrade our Open Hardware license from Cern 1.2 to 2.0.

The first thing we noticed was that the second version is split into three variants called Strongly Reciprocal (S), Weakly Reciprocal (W) and Permissive (P). Basically, all three documents are structured in the same manner and, indeed, some sections are identical. The main differences are found in sections 3 Copying, Modifying and Conveying Covered Source, 4 Making and Conveying Products and 5 Research and Development (which does not exist in the Permissive license). The changes that can be found comparing one document to another also imply different concepts to be explained in section 1 Definitions. Most terms that appear on more than one document have the same definition. Important exceptions to this are 1.1 “License” which refers to the exact variant of the license on each document and 1.7 Available Component which is not exactly the same on R and W (and can’t be found on S)

As we understand, the variant should be chosen depending on the restrictions related to the components (Available Components) and additional parts that could be added by the Licensee (External Materials). OHL-S specifies that all “the Complete Source is the Covered Source” and any modification should be distributed using the same license. On the other hand, OHL-W allows the inclusion of External Materials, which means that you can add some parts to the design using a different license. Finally, the Permissive license does not mention anything about Available Components and External Materials but allows to make or publish a Product only including all the Notices of the Licensor.

To better understand the differences, the examples provided in the Open Hardware repository FAQ page are quite explanatory:

What are all these suffixes?

In the software domain, there are three generally acknowledged licensing regimes for free and open source software: permissive, weak copyleft and strong copyleft. There are tastes and use cases for each option, and the same happens in hardware. We use the word “reciprocal” instead of “copyleft” because the underlying rights in our case are not restricted to copyright. So, when you use the licence, you need to add a Notice to your designs with one of the three following suffixes: S, W or P:

  • CERN-OHL-S is a strongly reciprocal licence. For example, if you release HDL files under CERN-OHL-S and then somebody uses those files in their FPGA, when they distribute the bitstream (either putting it online or shipping a product with it) they need to make the rest of the HDL design available under CERN-OHL-S as well.
  • CERN-OHL-W is a weakly reciprocal licence. For the example above, if you release your part of the design under CERN-OHL-W, somebody who distributes a bitstream which includes your part does not need to distribute the rest of the design files as well.
  • CERN-OHL-P is a permissive licence. It allows people to take your code, relicense it and use it without any obligation to distribute the sources when they ship a product.

Compared to this second version, OSHL v1.2 does not include “Available Components” and “External Materials” terms, making it difficult to establish a direct relation with any of these variants. This makes us think that it is probably more similar to OHL-S.

Regarding the Disclaimer section, which is the one that protects the Licensor from legal issues and warns the Licensee about his responsibility, both versions have a very similar writing. Version 2 is slightly more detailed specifying that “The Licensor shall, to the maximum extent permitted by law, have no liability for […]” while the previous version did not mention the limits created by law. In any case, we think these limits are oblivious and the meaning of the Disclaimer section is equivalent.

Again, to compare OSHL v1.2 and OHL v2 we can make use of one question in the FAQ section:

I am a user of CERN OHL version 1.2. What are the main changes introduced by this new version?

Version 2 of the CERN OHL improves on version 1.2 in various respects:

  • The new version comes in three variants: strongly reciprocal, weakly reciprocal and permissive. Reciprocal licences stipulate that changes to a design must be fed back to the community, for everybody to benefit from them. Permissive licences do not impose this condition. In this way, CERN OHL v2 caters for the different collaborative models currently used in Open Source Hardware projects.
  • In the reciprocal variants, it is very important to clarify the scope of reciprocal obligations. By introducing the concepts of “Available Component” and “External Material”, plus the already-existing concept of “Product”, the new version makes a special effort to clarify what sources should be shared in both the -S and -W variants.
  • CERN OHL version 1.2 included a patent licence, i.e. a promise by the licensor that (s)he will not sue a licensee for patent infringement as regards the design licensed under CERN OHL. Version 2 adds a reciprocal clause for this patent licence: if a licensee sues a licensor for patent infringement, (s)he loses all the rights granted by the licence.
  • In licence 1.2 we did not make a special effort to cater for Hardware Description Language (HDL) development as used in Field Programmable Gate Array (FPGA) and Application-Specific Integrated Circuit (ASIC) design. As we became convinced that there was no appropriate reciprocal licensing regime for HDL, we made sure that CERN-OHL-S and CERN-OHL-W can provide a good solution for FPGA and ASIC designers with a reciprocal mindset.
  • Version 2 makes a special effort to maximise the chances that the recipient of a physical product will get access to the design files for that product. It does this by granting the licensor the possibility of embedding a URL or another reference in the object itself, and establishing that downstream licensees should respect that notice and update it as applicable if the design is changed.
  • The new version provides a grace period of 30 days for licensees which infringe in terms. If they come into compliance within 30 days after receiving a notification from the licensor, their rights are reinstated. This is meant to help with cases in which a licensee infringes the terms of the licence inadvertently.

