June 16, 2024, 02:35:42 PM

Development of Teres-2

Started by kreyren, December 13, 2023, 11:41:41 PM

Previous topic - Next topic

DiTBho

Quote from: kreyren on March 18, 2024, 03:01:58 PMOLIMEX has a solution for USB-to-Ethernet which is open-source

For the teres-1 I prefer the USB-ethernet adapter sold by Apple for their old Book-Air/Intel.

It's a chip that works very well with Linux (I still have to understand if uboot has support), consumes very little energy and is much more aesthetically pleasing too.

DiTBho

Just checked, the pinetab2 uses the Rockchip RK3566 SoC

kreyren

Quote from: DiTBho on March 20, 2024, 05:34:51 PMJust checked, the pinetab2 uses the Rockchip RK3566 SoC

Afaik the reason why OLIMEX didn't provide the RK3399 and RK3566 is that they were very unreliable software-wise (afaik they release the chip and then rely on the community to mainline them while not providing sufficient amount of documentation?) and tsvetan not wanting to support that as the resulting mainline is often very problematic and takes long time to get implemented in a way that is acceptable in industrial settings (OLIMEX's main focus)

For me in terms of arm architecture only use Cortex A7/53/55 and consider everything else garbage due to the CPU vulnerabilities, but i plan on supporting all chips as long as the required docs to make boards for these are available or unless someone wants to do the adventure of reverse-engineering them and contributing that (that's what the SOM management is meant to be for)
* Community Lead Developer of TeresDevelopment
* Downstream Maintainer of Linux and BSD* compatibility for OLIMEX Teres-I
* World record holder in annoying Tsvetan with questions about OLIMEX Teres-I hacking

kreyren

Quote from: DiTBho on March 18, 2024, 05:20:08 PM
Quote from: kreyren on March 18, 2024, 02:51:20 PM
Quote from: olimex on March 14, 2024, 08:01:19 PMA64 Ethernet is shared with the LCD, so if you want to use the build in Ethernet you lose the LCD, quite annoying decision but Allwinner made A64 thinking for tablets where Ethernet is not mandatory

As far as i was told that applies only for the LCD interface, you could still use the display over MIPI-DSI to enable the ethernet which might even be the more power efficient solution that could do touchscreen over e.g. I2C

So, could you check, and it's possible, could you do it for Teres-v1/Rev${next} and/or Teres-v1.5?


Not for teres 1.5 as it would require a redesign of the whole mainboard and the resulting product would be too costly for fabrication in small quantities due to the amount of mainboard layers and size (~250 EUR for 5 boards that don't have components on them).

For teres 1.5 i want to mainly extend the functionality and life of the existing devices as teres by itself is very functional netbook once you put a faster eMMC into it and i want it to be a reliable device for an open-source development so that it's by it's own a portable laboratory.

The mainboard redesign is meant for teres 2.0 as it will enable additional form-factors and i want the SOM solution to always provide a method to wire up an ethernet port while the user should be able to easily adjust the platform board to the computing they want (ideally at home as i am working on a cheap 2-layer PCB fabricator)

--

In terms of engineering the IO board is nearly done and I am researching eMMCs that are compatible with teres and trying to figure out how to document them for users to be able to easily find and replace while making a catalog of open-source friendly technicians that would for a small fee (decided that up to 15 EUR max is good deal especially if the user removes the mainboard from the case and hands that out to the technician to work on) replace the chip for them if the user lacks the skill to reflow the chip on a hot air station.

The 3D printable case has a bottom part nearly done just needs a better management for the keyboard to maintain the ~2mm key travel (adjustable) and then i need to adjust the display part. If you have access to the 3D printer (or want to get the parts fabricated locally which is projected to be around 8 EUR) to test the case design then let me know and i try to publish the prototype of the case soon. (I have to take care of a sick family member so that makes the work take longer than i would like atm)

Optionally if you have a recommendation for laptop hinges that would also be helpful as i am trying to make a design that re-uses the original as those seems to be superior to the hinges in consumer common laptops (based on me using the laptop as a workout tool and continuously opening and closing it to see if it will cause any damage for multiple weeks and it surviving without any noticable change in behavior), but it would be a good information to have for comparison in case there is one that is sufficiently economical and performs better.

