site  contact  subhomenews

EOMA68 A10 boards working

April 13, 2013 — BarryK
This is good news. I reported on this earlier:

Now the boards are working:

I was getting a bit disinterested in the project, as the board is based on the Allwinner A10, which is a single-core Cortex-A8 chip. I want more speed!

However, this first board is proof-of-concept, and another with Freescale imx6 SoC is in the pipeline:

Also, boards based on Allwinner A20 and A31 are planned.
There is also a baseboard (with case) for a 7inch tablet being developed.


Flying Squirrel
Username: BarryK
The tablet board is codenamed Flying Squirrel:

Mele A1000G quad core
Username: don570
"from Mele is offering a new model with a faster processor and more storage and memory. The new model is called the Mele A1000G, and it features an Allwinner A31 ARM Cortex-A7 quad-core processor, PowerVR SGX544MP2 graphics, 2GB of RAM, and 8GB of storage. For comparison, the original Mele A1000 had a single core ARM Cortex-A8 chip, 512MB of memory, and just 2GB of storage.

Username: Raffy
"This is excellent news for ARM-Linux community. My Mele A1000 (single-core A10) is a proven light, rugged classroom companion.

Reply from Luke Leighton
Username: BarryK
"Luke is the main guy at Rhombus Tech. He sent me an email with a response to my blog posts, in particular an explanation about the 68-pin connector. Here is Luke's reply: [i]barry, hi i've only just encountered this entry, apologies! took me a while. to understand why only 68 pins, you need to look closely at the interfaces selected on them, and yes i think you already mentioned, the "limited" range of intended applications. the target products are mass-volume appliances, such as laptops (9in to 17in), tablets (5in to 12in), small desktop pcs, HDTVs, all-in-one LCD computers (15in to 40in), all-in-one keyboard computers, NAS Boxes, smart routers, games consoles, that sort of thing. [b]*[/b] most of those products need a hard drive (not all). they don't need multiple hard drives, or RAID. [b]*[/b] most of those products need network access (not all) i.e. ethernet. they don't need 10gigE, and they don't need dual NICs. [if they do, in the case of the smart router product, they can do it as a 4 or 8 or more port switch IC which they control using e.g. I2C for out-of-band communication such as "please filter this MAC address" etc.] [b]*[/b] most of those products need an LCD screen (not all of them). the resolution those products need is between 320x240 and 1920x1080. [b]*[/b] most of those products need to connect to USB peripherals. do they need to be ultra-ultra-ultra-ultra high-speed peripherals? not really. [b]*[/b] all of them need sensors of some kind. so that covers SATA, Ethernet, RGB/TTL, USB and I2C. done. you see how deceptively simple that is? ... why do you need more? seriously - why would you ever need more than that on the range of mass-volume appliances being covered by this standard? and it's a *big* range of mass-volume products. and, remember also, as you can see from the above picture you published, the interfaces that *can't* fit onto 68 pins, they can come out the other end as user-facing connectors. HDMI (19 pins), USB-OTG (5 pins), MicroSD (8 pins), Audio (ok we couldn't cram it in for this version but it's another 5 pins). if they won't fit on the user-facing connector, you can put a hatch in the top of the case, just like you get no *exactly* like you get for PCMCIA 3G cards, you could put a hatch for a top-loading SIM card, or a top-loading MicroSD card, or both.[/i] continued...

Reply from Luke, continued
Username: BarryK
"[i] you could put... a DisplayPort on there, you could put a... TV Antenna on there and build in a DVB-T TV chipset, you could put a 3G antenna on there, you could put... S-Video out, or... anything you like! so it's *not* as restrictive as it seems. you just have to think about it for a moment, and it starts to make a lot of sense. the only bug-bear is audio. audio i debated with my associates for some considerable time (several months in fact). the issue was that we wanted EOMA-68 to be able to go all the way back to some really ultra-low-cost ARM SoCs that would *not* necessarily have the "standard" AC97 I2S audio, and even if they did, they would not necessarily have all the capabilities of other audio systems. some of them would have say... 7.1 super-duper high-speed audio, and others would be able to cope with stereo at 32k. in the end we went, "y'know what? it's too many pins. plus, there's always the front edge of the PCMCIA case: they can put high-speed audio (SPDIF etc. etc.) out of that". and so we decided that audio should be done as something like a USB device. in that way, it STAYS GENERIC. USB AC97 audio ICs are low-cost, and we can guarantee that no matter what CPU Card is plugged in, a USB-based AC97 Audio chip will work across *all* of them, now and forever of the lifetime of the EOMA-68 standard. then for the flying squirrel i tried getting hold of one of these magic USB AC97 audio ICs, and could i get one? could i buggery :) so that was when i decided to implement audio using PWM in software (GPL licensed of course) from an STM32F, to drive an ultra-low-cost pair of $0.10 Class A 0.5 watt amplifiers that have been sold in the hundreds of millions if not billions of units over the years, in low-cost chinese-made radios and cassette players. it turns out that the cost of an STM32F is around $2.36 in 10k volumes, and it can do a LOT. one $2.36 little processor can do the job of $8 to $10 worth of discrete ICs, from audio, camera, LCD backlight, power management, SD/MMC card - you name it, incredibly it can do it. and the flying squirrel - schematics of the early version are available - will be the basis in variants of other future products as well. so basically, i've thought, "what's actually *really* needed across a mass-volume product range" and then thought outside the box on how to achieve that, and designed the standard accordingly.[/i] ...quite a reasonable explanation, I think! I am looking forward to putting Puppy onto a Flying Squirrel, and then upgrade the CPU as new boards get released. Quite exciting to look forward to.

Flying Puppy
Username: BarryK
"After all, we already have Puppy on a seagull... [img][/img]

circuit boards are so 20th century
Username: technosaurus
"Now that most devices are serial and we have fpga, 68 pins seems like overkill to me. Projects like openrisc and software defined radio are advancing open hardware into the future ... unfortunately there is no 2d/3d video card equivalent - even just a vesa compliant one, much less opengl(es)... anyone want an instantly fundable kickstarter? The possibility already exists to embed ram wireless and an lcd screen directly onto chips, as well as wireless power so the next big research _should_ involve the ability to embed battery technology into the chips directly. At that point it will be possible to have an _entire_ system on a chip at a disposable cost (eventually - a lot of R&D cost upfront) This is counter to how chip development is actually going - everyone is trying to go smaller to reduce the silicon cost, but putting more on one chip and having it big enough to embed a usable screen - the additional cost is an enclosure with virtually no labor. Circuit boards are so 20th century.

Tags: general