We are still studying which alternative is better from OHL-S, OHL-W and OHL-P. The decision should take into account how free we want the licensees to be when producing or modifying our design.

Join our Talks at OpenPower Summit 2021, Linux Day Online Italy 2021 and the Sfscon 2021

For more news and updates about our project and future plans, join our talks during:

Linux Day 2021 Online (23rd October 2021 3 PM CEST) “When the community produces an Open Hardware laptop”  in italian .

At the OpenPOWER Summit NA 2021 (28th October 2021 at 2.45 Central Daylight Time (CDT), UTC -5 ) “Prepare yourself to switch computing to Open Hardware Power Architecture”

At the Sfscon.it PPC64 Open Hardware Notebook prototypes around the corner (12th November 2021 2.30 pm CEST).

Hurry up, vote for the motherboard name!

As of the 22th of October 2021, we’ve reached 987 votes out of the 1000 required to take a final decision on the motherboard name. We would like to close the pools as soon as we have reached that number to start working on the drafting of logos and other communication materials.

Freedesktop-sdk merge requests to support PPC64 Big Endian

Thanks to Charles and Manuel, members of our software team, we have submitted a merge request to Freedesktp-sdk with a patch to allow the compilation on PPC64 Big Endian. That was a major achievement and an enormous effort by our volunteers. A very good job guys! Thank you!!

Our ppc64 Be branch of Freedesktop-sdk.

Freedesktop-sdk was going to stop supporting the PowerPC architecture due to the lack of a builder . The situation has now changed, and we are now very happy to inform that even thanks to help from OpenPower members ( from OSUOSL), in terms of updates and improvements to the ppc64le branch of freedesktop-sdk.

Dummy board being finalized

It is with great joy that we present you the first tangible result after years of spending time on planning, ideas, projects and schematics. Below you see pictures of the dummy board, a non-working prototype that was printed with a two-layers PCB that was paid thanks to the ongoing donation campaign.

Top side of the dummy board.
Bottom side of the dummy board.

The primary use of this dummy board is to perform mechanical checks in conjunction with the Slimbook notebook chassis. The board is not finished yet, the PCB designer still has to mount additional mechanical components such as connectors to ensure the final working prototypes will fit perfectly in the Slimbook Eclipse chassis.

The PCB designer in charge of the job is carefully working to fine tune the gerber design files and already adjusted some minor details, proving that a preliminary dummy board was very much needed.

We would like to thank Gerard Schneider that kindly offered us a ATI Radeon 7970 MXM card, it will surely help us testing the working prototypes that will be produced later on. We welcome anybody else willing to send us other Radeon MXM cards that may lay unused in a corner, we would like to start as soon as possible to test various GPUs in the upcoming working hardware.
[UPDATE 2021-04-22] Unfortunately our notebook board is set to work exclusively with MXM-A 3.0 (type A) with a size of 82mm x 70 mm and with a maximum power consumption of 55W, whereas the MXM card provided by that Gerard Schneider is an MXM-B (type B) with a size of 82mm x 105mm and a maximum power of 200W. Thank you anyway Gerard, your card will be useful to check and eventually fix the video drivers but it will not be used inside the prototypes.

ATI Radeon 7970 MXM card.

Even if it is “just” a dummy board, this is a great milestone, and we are really happy about it because we can finally touch something with our hands. We would like to thank all the people that made it possible to reach this point, and we really hope that the donation campaign financing the final prototypes will speed up because now we all want to see more!!

Are you willing to help?