In terms of teres 2.0 development I read through the https://neo900.org project and trying to get a hold of the designers.
If you can get a hold of any of them and ask them what is the state of the gerber files meaning if they are expected to provide basic functionality once fabricated (judging by the kicad files) then that would help a lot as the kicads seems to be exported from eagle and it seems very sketchy as they require minor adjustments to show up correctly in kicad and it seems like there are probably a lot of fabrication adjustments for those to work without issues.

Few days ago i was able to source the spare parts for Nokia N900 to partially re-construct the body to get reference for the dimensions and placements of IO and other components and I am currently trying to figure if:
* Re-use the Nokia N900 case and make a PCB that fits into the existing phone
* Make a new design inspried by the Nokia N900

The re-using of the Nokia N900 case seems kinda like a bad idea, it's using a very ancient display technology and it looks very comical in my large hands :D If i were to use it i would probably scale that up 150% for me, but i was thinking that if there is an interest e.g. in maemo community that i could try making that

Alternatively I feel like making a Nokia N900-inspired design that can be 3D printed is a better option to have a constraint for the System On Module solution to be able to fit in the chip while being economical to self-fabricate and probably making 2 versions one that tries to match the dimensions of N900 and other that better fits in my hands.

Beyond this last year i talk with beagledevs (non-profit open-source hardware developer) about SOM and there seemed to be an interest in at least exploring that option so i plan on making a proposal to pitch to them so that in ideal scenario they would be providing a BeagleV-Ahead-SOM that fits in the teres design as that would outsource lot of management-induced pain off of me and having beagledevs and OLIMEX contributing and helping designing the SOM standard would be the dream.

I also talked with community member about the ability to stick a beagle cape at the bottom of teres (mainly about placing a CNC cape to aid the design of CNC router), which doesn't seem to be realistic for teres 1 and 1.5 as it would require jumper wiring to the mainboard, but it's an interesting direction for the SOM standard so that teres 2.0 could use it e.g. standard connector on the module board that can be connected to the pcb that accepts beagle cape or making a reference designs with platform boards that have this integrated.. or (their proposal) taking my idea of turning all OLIMEX's SBC into a thing that can slot into an MxM slot (from https://git.dotya.ml/kreyren/OSHW-System-On-Module#use-of-edge-connectors) and placing pins all over it that can slot into beagle-cape but that would probably look hilarious and wouldn't be too functional and/or require a lot of additional engineering bcs some chips need cooling and the size is very important to be kept as compact as possible, for teres 2.0 it might work if we make a custom heatsinks and place the module board below the platform board, but i am not sure that is a good idea

In terms of software for A64: https://git.dotya.ml/KREYLIMEX/TERES/issues/19 and https://github.com/NixOS/nixpkgs/issues/297358 which seems to be the same issue are currently the major problem as it seems that u-boot introduced a regression that broke regulators during u-boot phase that also affect the OS even when the kernel seems to have them implemented correctly, it seems that (with a major help from sunxi-linux people) the problem was found and i am working on a patch.

https://git.dotya.ml/KREYLIMEX/TERES/issues/17 (Adjust the kernel for all distros) i am struggling with getting CEDRUS to work, but beyond that the configuration seems to be understood and i will update the sunxi wiki with it soon

https://git.dotya.ml/KREYLIMEX/TERES/issues/2 (Adjust the DTS for sun50i to fix thermal trips for all A64 chips) -- I have a patch kinda ready, but it seems that diego is handling that so i am monitoring it.

https://git.dotya.ml/KREYLIMEX/TERES/issues/1 (Make HDMI To Work On All Distros Reliably) -- Can't get that to work on nixos, itycodes said that she's interested in trying to fix it, but doesn't have the device so i am trying to adjust my infrastructure to enable the development for it.

https://git.dotya.ml/KREYLIMEX/TERES/issues/4 (Fix Audio on NixOS) -- This seems to be a distro-related problem that has lower priority over the mentioned atm
* Community Lead Developer of TeresDevelopment
* Downstream Maintainer of Linux and BSD* compatibility for OLIMEX Teres-I
* World record holder in annoying Tsvetan with questions about OLIMEX Teres-I hacking