Being part of a project like this could be an amazing experience, you meet new people, volunteers of other projects, companies devoted to open source and everyone is willing to help. We are continuously giving examples of this in our blog posts but, in the last weeks, we are especially grateful about the support received from KiCad developers and Slimbook.

Two additional enclosures for our prototypes.

Slimbook is a company making a huge effort in promoting an Open Source environment. They produce notebooks, mini-PCs and desktop computers targeting mainly Linux users. As an example of their commitment to the open source community, they have a very have a good relationship with the KDE project and together they collaborate on the creation of laptops meant to use primarily KDE. Despite being a small company, they are having success selling their products worldwide and these are very appreciated by the Linux community.
As you may know, we started our collaboration with Slimbook more than two years ago and they have been always quite helpful promptly responding to our requests and providing information about the enclosure design or the related components that will be also used in our notebook (screen, keyboard, dissipation devices, etc.). All their support and time was kindly offered for free. Besides that support, we have received two Slimbook Eclipse enclosures to continue our tests. This will make possible to assemble three prototypes of our PPC Notebook. Again, they did it for free. We have no words.

A Slimbook Eclipse enclosure kindly donated by Slimbook.

Export our PCB to KiCad, a difficult journey

At the very beginning of this adventure, we were trying to find hardware experts to design the motherboard but the level of expertise required for such type of hardware made this challenge unachievable for us. Of course, we have experts on that field but the complexity of this design demands quite a lot of time, impossible to carry out solely using the volunteers spare time. So we opted to look for a company experienced in motherboard design and even more difficult, a company that was experienced with the PowerPC architecture.

We were lucky enough to meet ACube Systems and its circle of collaborators. However, as most for-profit companies do, the ACube System subcontractor company had its own proprietary software tools which generates file encoded using non-open outputs formats. In our case we end up with files created using Mentor Xpedition, a software that cannot exporting to KiCad. To convert our Mentor Xpedition source files we were told to import them into Altium, and import the converted Altium files into KiCad.

Unfortunately, the KiCad importer for Altium files is still heavily under development, and it is far from being complete. We contacted the KiCad developers and they kindly accepted to perform some testing with our Altium PCB files and that helped spotting various errors in the conversion procedure. These error were identified by the developer in charge of the Altium import module for KiCad and he is currently addressing the encountered issues. Regarding the BOM (Bill Of Materials) the guys at KiCad recommended to import the Altium schematics to KiCad, and generate the BOM from there.

Obtaining an open source format for publishing our motherboard PCB is very important for us, as it allow anyone to easily access the result of our efforts to deliver a truly and fully compliant Open Hardware design.

After a few attempts, the guys at KiCad suggested another option: instead of converting the original Mentor Xpedition files to Altium, they suggested to load them using FabMaster. In fact, KiCad has another importer dedicated to FabMaster (for the board only) and the result of this import module should be useful to understand the level of accuracy of the Altium importer. In theory, the Altium import should produce better results with respects to the FabMaster importer as it is a newer. We are currently investigating if we can follow this path, as it seems to require a full Xpedition license, therefore we are in the process of contacting the subcontractor engineer to explore this solution.

An AmigaOS4 AHI driver for our sound chipset

Our notebook motherboard is open to any operating system supporting PowerPC. Among the operating system that could possibly work, there is AmigaOS 4, a closed-source system that already works on the E-AON AmigaOne X5000 that mount either a NXP P5020 or a P5040, which are both PowerPC Book3e e5500 CPUs. These CPUs can be considered the previous generation CPUs with respect to our T2080 (PowerPC Book3e e6500), one of the main differences is that they lack the Altivec unit, which the T2080 has.

On the April 1st the Dutch developer H. Kanning (nickname “geen_naam”) announced the availability of an AHI sound driver supporting HD Audio compliant chips, and explicitly supporting the C-MEDIA C8828 that we selected for our motherboard. At first we though it was an April fool, but then it was confirmed to actually exists and work, meaning that another operating system is one step closer to being supported. Great job!

PCB Complete sources published! Mentor Expedition, Altium and Kicad formats

At the beginning of March the consultant engineer paid using the donation campaign provided us with the Mentor Xpedition source files of our PCB. Providing source files created with a proprietary software is not ideal, therefore we have worked to convert the sources to the Open Source KiCad format.

To achieve the porting of the sources, we first tried loading the Mentor Xpedition sources using the PCB Design Software Altium, and from there, export the sources to KiCad.

We were pleasantly surprised by members of the KiCad team that promptly answered our call for volunteers able to help us in the source translation process, thank you guys, it was really much appreciated!

In our task, we found a very useful post on the KiCad blog that explains how to import an Altium PCB design file in Kicad.

Apparently the Altium Importer is not available if you start PCB window from the KiCad project manager window, you have  to start it from the command line as pcbnew-nightly to get the KiCad import feature for loading alien formats.

KiCad eeschema-nightly currently does not support importing Altium schematics. There is an ongoing discussion, so perhaps there are some alpha/beta-testers of it.

For BOM – we are finding information in the Altium database as well as KiCad. The KiCad export info we obtained in our first attempt is simplistic, and it does miss instance identifiers (c43, u17, r9 sorts of designators, which are present in Altium info). We do not see anything yet about enabling/disabling details types from the KiCad BOM export, so we are unsure whether more detailed columns can be obtained.

our gitlab repository with motherboard design sources

You may find the original Mentor Expedition sources in our GitLab repository. We were unable to run the visECAD Viewer as even the provided free license version doesn’t seem to work. In fact, visECAD Viewer appears to be withdrawn from the market and it not anymore available to download, and we did not not receive any answer to our requests of support from the Siemens team.

We were able to view the Altium import with the Altium online viewer thanks to the support of the Altium team.

After all our attempts, we are now pleased to announce that it is now possible to load the source files of the notebook motherboard PCB in KiCad using its kicad-nightly version.

Another recent news about the PowerPC notebook project, is that Slimbook will kindly provide us with two additional Slimbook Eclipse empty chassis. These will be used to test that our prototypes will correctly fit. At the same time, the guys at ACube Systems are investigating  suitable MXM video cards mounting AMD chips that could be used in the prototypes. We are investigating how we could collect  the additional funding required to pay for the various MXM video cards that will be used for testing and for the two additional Slimbook Eclipse chassis.Thanks to the kind contribution of the donors, the preparation of all components for the prototypes is progressing well.

We would like to reach the 50% of the final goal of the current campaign as soon as possible to avoid slowdowns in the current prototypes production phase. We currently need your financial support more than ever!

Another success: PCB completely financed! It’s time for the prototypes campaign!

Thanks to the kind contribution of the donors, we just reached the goal of the campaign for financing Phase 1B “Fast simulation bus”.

The PCB (Printed Circuit Board) design is currently being finalized, as soon as we have reviewed it, we will publish it in our GitLab repository. This last phase was of fundamental importance as we could test the correctness of the design, paving the way for the next campaign.

We got through the hardest parts with regards to the Research and Development choices. We past the uncertain ground with its many open challenges and many solutions have been explored. We also past the economical goal of previous campaigns that were quite heavy, and thanks to many people we have managed to get this far, again, thank you all!

We are now ready to launch Phase 2 “Production and delivery of five working prototypes” with a goal of 10500 Euros (around 12720 USD).

Our target is to complete this new campaign by Spring 2021, and we are working again with the patient guys at ACube Systems that will assist us in making the five prototypes.

printed circuit check

These prototypes will be very important as they will be used to

  • put the PCB design to the test
  • learn how to correctly initialize the hardware
  • fine-tune the configuration of U-Boot
  • fit the motherboard in the Eclipse Slimbook body

So far, we worked with U-Boot on the NXP T2080 RDB Devkit but that is quite different to our motherboard, which is quite more complex. We have to fine-tune U-Boot directly on the final hardware, and the prototypes will be essential for this. In addition, work on U-Uoot is still required even with the Devkit in order to correctly set up the framebuffer. Right now we can access the U-Boot console in serial mode only. ACube Systems will assist us on this task.

The motherboard is designed (screw and ports positions) to fit inside the Eclipse Notebook body that the prototypes should be mounted on.

Moreover, we need to redesign the dissipation heat pipes that will be slightly different from what was originally in place for the Eclipse Notebook.

Maybe some of you didn’t even think it was possible but we are progressing. The journey is still not finished. We need you to tell about this project all around you as we need more interested people, more donations.

2021 is our year! Let’s make this project a reality!